diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14343 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.999985976918007,
+  "global_step": 106965,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9985976721357454e-05,
+      "loss": 3.2989,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9971953442714907e-05,
+      "loss": 2.6398,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.995793016407236e-05,
+      "loss": 2.3782,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9943906885429814e-05,
+      "loss": 2.2384,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9929883606787268e-05,
+      "loss": 2.1279,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.991586032814472e-05,
+      "loss": 1.9667,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9901837049502174e-05,
+      "loss": 1.9143,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9887813770859628e-05,
+      "loss": 1.8277,
+      "step": 400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.987379049221708e-05,
+      "loss": 1.6988,
+      "step": 450
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9859767213574534e-05,
+      "loss": 1.637,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9845743934931988e-05,
+      "loss": 1.5378,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.983172065628944e-05,
+      "loss": 1.4824,
+      "step": 600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9817697377646894e-05,
+      "loss": 1.4092,
+      "step": 650
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9803674099004348e-05,
+      "loss": 1.294,
+      "step": 700
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.97896508203618e-05,
+      "loss": 1.2394,
+      "step": 750
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9775627541719254e-05,
+      "loss": 1.1931,
+      "step": 800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9761604263076708e-05,
+      "loss": 1.1169,
+      "step": 850
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.974758098443416e-05,
+      "loss": 1.0487,
+      "step": 900
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9733557705791615e-05,
+      "loss": 0.9585,
+      "step": 950
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9719534427149068e-05,
+      "loss": 0.9143,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "eval_bleu": 79.3692,
+      "eval_gen_len": 61.919,
+      "eval_loss": 0.5720869302749634,
+      "eval_rouge1": 87.2325,
+      "eval_rouge2": 78.8924,
+      "eval_rougeL": 85.959,
+      "eval_rougeLsum": 86.2188,
+      "eval_runtime": 3277.2108,
+      "eval_samples_per_second": 0.915,
+      "eval_steps_per_second": 0.229,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.970551114850652e-05,
+      "loss": 0.8543,
+      "step": 1050
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9691487869863975e-05,
+      "loss": 0.813,
+      "step": 1100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9677464591221428e-05,
+      "loss": 0.7565,
+      "step": 1150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.966344131257888e-05,
+      "loss": 0.6924,
+      "step": 1200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9649418033936335e-05,
+      "loss": 0.6407,
+      "step": 1250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9635394755293788e-05,
+      "loss": 0.6138,
+      "step": 1300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.962137147665124e-05,
+      "loss": 0.5659,
+      "step": 1350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9607348198008695e-05,
+      "loss": 0.5688,
+      "step": 1400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9593324919366148e-05,
+      "loss": 0.516,
+      "step": 1450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.95793016407236e-05,
+      "loss": 0.4945,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9565278362081055e-05,
+      "loss": 0.4414,
+      "step": 1550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9551255083438508e-05,
+      "loss": 0.4591,
+      "step": 1600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9537231804795965e-05,
+      "loss": 0.4378,
+      "step": 1650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9523208526153415e-05,
+      "loss": 0.4287,
+      "step": 1700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9509185247510868e-05,
+      "loss": 0.4494,
+      "step": 1750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.949516196886832e-05,
+      "loss": 0.4241,
+      "step": 1800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9481138690225775e-05,
+      "loss": 0.4092,
+      "step": 1850
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.946711541158323e-05,
+      "loss": 0.4149,
+      "step": 1900
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9453092132940682e-05,
+      "loss": 0.3603,
+      "step": 1950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9439068854298135e-05,
+      "loss": 0.3801,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "eval_bleu": 93.6507,
+      "eval_gen_len": 64.828,
+      "eval_loss": 0.4033011794090271,
+      "eval_rouge1": 91.7405,
+      "eval_rouge2": 86.6147,
+      "eval_rougeL": 91.4332,
+      "eval_rougeLsum": 91.4208,
+      "eval_runtime": 3199.5288,
+      "eval_samples_per_second": 0.938,
+      "eval_steps_per_second": 0.234,
+      "step": 2000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9425045575655592e-05,
+      "loss": 0.3768,
+      "step": 2050
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9411022297013042e-05,
+      "loss": 0.3616,
+      "step": 2100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9396999018370495e-05,
+      "loss": 0.3987,
+      "step": 2150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.938297573972795e-05,
+      "loss": 0.3601,
+      "step": 2200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9368952461085402e-05,
+      "loss": 0.3416,
+      "step": 2250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9354929182442855e-05,
+      "loss": 0.3517,
+      "step": 2300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.934090590380031e-05,
+      "loss": 0.3315,
+      "step": 2350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9326882625157762e-05,
+      "loss": 0.3501,
+      "step": 2400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.931285934651522e-05,
+      "loss": 0.3275,
+      "step": 2450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.929883606787267e-05,
+      "loss": 0.3098,
+      "step": 2500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9284812789230122e-05,
+      "loss": 0.324,
+      "step": 2550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9270789510587575e-05,
+      "loss": 0.3129,
+      "step": 2600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.925676623194503e-05,
+      "loss": 0.3394,
+      "step": 2650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9242742953302482e-05,
+      "loss": 0.2813,
+      "step": 2700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9228719674659935e-05,
+      "loss": 0.3168,
+      "step": 2750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.921469639601739e-05,
+      "loss": 0.3266,
+      "step": 2800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9200673117374846e-05,
+      "loss": 0.2833,
+      "step": 2850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9186649838732296e-05,
+      "loss": 0.3025,
+      "step": 2900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.917262656008975e-05,
+      "loss": 0.3095,
+      "step": 2950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9158603281447202e-05,
+      "loss": 0.2844,
+      "step": 3000
+    },
+    {
+      "epoch": 0.08,
+      "eval_bleu": 94.365,
+      "eval_gen_len": 64.905,
+      "eval_loss": 0.3663557171821594,
+      "eval_rouge1": 91.9207,
+      "eval_rouge2": 87.0399,
+      "eval_rougeL": 91.6573,
+      "eval_rougeLsum": 91.6307,
+      "eval_runtime": 3129.011,
+      "eval_samples_per_second": 0.959,
+      "eval_steps_per_second": 0.24,
+      "step": 3000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9144580002804656e-05,
+      "loss": 0.3001,
+      "step": 3050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9130556724162112e-05,
+      "loss": 0.2903,
+      "step": 3100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9116533445519562e-05,
+      "loss": 0.2751,
+      "step": 3150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9102510166877016e-05,
+      "loss": 0.2725,
+      "step": 3200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9088486888234472e-05,
+      "loss": 0.2897,
+      "step": 3250
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9074463609591922e-05,
+      "loss": 0.2878,
+      "step": 3300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9060440330949376e-05,
+      "loss": 0.2966,
+      "step": 3350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.904641705230683e-05,
+      "loss": 0.3079,
+      "step": 3400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9032393773664282e-05,
+      "loss": 0.2598,
+      "step": 3450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.901837049502174e-05,
+      "loss": 0.2982,
+      "step": 3500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.900434721637919e-05,
+      "loss": 0.2634,
+      "step": 3550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8990323937736643e-05,
+      "loss": 0.2896,
+      "step": 3600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.89763006590941e-05,
+      "loss": 0.2822,
+      "step": 3650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.896227738045155e-05,
+      "loss": 0.278,
+      "step": 3700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8948254101809003e-05,
+      "loss": 0.2818,
+      "step": 3750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8934230823166456e-05,
+      "loss": 0.2788,
+      "step": 3800
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.892020754452391e-05,
+      "loss": 0.2989,
+      "step": 3850
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8906184265881366e-05,
+      "loss": 0.3021,
+      "step": 3900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8892160987238816e-05,
+      "loss": 0.281,
+      "step": 3950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.887813770859627e-05,
+      "loss": 0.2612,
+      "step": 4000
+    },
+    {
+      "epoch": 0.11,
+      "eval_bleu": 94.6072,
+      "eval_gen_len": 64.9013,
+      "eval_loss": 0.3458440899848938,
+      "eval_rouge1": 92.2699,
+      "eval_rouge2": 87.5817,
+      "eval_rougeL": 92.003,
+      "eval_rougeLsum": 91.9759,
+      "eval_runtime": 3105.6443,
+      "eval_samples_per_second": 0.966,
+      "eval_steps_per_second": 0.241,
+      "step": 4000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8864114429953726e-05,
+      "loss": 0.2749,
+      "step": 4050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8850091151311176e-05,
+      "loss": 0.2948,
+      "step": 4100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.883606787266863e-05,
+      "loss": 0.2429,
+      "step": 4150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8822044594026083e-05,
+      "loss": 0.2793,
+      "step": 4200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8808021315383536e-05,
+      "loss": 0.2899,
+      "step": 4250
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8793998036740993e-05,
+      "loss": 0.2384,
+      "step": 4300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8779974758098443e-05,
+      "loss": 0.2783,
+      "step": 4350
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8765951479455896e-05,
+      "loss": 0.2709,
+      "step": 4400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8751928200813353e-05,
+      "loss": 0.2453,
+      "step": 4450
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8737904922170803e-05,
+      "loss": 0.2626,
+      "step": 4500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.872388164352826e-05,
+      "loss": 0.2476,
+      "step": 4550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.870985836488571e-05,
+      "loss": 0.2207,
+      "step": 4600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8695835086243163e-05,
+      "loss": 0.2553,
+      "step": 4650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.868181180760062e-05,
+      "loss": 0.2624,
+      "step": 4700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.866778852895807e-05,
+      "loss": 0.2505,
+      "step": 4750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8653765250315523e-05,
+      "loss": 0.2356,
+      "step": 4800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.863974197167298e-05,
+      "loss": 0.2529,
+      "step": 4850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.862571869303043e-05,
+      "loss": 0.2331,
+      "step": 4900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8611695414387887e-05,
+      "loss": 0.2487,
+      "step": 4950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8597672135745337e-05,
+      "loss": 0.2609,
+      "step": 5000
+    },
+    {
+      "epoch": 0.14,
+      "eval_bleu": 94.6745,
+      "eval_gen_len": 64.809,
+      "eval_loss": 0.33044180274009705,
+      "eval_rouge1": 92.3236,
+      "eval_rouge2": 87.7132,
+      "eval_rougeL": 92.0699,
+      "eval_rougeLsum": 92.0344,
+      "eval_runtime": 3150.3076,
+      "eval_samples_per_second": 0.952,
+      "eval_steps_per_second": 0.238,
+      "step": 5000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.858364885710279e-05,
+      "loss": 0.2444,
+      "step": 5050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8569625578460247e-05,
+      "loss": 0.2717,
+      "step": 5100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.8555602299817697e-05,
+      "loss": 0.2372,
+      "step": 5150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.854157902117515e-05,
+      "loss": 0.2448,
+      "step": 5200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8527555742532607e-05,
+      "loss": 0.2752,
+      "step": 5250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8513532463890057e-05,
+      "loss": 0.2762,
+      "step": 5300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8499509185247513e-05,
+      "loss": 0.2355,
+      "step": 5350
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8485485906604963e-05,
+      "loss": 0.2565,
+      "step": 5400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8471462627962417e-05,
+      "loss": 0.2795,
+      "step": 5450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.8457439349319873e-05,
+      "loss": 0.2305,
+      "step": 5500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8443416070677323e-05,
+      "loss": 0.2515,
+      "step": 5550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8429392792034777e-05,
+      "loss": 0.2473,
+      "step": 5600
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8415369513392234e-05,
+      "loss": 0.2531,
+      "step": 5650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8401346234749684e-05,
+      "loss": 0.2243,
+      "step": 5700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.838732295610714e-05,
+      "loss": 0.2272,
+      "step": 5750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.837329967746459e-05,
+      "loss": 0.2601,
+      "step": 5800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.8359276398822044e-05,
+      "loss": 0.2539,
+      "step": 5850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.83452531201795e-05,
+      "loss": 0.2445,
+      "step": 5900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.833122984153695e-05,
+      "loss": 0.213,
+      "step": 5950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8317206562894407e-05,
+      "loss": 0.2173,
+      "step": 6000
+    },
+    {
+      "epoch": 0.17,
+      "eval_bleu": 94.886,
+      "eval_gen_len": 64.737,
+      "eval_loss": 0.32196304202079773,
+      "eval_rouge1": 92.4893,
+      "eval_rouge2": 87.9435,
+      "eval_rougeL": 92.23,
+      "eval_rougeLsum": 92.1972,
+      "eval_runtime": 3194.0894,
+      "eval_samples_per_second": 0.939,
+      "eval_steps_per_second": 0.235,
+      "step": 6000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.830318328425186e-05,
+      "loss": 0.2123,
+      "step": 6050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.828916000560931e-05,
+      "loss": 0.2351,
+      "step": 6100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8275136726966767e-05,
+      "loss": 0.2546,
+      "step": 6150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.8261113448324217e-05,
+      "loss": 0.2388,
+      "step": 6200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.824709016968167e-05,
+      "loss": 0.2309,
+      "step": 6250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8233066891039127e-05,
+      "loss": 0.2379,
+      "step": 6300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8219043612396577e-05,
+      "loss": 0.2269,
+      "step": 6350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8205020333754034e-05,
+      "loss": 0.2362,
+      "step": 6400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8190997055111487e-05,
+      "loss": 0.2325,
+      "step": 6450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8176973776468937e-05,
+      "loss": 0.2519,
+      "step": 6500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.8162950497826394e-05,
+      "loss": 0.215,
+      "step": 6550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8148927219183844e-05,
+      "loss": 0.2301,
+      "step": 6600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8134903940541297e-05,
+      "loss": 0.247,
+      "step": 6650
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8120880661898754e-05,
+      "loss": 0.2261,
+      "step": 6700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8106857383256204e-05,
+      "loss": 0.2349,
+      "step": 6750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.809283410461366e-05,
+      "loss": 0.2253,
+      "step": 6800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8078810825971114e-05,
+      "loss": 0.2186,
+      "step": 6850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.8064787547328564e-05,
+      "loss": 0.238,
+      "step": 6900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.805076426868602e-05,
+      "loss": 0.2308,
+      "step": 6950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.803674099004347e-05,
+      "loss": 0.2252,
+      "step": 7000
+    },
+    {
+      "epoch": 0.2,
+      "eval_bleu": 94.9526,
+      "eval_gen_len": 64.719,
+      "eval_loss": 0.3162732422351837,
+      "eval_rouge1": 92.5629,
+      "eval_rouge2": 88.0962,
+      "eval_rougeL": 92.3033,
+      "eval_rougeLsum": 92.2707,
+      "eval_runtime": 3111.1893,
+      "eval_samples_per_second": 0.964,
+      "eval_steps_per_second": 0.241,
+      "step": 7000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.8022717711400924e-05,
+      "loss": 0.2267,
+      "step": 7050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.800869443275838e-05,
+      "loss": 0.2197,
+      "step": 7100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.799467115411583e-05,
+      "loss": 0.195,
+      "step": 7150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7980647875473288e-05,
+      "loss": 0.1975,
+      "step": 7200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.796662459683074e-05,
+      "loss": 0.2261,
+      "step": 7250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.795260131818819e-05,
+      "loss": 0.2276,
+      "step": 7300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7938578039545648e-05,
+      "loss": 0.2236,
+      "step": 7350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7924554760903098e-05,
+      "loss": 0.2362,
+      "step": 7400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7910531482260554e-05,
+      "loss": 0.1933,
+      "step": 7450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7896508203618008e-05,
+      "loss": 0.256,
+      "step": 7500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7882484924975458e-05,
+      "loss": 0.22,
+      "step": 7550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7868461646332915e-05,
+      "loss": 0.2285,
+      "step": 7600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.7854438367690368e-05,
+      "loss": 0.2297,
+      "step": 7650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7840415089047818e-05,
+      "loss": 0.2198,
+      "step": 7700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7826391810405275e-05,
+      "loss": 0.219,
+      "step": 7750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7812368531762725e-05,
+      "loss": 0.264,
+      "step": 7800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.779834525312018e-05,
+      "loss": 0.2218,
+      "step": 7850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7784321974477635e-05,
+      "loss": 0.2343,
+      "step": 7900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.7770298695835085e-05,
+      "loss": 0.223,
+      "step": 7950
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.775627541719254e-05,
+      "loss": 0.2111,
+      "step": 8000
+    },
+    {
+      "epoch": 0.22,
+      "eval_bleu": 95.0762,
+      "eval_gen_len": 64.6663,
+      "eval_loss": 0.31111225485801697,
+      "eval_rouge1": 92.7158,
+      "eval_rouge2": 88.3325,
+      "eval_rougeL": 92.4682,
+      "eval_rougeLsum": 92.4352,
+      "eval_runtime": 3140.711,
+      "eval_samples_per_second": 0.955,
+      "eval_steps_per_second": 0.239,
+      "step": 8000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7742252138549995e-05,
+      "loss": 0.2171,
+      "step": 8050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7728228859907445e-05,
+      "loss": 0.2263,
+      "step": 8100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.77142055812649e-05,
+      "loss": 0.2359,
+      "step": 8150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.770018230262235e-05,
+      "loss": 0.2058,
+      "step": 8200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.7686159023979808e-05,
+      "loss": 0.2044,
+      "step": 8250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.767213574533726e-05,
+      "loss": 0.2322,
+      "step": 8300
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.765811246669471e-05,
+      "loss": 0.2289,
+      "step": 8350
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7644089188052168e-05,
+      "loss": 0.2228,
+      "step": 8400
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.763006590940962e-05,
+      "loss": 0.2211,
+      "step": 8450
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.761604263076707e-05,
+      "loss": 0.2297,
+      "step": 8500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.760201935212453e-05,
+      "loss": 0.2087,
+      "step": 8550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7587996073481978e-05,
+      "loss": 0.2266,
+      "step": 8600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7573972794839435e-05,
+      "loss": 0.2259,
+      "step": 8650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.755994951619689e-05,
+      "loss": 0.1985,
+      "step": 8700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.754592623755434e-05,
+      "loss": 0.2023,
+      "step": 8750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7531902958911795e-05,
+      "loss": 0.2297,
+      "step": 8800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.751787968026925e-05,
+      "loss": 0.2179,
+      "step": 8850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7503856401626702e-05,
+      "loss": 0.1993,
+      "step": 8900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7489833122984155e-05,
+      "loss": 0.1946,
+      "step": 8950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7475809844341605e-05,
+      "loss": 0.195,
+      "step": 9000
+    },
+    {
+      "epoch": 0.25,
+      "eval_bleu": 95.1941,
+      "eval_gen_len": 64.634,
+      "eval_loss": 0.3037899434566498,
+      "eval_rouge1": 92.7974,
+      "eval_rouge2": 88.5146,
+      "eval_rougeL": 92.5398,
+      "eval_rougeLsum": 92.5031,
+      "eval_runtime": 3164.9424,
+      "eval_samples_per_second": 0.948,
+      "eval_steps_per_second": 0.237,
+      "step": 9000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7461786565699062e-05,
+      "loss": 0.2171,
+      "step": 9050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7447763287056515e-05,
+      "loss": 0.2282,
+      "step": 9100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7433740008413965e-05,
+      "loss": 0.2159,
+      "step": 9150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7419716729771422e-05,
+      "loss": 0.221,
+      "step": 9200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7405693451128875e-05,
+      "loss": 0.2357,
+      "step": 9250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.739167017248633e-05,
+      "loss": 0.221,
+      "step": 9300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7377646893843782e-05,
+      "loss": 0.2064,
+      "step": 9350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.7363623615201232e-05,
+      "loss": 0.2112,
+      "step": 9400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.734960033655869e-05,
+      "loss": 0.2489,
+      "step": 9450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7335577057916142e-05,
+      "loss": 0.2209,
+      "step": 9500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7321553779273592e-05,
+      "loss": 0.2435,
+      "step": 9550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.730753050063105e-05,
+      "loss": 0.2135,
+      "step": 9600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7293507221988502e-05,
+      "loss": 0.2398,
+      "step": 9650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.7279483943345956e-05,
+      "loss": 0.2005,
+      "step": 9700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.726546066470341e-05,
+      "loss": 0.2122,
+      "step": 9750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.725143738606086e-05,
+      "loss": 0.2027,
+      "step": 9800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7237414107418316e-05,
+      "loss": 0.199,
+      "step": 9850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.722339082877577e-05,
+      "loss": 0.222,
+      "step": 9900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7209367550133222e-05,
+      "loss": 0.2185,
+      "step": 9950
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7195344271490676e-05,
+      "loss": 0.1982,
+      "step": 10000
+    },
+    {
+      "epoch": 0.28,
+      "eval_bleu": 95.2522,
+      "eval_gen_len": 64.626,
+      "eval_loss": 0.3007320463657379,
+      "eval_rouge1": 92.7899,
+      "eval_rouge2": 88.4907,
+      "eval_rougeL": 92.5338,
+      "eval_rougeLsum": 92.4971,
+      "eval_runtime": 3103.217,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 10000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.718132099284813e-05,
+      "loss": 0.2071,
+      "step": 10050
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7167297714205582e-05,
+      "loss": 0.204,
+      "step": 10100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.7153274435563036e-05,
+      "loss": 0.2138,
+      "step": 10150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7139251156920486e-05,
+      "loss": 0.2045,
+      "step": 10200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7125227878277942e-05,
+      "loss": 0.2005,
+      "step": 10250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7111204599635396e-05,
+      "loss": 0.2392,
+      "step": 10300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.709718132099285e-05,
+      "loss": 0.2126,
+      "step": 10350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7083158042350303e-05,
+      "loss": 0.2097,
+      "step": 10400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.7069134763707756e-05,
+      "loss": 0.2197,
+      "step": 10450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.705511148506521e-05,
+      "loss": 0.2013,
+      "step": 10500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7041088206422663e-05,
+      "loss": 0.2208,
+      "step": 10550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.7027064927780113e-05,
+      "loss": 0.1938,
+      "step": 10600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.701304164913757e-05,
+      "loss": 0.2048,
+      "step": 10650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6999018370495023e-05,
+      "loss": 0.2065,
+      "step": 10700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6984995091852476e-05,
+      "loss": 0.2075,
+      "step": 10750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.697097181320993e-05,
+      "loss": 0.2279,
+      "step": 10800
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.6956948534567383e-05,
+      "loss": 0.2072,
+      "step": 10850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.6942925255924836e-05,
+      "loss": 0.1998,
+      "step": 10900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.692890197728229e-05,
+      "loss": 0.1904,
+      "step": 10950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.691487869863974e-05,
+      "loss": 0.1762,
+      "step": 11000
+    },
+    {
+      "epoch": 0.31,
+      "eval_bleu": 95.3107,
+      "eval_gen_len": 64.546,
+      "eval_loss": 0.29688259959220886,
+      "eval_rouge1": 92.939,
+      "eval_rouge2": 88.7461,
+      "eval_rougeL": 92.6735,
+      "eval_rougeLsum": 92.6483,
+      "eval_runtime": 3120.8781,
+      "eval_samples_per_second": 0.961,
+      "eval_steps_per_second": 0.24,
+      "step": 11000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.6900855419997196e-05,
+      "loss": 0.2241,
+      "step": 11050
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.688683214135465e-05,
+      "loss": 0.2087,
+      "step": 11100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.6872808862712103e-05,
+      "loss": 0.1918,
+      "step": 11150
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.6858785584069556e-05,
+      "loss": 0.2117,
+      "step": 11200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.684476230542701e-05,
+      "loss": 0.213,
+      "step": 11250
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.6830739026784463e-05,
+      "loss": 0.2338,
+      "step": 11300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.6816715748141916e-05,
+      "loss": 0.2056,
+      "step": 11350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.680269246949937e-05,
+      "loss": 0.2087,
+      "step": 11400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.6788669190856823e-05,
+      "loss": 0.2014,
+      "step": 11450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.6774645912214276e-05,
+      "loss": 0.2464,
+      "step": 11500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.676062263357173e-05,
+      "loss": 0.2132,
+      "step": 11550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.6746599354929183e-05,
+      "loss": 0.2199,
+      "step": 11600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.6732576076286636e-05,
+      "loss": 0.187,
+      "step": 11650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.671855279764409e-05,
+      "loss": 0.2053,
+      "step": 11700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.6704529519001543e-05,
+      "loss": 0.1936,
+      "step": 11750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.6690506240358997e-05,
+      "loss": 0.2214,
+      "step": 11800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.667648296171645e-05,
+      "loss": 0.2056,
+      "step": 11850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.6662459683073903e-05,
+      "loss": 0.181,
+      "step": 11900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.6648436404431357e-05,
+      "loss": 0.2127,
+      "step": 11950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.663441312578881e-05,
+      "loss": 0.1798,
+      "step": 12000
+    },
+    {
+      "epoch": 0.34,
+      "eval_bleu": 95.3546,
+      "eval_gen_len": 64.4707,
+      "eval_loss": 0.2918665111064911,
+      "eval_rouge1": 92.9828,
+      "eval_rouge2": 88.8459,
+      "eval_rougeL": 92.7418,
+      "eval_rougeLsum": 92.6996,
+      "eval_runtime": 3099.2674,
+      "eval_samples_per_second": 0.968,
+      "eval_steps_per_second": 0.242,
+      "step": 12000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.6620389847146263e-05,
+      "loss": 0.207,
+      "step": 12050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.6606366568503717e-05,
+      "loss": 0.1874,
+      "step": 12100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.659234328986117e-05,
+      "loss": 0.1933,
+      "step": 12150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.6578320011218623e-05,
+      "loss": 0.2117,
+      "step": 12200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.6564296732576077e-05,
+      "loss": 0.2147,
+      "step": 12250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.655027345393353e-05,
+      "loss": 0.2206,
+      "step": 12300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6536250175290983e-05,
+      "loss": 0.2034,
+      "step": 12350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6522226896648437e-05,
+      "loss": 0.2019,
+      "step": 12400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.650820361800589e-05,
+      "loss": 0.1982,
+      "step": 12450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6494180339363344e-05,
+      "loss": 0.1851,
+      "step": 12500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6480157060720797e-05,
+      "loss": 0.2089,
+      "step": 12550
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.646613378207825e-05,
+      "loss": 0.1985,
+      "step": 12600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.6452110503435704e-05,
+      "loss": 0.1879,
+      "step": 12650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6438087224793157e-05,
+      "loss": 0.2285,
+      "step": 12700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.642406394615061e-05,
+      "loss": 0.2068,
+      "step": 12750
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6410040667508064e-05,
+      "loss": 0.2248,
+      "step": 12800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6396017388865517e-05,
+      "loss": 0.2017,
+      "step": 12850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.638199411022297e-05,
+      "loss": 0.2064,
+      "step": 12900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6367970831580424e-05,
+      "loss": 0.1967,
+      "step": 12950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.6353947552937877e-05,
+      "loss": 0.2037,
+      "step": 13000
+    },
+    {
+      "epoch": 0.36,
+      "eval_bleu": 95.472,
+      "eval_gen_len": 64.5277,
+      "eval_loss": 0.28798067569732666,
+      "eval_rouge1": 93.1329,
+      "eval_rouge2": 89.0529,
+      "eval_rougeL": 92.8733,
+      "eval_rougeLsum": 92.8293,
+      "eval_runtime": 3148.5331,
+      "eval_samples_per_second": 0.953,
+      "eval_steps_per_second": 0.238,
+      "step": 13000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.633992427429533e-05,
+      "loss": 0.1902,
+      "step": 13050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6325900995652784e-05,
+      "loss": 0.1787,
+      "step": 13100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6311877717010237e-05,
+      "loss": 0.1822,
+      "step": 13150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.629785443836769e-05,
+      "loss": 0.2024,
+      "step": 13200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6283831159725147e-05,
+      "loss": 0.2068,
+      "step": 13250
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.6269807881082597e-05,
+      "loss": 0.2084,
+      "step": 13300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.625578460244005e-05,
+      "loss": 0.1967,
+      "step": 13350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6241761323797504e-05,
+      "loss": 0.1902,
+      "step": 13400
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6227738045154957e-05,
+      "loss": 0.1975,
+      "step": 13450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.621371476651241e-05,
+      "loss": 0.2069,
+      "step": 13500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6199691487869864e-05,
+      "loss": 0.2188,
+      "step": 13550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6185668209227317e-05,
+      "loss": 0.2088,
+      "step": 13600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6171644930584774e-05,
+      "loss": 0.1875,
+      "step": 13650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.6157621651942224e-05,
+      "loss": 0.1881,
+      "step": 13700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6143598373299678e-05,
+      "loss": 0.1907,
+      "step": 13750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.612957509465713e-05,
+      "loss": 0.1968,
+      "step": 13800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6115551816014584e-05,
+      "loss": 0.2087,
+      "step": 13850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6101528537372038e-05,
+      "loss": 0.1805,
+      "step": 13900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.608750525872949e-05,
+      "loss": 0.1897,
+      "step": 13950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.6073481980086944e-05,
+      "loss": 0.2091,
+      "step": 14000
+    },
+    {
+      "epoch": 0.39,
+      "eval_bleu": 95.499,
+      "eval_gen_len": 64.423,
+      "eval_loss": 0.2838546335697174,
+      "eval_rouge1": 93.0724,
+      "eval_rouge2": 89.0302,
+      "eval_rougeL": 92.8261,
+      "eval_rougeLsum": 92.7827,
+      "eval_runtime": 3139.4968,
+      "eval_samples_per_second": 0.956,
+      "eval_steps_per_second": 0.239,
+      "step": 14000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.60594587014444e-05,
+      "loss": 0.1831,
+      "step": 14050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.604543542280185e-05,
+      "loss": 0.1918,
+      "step": 14100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6031412144159304e-05,
+      "loss": 0.1973,
+      "step": 14150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.6017388865516758e-05,
+      "loss": 0.1948,
+      "step": 14200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.600336558687421e-05,
+      "loss": 0.202,
+      "step": 14250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5989342308231664e-05,
+      "loss": 0.2092,
+      "step": 14300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.5975319029589118e-05,
+      "loss": 0.189,
+      "step": 14350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.596129575094657e-05,
+      "loss": 0.2014,
+      "step": 14400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.5947272472304028e-05,
+      "loss": 0.1835,
+      "step": 14450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.5933249193661478e-05,
+      "loss": 0.1941,
+      "step": 14500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.591922591501893e-05,
+      "loss": 0.2027,
+      "step": 14550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.5905202636376385e-05,
+      "loss": 0.1794,
+      "step": 14600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.5891179357733838e-05,
+      "loss": 0.2,
+      "step": 14650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.5877156079091295e-05,
+      "loss": 0.2091,
+      "step": 14700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.5863132800448745e-05,
+      "loss": 0.1813,
+      "step": 14750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5849109521806198e-05,
+      "loss": 0.1793,
+      "step": 14800
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5835086243163655e-05,
+      "loss": 0.1852,
+      "step": 14850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5821062964521105e-05,
+      "loss": 0.2122,
+      "step": 14900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5807039685878558e-05,
+      "loss": 0.1758,
+      "step": 14950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.579301640723601e-05,
+      "loss": 0.1998,
+      "step": 15000
+    },
+    {
+      "epoch": 0.42,
+      "eval_bleu": 95.5491,
+      "eval_gen_len": 64.419,
+      "eval_loss": 0.28141552209854126,
+      "eval_rouge1": 93.1634,
+      "eval_rouge2": 89.1339,
+      "eval_rougeL": 92.9186,
+      "eval_rougeLsum": 92.904,
+      "eval_runtime": 3158.2947,
+      "eval_samples_per_second": 0.95,
+      "eval_steps_per_second": 0.237,
+      "step": 15000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.5778993128593465e-05,
+      "loss": 0.1902,
+      "step": 15050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.576496984995092e-05,
+      "loss": 0.1829,
+      "step": 15100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.575094657130837e-05,
+      "loss": 0.2212,
+      "step": 15150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5736923292665825e-05,
+      "loss": 0.1878,
+      "step": 15200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.572290001402328e-05,
+      "loss": 0.2088,
+      "step": 15250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.570887673538073e-05,
+      "loss": 0.217,
+      "step": 15300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.5694853456738185e-05,
+      "loss": 0.1859,
+      "step": 15350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.568083017809564e-05,
+      "loss": 0.1975,
+      "step": 15400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.566680689945309e-05,
+      "loss": 0.1848,
+      "step": 15450
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.565278362081055e-05,
+      "loss": 0.2097,
+      "step": 15500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5638760342168e-05,
+      "loss": 0.1984,
+      "step": 15550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5624737063525452e-05,
+      "loss": 0.1905,
+      "step": 15600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.561071378488291e-05,
+      "loss": 0.2155,
+      "step": 15650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.559669050624036e-05,
+      "loss": 0.1933,
+      "step": 15700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5582667227597815e-05,
+      "loss": 0.1853,
+      "step": 15750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.5568643948955265e-05,
+      "loss": 0.1849,
+      "step": 15800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.555462067031272e-05,
+      "loss": 0.179,
+      "step": 15850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5540597391670175e-05,
+      "loss": 0.1907,
+      "step": 15900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5526574113027625e-05,
+      "loss": 0.2032,
+      "step": 15950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.551255083438508e-05,
+      "loss": 0.1855,
+      "step": 16000
+    },
+    {
+      "epoch": 0.45,
+      "eval_bleu": 95.5711,
+      "eval_gen_len": 64.4307,
+      "eval_loss": 0.27981552481651306,
+      "eval_rouge1": 93.2066,
+      "eval_rouge2": 89.2588,
+      "eval_rougeL": 92.9732,
+      "eval_rougeLsum": 92.9506,
+      "eval_runtime": 3083.9167,
+      "eval_samples_per_second": 0.973,
+      "eval_steps_per_second": 0.243,
+      "step": 16000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5498527555742535e-05,
+      "loss": 0.19,
+      "step": 16050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5484504277099985e-05,
+      "loss": 0.1877,
+      "step": 16100
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5470480998457442e-05,
+      "loss": 0.1868,
+      "step": 16150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.5456457719814892e-05,
+      "loss": 0.179,
+      "step": 16200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.5442434441172345e-05,
+      "loss": 0.2247,
+      "step": 16250
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.5428411162529802e-05,
+      "loss": 0.1745,
+      "step": 16300
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.5414387883887252e-05,
+      "loss": 0.2015,
+      "step": 16350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.5400364605244705e-05,
+      "loss": 0.212,
+      "step": 16400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.5386341326602162e-05,
+      "loss": 0.2018,
+      "step": 16450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.5372318047959612e-05,
+      "loss": 0.1888,
+      "step": 16500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.535829476931707e-05,
+      "loss": 0.1806,
+      "step": 16550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.534427149067452e-05,
+      "loss": 0.1791,
+      "step": 16600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.5330248212031972e-05,
+      "loss": 0.1949,
+      "step": 16650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.531622493338943e-05,
+      "loss": 0.1858,
+      "step": 16700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.530220165474688e-05,
+      "loss": 0.1879,
+      "step": 16750
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.5288178376104332e-05,
+      "loss": 0.1936,
+      "step": 16800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.527415509746179e-05,
+      "loss": 0.1692,
+      "step": 16850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.526013181881924e-05,
+      "loss": 0.1852,
+      "step": 16900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5246108540176696e-05,
+      "loss": 0.2128,
+      "step": 16950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5232085261534146e-05,
+      "loss": 0.184,
+      "step": 17000
+    },
+    {
+      "epoch": 0.48,
+      "eval_bleu": 95.6325,
+      "eval_gen_len": 64.3623,
+      "eval_loss": 0.2761005759239197,
+      "eval_rouge1": 93.269,
+      "eval_rouge2": 89.3661,
+      "eval_rougeL": 93.0637,
+      "eval_rougeLsum": 93.0433,
+      "eval_runtime": 3126.3939,
+      "eval_samples_per_second": 0.96,
+      "eval_steps_per_second": 0.24,
+      "step": 17000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.52180619828916e-05,
+      "loss": 0.2114,
+      "step": 17050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5204038704249056e-05,
+      "loss": 0.1933,
+      "step": 17100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5190015425606506e-05,
+      "loss": 0.1978,
+      "step": 17150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5175992146963963e-05,
+      "loss": 0.2033,
+      "step": 17200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.5161968868321416e-05,
+      "loss": 0.1854,
+      "step": 17250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5147945589678866e-05,
+      "loss": 0.1792,
+      "step": 17300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5133922311036323e-05,
+      "loss": 0.1804,
+      "step": 17350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5119899032393773e-05,
+      "loss": 0.1901,
+      "step": 17400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5105875753751226e-05,
+      "loss": 0.2045,
+      "step": 17450
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5091852475108683e-05,
+      "loss": 0.1796,
+      "step": 17500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5077829196466133e-05,
+      "loss": 0.1763,
+      "step": 17550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.506380591782359e-05,
+      "loss": 0.1882,
+      "step": 17600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5049782639181043e-05,
+      "loss": 0.191,
+      "step": 17650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5035759360538493e-05,
+      "loss": 0.1646,
+      "step": 17700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.502173608189595e-05,
+      "loss": 0.1898,
+      "step": 17750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.50077128032534e-05,
+      "loss": 0.1717,
+      "step": 17800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4993689524610853e-05,
+      "loss": 0.1676,
+      "step": 17850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.497966624596831e-05,
+      "loss": 0.1978,
+      "step": 17900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.496564296732576e-05,
+      "loss": 0.216,
+      "step": 17950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4951619688683216e-05,
+      "loss": 0.2025,
+      "step": 18000
+    },
+    {
+      "epoch": 0.5,
+      "eval_bleu": 95.6412,
+      "eval_gen_len": 64.4067,
+      "eval_loss": 0.27226048707962036,
+      "eval_rouge1": 93.2987,
+      "eval_rouge2": 89.3773,
+      "eval_rougeL": 93.084,
+      "eval_rougeLsum": 93.0595,
+      "eval_runtime": 3091.5775,
+      "eval_samples_per_second": 0.97,
+      "eval_steps_per_second": 0.243,
+      "step": 18000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.493759641004067e-05,
+      "loss": 0.1948,
+      "step": 18050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.492357313139812e-05,
+      "loss": 0.1896,
+      "step": 18100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4909549852755576e-05,
+      "loss": 0.1767,
+      "step": 18150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4895526574113026e-05,
+      "loss": 0.2107,
+      "step": 18200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.488150329547048e-05,
+      "loss": 0.1878,
+      "step": 18250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4867480016827936e-05,
+      "loss": 0.1735,
+      "step": 18300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4853456738185386e-05,
+      "loss": 0.1607,
+      "step": 18350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4839433459542843e-05,
+      "loss": 0.1611,
+      "step": 18400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4825410180900297e-05,
+      "loss": 0.1889,
+      "step": 18450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4811386902257746e-05,
+      "loss": 0.1962,
+      "step": 18500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4797363623615203e-05,
+      "loss": 0.1767,
+      "step": 18550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4783340344972653e-05,
+      "loss": 0.2027,
+      "step": 18600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.476931706633011e-05,
+      "loss": 0.1994,
+      "step": 18650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.4755293787687563e-05,
+      "loss": 0.1725,
+      "step": 18700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4741270509045013e-05,
+      "loss": 0.1931,
+      "step": 18750
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.472724723040247e-05,
+      "loss": 0.1897,
+      "step": 18800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4713223951759923e-05,
+      "loss": 0.191,
+      "step": 18850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4699200673117373e-05,
+      "loss": 0.1868,
+      "step": 18900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.468517739447483e-05,
+      "loss": 0.1738,
+      "step": 18950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.467115411583228e-05,
+      "loss": 0.1808,
+      "step": 19000
+    },
+    {
+      "epoch": 0.53,
+      "eval_bleu": 95.6749,
+      "eval_gen_len": 64.317,
+      "eval_loss": 0.2700382173061371,
+      "eval_rouge1": 93.4064,
+      "eval_rouge2": 89.5527,
+      "eval_rougeL": 93.2047,
+      "eval_rougeLsum": 93.1846,
+      "eval_runtime": 3097.9008,
+      "eval_samples_per_second": 0.968,
+      "eval_steps_per_second": 0.242,
+      "step": 19000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.4657130837189737e-05,
+      "loss": 0.1747,
+      "step": 19050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.464310755854719e-05,
+      "loss": 0.189,
+      "step": 19100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.462908427990464e-05,
+      "loss": 0.172,
+      "step": 19150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4615061001262097e-05,
+      "loss": 0.1793,
+      "step": 19200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.460103772261955e-05,
+      "loss": 0.1855,
+      "step": 19250
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4587014443977e-05,
+      "loss": 0.1845,
+      "step": 19300
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4572991165334457e-05,
+      "loss": 0.1907,
+      "step": 19350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.4558967886691907e-05,
+      "loss": 0.1713,
+      "step": 19400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4544944608049364e-05,
+      "loss": 0.1983,
+      "step": 19450
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4530921329406817e-05,
+      "loss": 0.1654,
+      "step": 19500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4516898050764267e-05,
+      "loss": 0.2049,
+      "step": 19550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4502874772121724e-05,
+      "loss": 0.1826,
+      "step": 19600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4488851493479177e-05,
+      "loss": 0.1622,
+      "step": 19650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4474828214836627e-05,
+      "loss": 0.196,
+      "step": 19700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.4460804936194084e-05,
+      "loss": 0.1805,
+      "step": 19750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4446781657551534e-05,
+      "loss": 0.176,
+      "step": 19800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.443275837890899e-05,
+      "loss": 0.1756,
+      "step": 19850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4418735100266444e-05,
+      "loss": 0.2069,
+      "step": 19900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4404711821623894e-05,
+      "loss": 0.2099,
+      "step": 19950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.439068854298135e-05,
+      "loss": 0.1844,
+      "step": 20000
+    },
+    {
+      "epoch": 0.56,
+      "eval_bleu": 95.7094,
+      "eval_gen_len": 64.34,
+      "eval_loss": 0.2679564356803894,
+      "eval_rouge1": 93.3936,
+      "eval_rouge2": 89.5638,
+      "eval_rougeL": 93.2067,
+      "eval_rougeLsum": 93.1818,
+      "eval_runtime": 3170.5236,
+      "eval_samples_per_second": 0.946,
+      "eval_steps_per_second": 0.237,
+      "step": 20000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4376665264338804e-05,
+      "loss": 0.1718,
+      "step": 20050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.4362641985696257e-05,
+      "loss": 0.1938,
+      "step": 20100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.434861870705371e-05,
+      "loss": 0.1808,
+      "step": 20150
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.433459542841116e-05,
+      "loss": 0.1805,
+      "step": 20200
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.4320572149768617e-05,
+      "loss": 0.1761,
+      "step": 20250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.430654887112607e-05,
+      "loss": 0.1826,
+      "step": 20300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.429252559248352e-05,
+      "loss": 0.1822,
+      "step": 20350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.4278502313840977e-05,
+      "loss": 0.1838,
+      "step": 20400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.426447903519843e-05,
+      "loss": 0.1723,
+      "step": 20450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.4250455756555884e-05,
+      "loss": 0.1816,
+      "step": 20500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4236432477913338e-05,
+      "loss": 0.1973,
+      "step": 20550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4222409199270788e-05,
+      "loss": 0.2059,
+      "step": 20600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4208385920628244e-05,
+      "loss": 0.1743,
+      "step": 20650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4194362641985698e-05,
+      "loss": 0.2011,
+      "step": 20700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4180339363343148e-05,
+      "loss": 0.1914,
+      "step": 20750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4166316084700604e-05,
+      "loss": 0.1716,
+      "step": 20800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.4152292806058058e-05,
+      "loss": 0.179,
+      "step": 20850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.413826952741551e-05,
+      "loss": 0.2066,
+      "step": 20900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4124246248772964e-05,
+      "loss": 0.1964,
+      "step": 20950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4110222970130414e-05,
+      "loss": 0.1892,
+      "step": 21000
+    },
+    {
+      "epoch": 0.59,
+      "eval_bleu": 95.7337,
+      "eval_gen_len": 64.2923,
+      "eval_loss": 0.2668701708316803,
+      "eval_rouge1": 93.4236,
+      "eval_rouge2": 89.643,
+      "eval_rougeL": 93.2265,
+      "eval_rougeLsum": 93.2057,
+      "eval_runtime": 3109.7477,
+      "eval_samples_per_second": 0.965,
+      "eval_steps_per_second": 0.241,
+      "step": 21000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.409619969148787e-05,
+      "loss": 0.1852,
+      "step": 21050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4082176412845324e-05,
+      "loss": 0.1833,
+      "step": 21100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.4068153134202774e-05,
+      "loss": 0.1668,
+      "step": 21150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.405412985556023e-05,
+      "loss": 0.1971,
+      "step": 21200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4040106576917685e-05,
+      "loss": 0.1819,
+      "step": 21250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.4026083298275138e-05,
+      "loss": 0.1922,
+      "step": 21300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.401206001963259e-05,
+      "loss": 0.1799,
+      "step": 21350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.399803674099004e-05,
+      "loss": 0.1892,
+      "step": 21400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.3984013462347498e-05,
+      "loss": 0.1631,
+      "step": 21450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.396999018370495e-05,
+      "loss": 0.1728,
+      "step": 21500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.3955966905062405e-05,
+      "loss": 0.1833,
+      "step": 21550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.3941943626419858e-05,
+      "loss": 0.1988,
+      "step": 21600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.392792034777731e-05,
+      "loss": 0.176,
+      "step": 21650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.3913897069134765e-05,
+      "loss": 0.1664,
+      "step": 21700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.3899873790492218e-05,
+      "loss": 0.1749,
+      "step": 21750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.3885850511849668e-05,
+      "loss": 0.1856,
+      "step": 21800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.3871827233207125e-05,
+      "loss": 0.1845,
+      "step": 21850
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.3857803954564578e-05,
+      "loss": 0.1823,
+      "step": 21900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.384378067592203e-05,
+      "loss": 0.1882,
+      "step": 21950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3829757397279485e-05,
+      "loss": 0.1754,
+      "step": 22000
+    },
+    {
+      "epoch": 0.62,
+      "eval_bleu": 95.7216,
+      "eval_gen_len": 64.29,
+      "eval_loss": 0.26628053188323975,
+      "eval_rouge1": 93.494,
+      "eval_rouge2": 89.7041,
+      "eval_rougeL": 93.2775,
+      "eval_rougeLsum": 93.255,
+      "eval_runtime": 3108.636,
+      "eval_samples_per_second": 0.965,
+      "eval_steps_per_second": 0.241,
+      "step": 22000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3815734118636938e-05,
+      "loss": 0.1727,
+      "step": 22050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.380171083999439e-05,
+      "loss": 0.2006,
+      "step": 22100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3787687561351845e-05,
+      "loss": 0.1731,
+      "step": 22150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3773664282709295e-05,
+      "loss": 0.1718,
+      "step": 22200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.3759641004066752e-05,
+      "loss": 0.1935,
+      "step": 22250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3745617725424205e-05,
+      "loss": 0.1855,
+      "step": 22300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.373159444678166e-05,
+      "loss": 0.1788,
+      "step": 22350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3717571168139112e-05,
+      "loss": 0.1722,
+      "step": 22400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3703547889496565e-05,
+      "loss": 0.1685,
+      "step": 22450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.368952461085402e-05,
+      "loss": 0.1998,
+      "step": 22500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3675501332211472e-05,
+      "loss": 0.1891,
+      "step": 22550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.3661478053568922e-05,
+      "loss": 0.1865,
+      "step": 22600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.364745477492638e-05,
+      "loss": 0.2003,
+      "step": 22650
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3633431496283832e-05,
+      "loss": 0.1833,
+      "step": 22700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3619408217641285e-05,
+      "loss": 0.1709,
+      "step": 22750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.360538493899874e-05,
+      "loss": 0.1784,
+      "step": 22800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3591361660356192e-05,
+      "loss": 0.1976,
+      "step": 22850
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.3577338381713645e-05,
+      "loss": 0.2096,
+      "step": 22900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.35633151030711e-05,
+      "loss": 0.1795,
+      "step": 22950
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3549291824428552e-05,
+      "loss": 0.1843,
+      "step": 23000
+    },
+    {
+      "epoch": 0.65,
+      "eval_bleu": 95.7705,
+      "eval_gen_len": 64.281,
+      "eval_loss": 0.263296902179718,
+      "eval_rouge1": 93.5248,
+      "eval_rouge2": 89.7445,
+      "eval_rougeL": 93.3203,
+      "eval_rougeLsum": 93.3217,
+      "eval_runtime": 3090.4348,
+      "eval_samples_per_second": 0.971,
+      "eval_steps_per_second": 0.243,
+      "step": 23000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3535268545786005e-05,
+      "loss": 0.171,
+      "step": 23050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.352124526714346e-05,
+      "loss": 0.1777,
+      "step": 23100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3507221988500912e-05,
+      "loss": 0.2083,
+      "step": 23150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3493198709858365e-05,
+      "loss": 0.163,
+      "step": 23200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.347917543121582e-05,
+      "loss": 0.1884,
+      "step": 23250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3465152152573272e-05,
+      "loss": 0.1772,
+      "step": 23300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.3451128873930726e-05,
+      "loss": 0.1746,
+      "step": 23350
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.343710559528818e-05,
+      "loss": 0.1971,
+      "step": 23400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3423082316645632e-05,
+      "loss": 0.1697,
+      "step": 23450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3409059038003086e-05,
+      "loss": 0.1635,
+      "step": 23500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.339503575936054e-05,
+      "loss": 0.1725,
+      "step": 23550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3381012480717992e-05,
+      "loss": 0.1669,
+      "step": 23600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.3366989202075446e-05,
+      "loss": 0.1871,
+      "step": 23650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.33529659234329e-05,
+      "loss": 0.1641,
+      "step": 23700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3338942644790352e-05,
+      "loss": 0.1909,
+      "step": 23750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3324919366147806e-05,
+      "loss": 0.1782,
+      "step": 23800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.331089608750526e-05,
+      "loss": 0.1802,
+      "step": 23850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3296872808862713e-05,
+      "loss": 0.1669,
+      "step": 23900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3282849530220166e-05,
+      "loss": 0.1745,
+      "step": 23950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.326882625157762e-05,
+      "loss": 0.1607,
+      "step": 24000
+    },
+    {
+      "epoch": 0.67,
+      "eval_bleu": 95.7802,
+      "eval_gen_len": 64.2037,
+      "eval_loss": 0.26373326778411865,
+      "eval_rouge1": 93.5949,
+      "eval_rouge2": 89.9557,
+      "eval_rougeL": 93.4252,
+      "eval_rougeLsum": 93.3955,
+      "eval_runtime": 3090.8367,
+      "eval_samples_per_second": 0.971,
+      "eval_steps_per_second": 0.243,
+      "step": 24000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.3254802972935073e-05,
+      "loss": 0.1758,
+      "step": 24050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3240779694292526e-05,
+      "loss": 0.1948,
+      "step": 24100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.322675641564998e-05,
+      "loss": 0.1836,
+      "step": 24150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3212733137007433e-05,
+      "loss": 0.1999,
+      "step": 24200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3198709858364886e-05,
+      "loss": 0.181,
+      "step": 24250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.318468657972234e-05,
+      "loss": 0.1729,
+      "step": 24300
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3170663301079793e-05,
+      "loss": 0.1814,
+      "step": 24350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3156640022437246e-05,
+      "loss": 0.1751,
+      "step": 24400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3142616743794703e-05,
+      "loss": 0.1655,
+      "step": 24450
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3128593465152153e-05,
+      "loss": 0.1742,
+      "step": 24500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3114570186509606e-05,
+      "loss": 0.154,
+      "step": 24550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.310054690786706e-05,
+      "loss": 0.1758,
+      "step": 24600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3086523629224513e-05,
+      "loss": 0.1921,
+      "step": 24650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3072500350581966e-05,
+      "loss": 0.1803,
+      "step": 24700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.305847707193942e-05,
+      "loss": 0.2012,
+      "step": 24750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3044453793296873e-05,
+      "loss": 0.163,
+      "step": 24800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.303043051465433e-05,
+      "loss": 0.1679,
+      "step": 24850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.301640723601178e-05,
+      "loss": 0.1846,
+      "step": 24900
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3002383957369233e-05,
+      "loss": 0.1813,
+      "step": 24950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2988360678726686e-05,
+      "loss": 0.1583,
+      "step": 25000
+    },
+    {
+      "epoch": 0.7,
+      "eval_bleu": 95.7853,
+      "eval_gen_len": 64.1823,
+      "eval_loss": 0.26211991906166077,
+      "eval_rouge1": 93.6077,
+      "eval_rouge2": 89.8774,
+      "eval_rougeL": 93.4105,
+      "eval_rougeLsum": 93.3911,
+      "eval_runtime": 3088.3555,
+      "eval_samples_per_second": 0.971,
+      "eval_steps_per_second": 0.243,
+      "step": 25000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.297433740008414e-05,
+      "loss": 0.203,
+      "step": 25050
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2960314121441593e-05,
+      "loss": 0.1676,
+      "step": 25100
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2946290842799046e-05,
+      "loss": 0.1866,
+      "step": 25150
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.29322675641565e-05,
+      "loss": 0.1928,
+      "step": 25200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2918244285513957e-05,
+      "loss": 0.1836,
+      "step": 25250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2904221006871407e-05,
+      "loss": 0.1706,
+      "step": 25300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.289019772822886e-05,
+      "loss": 0.2018,
+      "step": 25350
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2876174449586313e-05,
+      "loss": 0.1771,
+      "step": 25400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2862151170943767e-05,
+      "loss": 0.1808,
+      "step": 25450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.284812789230122e-05,
+      "loss": 0.1968,
+      "step": 25500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2834104613658673e-05,
+      "loss": 0.1776,
+      "step": 25550
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2820081335016127e-05,
+      "loss": 0.1657,
+      "step": 25600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2806058056373583e-05,
+      "loss": 0.1627,
+      "step": 25650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2792034777731033e-05,
+      "loss": 0.1687,
+      "step": 25700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2778011499088487e-05,
+      "loss": 0.1691,
+      "step": 25750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.276398822044594e-05,
+      "loss": 0.1875,
+      "step": 25800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2749964941803393e-05,
+      "loss": 0.1757,
+      "step": 25850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.273594166316085e-05,
+      "loss": 0.2009,
+      "step": 25900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.27219183845183e-05,
+      "loss": 0.1762,
+      "step": 25950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2707895105875754e-05,
+      "loss": 0.1533,
+      "step": 26000
+    },
+    {
+      "epoch": 0.73,
+      "eval_bleu": 95.8266,
+      "eval_gen_len": 64.2703,
+      "eval_loss": 0.26142334938049316,
+      "eval_rouge1": 93.5866,
+      "eval_rouge2": 89.888,
+      "eval_rougeL": 93.3841,
+      "eval_rougeLsum": 93.3821,
+      "eval_runtime": 3090.7679,
+      "eval_samples_per_second": 0.971,
+      "eval_steps_per_second": 0.243,
+      "step": 26000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.269387182723321e-05,
+      "loss": 0.1756,
+      "step": 26050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.267984854859066e-05,
+      "loss": 0.1793,
+      "step": 26100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2665825269948114e-05,
+      "loss": 0.1646,
+      "step": 26150
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.2651801991305567e-05,
+      "loss": 0.1537,
+      "step": 26200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.263777871266302e-05,
+      "loss": 0.1593,
+      "step": 26250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.2623755434020477e-05,
+      "loss": 0.182,
+      "step": 26300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.2609732155377927e-05,
+      "loss": 0.1721,
+      "step": 26350
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.259570887673538e-05,
+      "loss": 0.1922,
+      "step": 26400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.2581685598092837e-05,
+      "loss": 0.1703,
+      "step": 26450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.2567662319450287e-05,
+      "loss": 0.1731,
+      "step": 26500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.255363904080774e-05,
+      "loss": 0.1689,
+      "step": 26550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2539615762165194e-05,
+      "loss": 0.1537,
+      "step": 26600
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2525592483522647e-05,
+      "loss": 0.161,
+      "step": 26650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2511569204880104e-05,
+      "loss": 0.1812,
+      "step": 26700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2497545926237554e-05,
+      "loss": 0.188,
+      "step": 26750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2483522647595007e-05,
+      "loss": 0.2057,
+      "step": 26800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2469499368952464e-05,
+      "loss": 0.1778,
+      "step": 26850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.2455476090309914e-05,
+      "loss": 0.1757,
+      "step": 26900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2441452811667367e-05,
+      "loss": 0.196,
+      "step": 26950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.242742953302482e-05,
+      "loss": 0.1552,
+      "step": 27000
+    },
+    {
+      "epoch": 0.76,
+      "eval_bleu": 95.8418,
+      "eval_gen_len": 64.2587,
+      "eval_loss": 0.2597999572753906,
+      "eval_rouge1": 93.5569,
+      "eval_rouge2": 89.8854,
+      "eval_rougeL": 93.3668,
+      "eval_rougeLsum": 93.3569,
+      "eval_runtime": 3122.2549,
+      "eval_samples_per_second": 0.961,
+      "eval_steps_per_second": 0.24,
+      "step": 27000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2413406254382274e-05,
+      "loss": 0.1963,
+      "step": 27050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.239938297573973e-05,
+      "loss": 0.1837,
+      "step": 27100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.238535969709718e-05,
+      "loss": 0.1722,
+      "step": 27150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.2371336418454634e-05,
+      "loss": 0.1698,
+      "step": 27200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.235731313981209e-05,
+      "loss": 0.1932,
+      "step": 27250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.234328986116954e-05,
+      "loss": 0.1519,
+      "step": 27300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2329266582526998e-05,
+      "loss": 0.1744,
+      "step": 27350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2315243303884448e-05,
+      "loss": 0.1672,
+      "step": 27400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.23012200252419e-05,
+      "loss": 0.1641,
+      "step": 27450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2287196746599358e-05,
+      "loss": 0.1815,
+      "step": 27500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.2273173467956808e-05,
+      "loss": 0.1567,
+      "step": 27550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.225915018931426e-05,
+      "loss": 0.1621,
+      "step": 27600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2245126910671718e-05,
+      "loss": 0.1489,
+      "step": 27650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2231103632029168e-05,
+      "loss": 0.1692,
+      "step": 27700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2217080353386624e-05,
+      "loss": 0.1856,
+      "step": 27750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2203057074744074e-05,
+      "loss": 0.169,
+      "step": 27800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2189033796101528e-05,
+      "loss": 0.1749,
+      "step": 27850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2175010517458985e-05,
+      "loss": 0.1764,
+      "step": 27900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.2160987238816434e-05,
+      "loss": 0.1578,
+      "step": 27950
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2146963960173888e-05,
+      "loss": 0.1749,
+      "step": 28000
+    },
+    {
+      "epoch": 0.79,
+      "eval_bleu": 95.8187,
+      "eval_gen_len": 64.232,
+      "eval_loss": 0.25857558846473694,
+      "eval_rouge1": 93.7097,
+      "eval_rouge2": 89.9991,
+      "eval_rougeL": 93.5074,
+      "eval_rougeLsum": 93.4927,
+      "eval_runtime": 3115.4186,
+      "eval_samples_per_second": 0.963,
+      "eval_steps_per_second": 0.241,
+      "step": 28000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2132940681531345e-05,
+      "loss": 0.1808,
+      "step": 28050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2118917402888795e-05,
+      "loss": 0.1781,
+      "step": 28100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.210489412424625e-05,
+      "loss": 0.1684,
+      "step": 28150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.20908708456037e-05,
+      "loss": 0.1749,
+      "step": 28200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2076847566961155e-05,
+      "loss": 0.1758,
+      "step": 28250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206282428831861e-05,
+      "loss": 0.1811,
+      "step": 28300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.204880100967606e-05,
+      "loss": 0.1652,
+      "step": 28350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.2034777731033515e-05,
+      "loss": 0.1735,
+      "step": 28400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.202075445239097e-05,
+      "loss": 0.1735,
+      "step": 28450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.200673117374842e-05,
+      "loss": 0.1881,
+      "step": 28500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1992707895105878e-05,
+      "loss": 0.1892,
+      "step": 28550
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1978684616463328e-05,
+      "loss": 0.1665,
+      "step": 28600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.196466133782078e-05,
+      "loss": 0.1866,
+      "step": 28650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1950638059178238e-05,
+      "loss": 0.1756,
+      "step": 28700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1936614780535688e-05,
+      "loss": 0.1945,
+      "step": 28750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1922591501893145e-05,
+      "loss": 0.1882,
+      "step": 28800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1908568223250598e-05,
+      "loss": 0.1693,
+      "step": 28850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1894544944608048e-05,
+      "loss": 0.1653,
+      "step": 28900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1880521665965505e-05,
+      "loss": 0.1688,
+      "step": 28950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.1866498387322955e-05,
+      "loss": 0.1725,
+      "step": 29000
+    },
+    {
+      "epoch": 0.81,
+      "eval_bleu": 95.8531,
+      "eval_gen_len": 64.207,
+      "eval_loss": 0.25807130336761475,
+      "eval_rouge1": 93.6865,
+      "eval_rouge2": 90.02,
+      "eval_rougeL": 93.4949,
+      "eval_rougeLsum": 93.4886,
+      "eval_runtime": 3103.2566,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 29000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.185247510868041e-05,
+      "loss": 0.1597,
+      "step": 29050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1838451830037865e-05,
+      "loss": 0.1639,
+      "step": 29100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1824428551395315e-05,
+      "loss": 0.1748,
+      "step": 29150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1810405272752772e-05,
+      "loss": 0.1734,
+      "step": 29200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1796381994110225e-05,
+      "loss": 0.1783,
+      "step": 29250
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1782358715467675e-05,
+      "loss": 0.1872,
+      "step": 29300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1768335436825132e-05,
+      "loss": 0.1727,
+      "step": 29350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.1754312158182582e-05,
+      "loss": 0.1607,
+      "step": 29400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.1740288879540035e-05,
+      "loss": 0.1702,
+      "step": 29450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.1726265600897492e-05,
+      "loss": 0.162,
+      "step": 29500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.1712242322254942e-05,
+      "loss": 0.2007,
+      "step": 29550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.16982190436124e-05,
+      "loss": 0.2098,
+      "step": 29600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.1684195764969852e-05,
+      "loss": 0.1707,
+      "step": 29650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.1670172486327302e-05,
+      "loss": 0.1658,
+      "step": 29700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.165614920768476e-05,
+      "loss": 0.1736,
+      "step": 29750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.164212592904221e-05,
+      "loss": 0.1759,
+      "step": 29800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.1628102650399662e-05,
+      "loss": 0.1915,
+      "step": 29850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.161407937175712e-05,
+      "loss": 0.1755,
+      "step": 29900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.160005609311457e-05,
+      "loss": 0.171,
+      "step": 29950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.1586032814472026e-05,
+      "loss": 0.1917,
+      "step": 30000
+    },
+    {
+      "epoch": 0.84,
+      "eval_bleu": 95.8729,
+      "eval_gen_len": 64.2307,
+      "eval_loss": 0.2573055922985077,
+      "eval_rouge1": 93.7203,
+      "eval_rouge2": 90.0218,
+      "eval_rougeL": 93.5266,
+      "eval_rougeLsum": 93.517,
+      "eval_runtime": 3112.1338,
+      "eval_samples_per_second": 0.964,
+      "eval_steps_per_second": 0.241,
+      "step": 30000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.157200953582948e-05,
+      "loss": 0.165,
+      "step": 30050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.155798625718693e-05,
+      "loss": 0.1712,
+      "step": 30100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1543962978544386e-05,
+      "loss": 0.1821,
+      "step": 30150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1529939699901836e-05,
+      "loss": 0.1696,
+      "step": 30200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1515916421259292e-05,
+      "loss": 0.1666,
+      "step": 30250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1501893142616746e-05,
+      "loss": 0.1935,
+      "step": 30300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1487869863974196e-05,
+      "loss": 0.1819,
+      "step": 30350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1473846585331652e-05,
+      "loss": 0.1606,
+      "step": 30400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.1459823306689106e-05,
+      "loss": 0.1591,
+      "step": 30450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1445800028046556e-05,
+      "loss": 0.1602,
+      "step": 30500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1431776749404012e-05,
+      "loss": 0.1569,
+      "step": 30550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1417753470761462e-05,
+      "loss": 0.1721,
+      "step": 30600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.140373019211892e-05,
+      "loss": 0.1757,
+      "step": 30650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1389706913476373e-05,
+      "loss": 0.192,
+      "step": 30700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.1375683634833823e-05,
+      "loss": 0.1957,
+      "step": 30750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.136166035619128e-05,
+      "loss": 0.1645,
+      "step": 30800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1347637077548733e-05,
+      "loss": 0.1751,
+      "step": 30850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1333613798906183e-05,
+      "loss": 0.1767,
+      "step": 30900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.131959052026364e-05,
+      "loss": 0.1697,
+      "step": 30950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.130556724162109e-05,
+      "loss": 0.1897,
+      "step": 31000
+    },
+    {
+      "epoch": 0.87,
+      "eval_bleu": 95.8468,
+      "eval_gen_len": 64.2037,
+      "eval_loss": 0.25732484459877014,
+      "eval_rouge1": 93.6811,
+      "eval_rouge2": 89.9877,
+      "eval_rougeL": 93.478,
+      "eval_rougeLsum": 93.4745,
+      "eval_runtime": 3156.1925,
+      "eval_samples_per_second": 0.951,
+      "eval_steps_per_second": 0.238,
+      "step": 31000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1291543962978546e-05,
+      "loss": 0.1742,
+      "step": 31050
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1277520684336e-05,
+      "loss": 0.1663,
+      "step": 31100
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.126349740569345e-05,
+      "loss": 0.1831,
+      "step": 31150
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1249474127050906e-05,
+      "loss": 0.1653,
+      "step": 31200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.123545084840836e-05,
+      "loss": 0.1739,
+      "step": 31250
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.122142756976581e-05,
+      "loss": 0.1741,
+      "step": 31300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1207404291123266e-05,
+      "loss": 0.1755,
+      "step": 31350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1193381012480716e-05,
+      "loss": 0.164,
+      "step": 31400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1179357733838173e-05,
+      "loss": 0.1624,
+      "step": 31450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1165334455195626e-05,
+      "loss": 0.1716,
+      "step": 31500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.1151311176553076e-05,
+      "loss": 0.1759,
+      "step": 31550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1137287897910533e-05,
+      "loss": 0.1845,
+      "step": 31600
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1123264619267986e-05,
+      "loss": 0.1637,
+      "step": 31650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.110924134062544e-05,
+      "loss": 0.1939,
+      "step": 31700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1095218061982893e-05,
+      "loss": 0.1747,
+      "step": 31750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1081194783340343e-05,
+      "loss": 0.1727,
+      "step": 31800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.10671715046978e-05,
+      "loss": 0.1626,
+      "step": 31850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1053148226055253e-05,
+      "loss": 0.1885,
+      "step": 31900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1039124947412703e-05,
+      "loss": 0.1812,
+      "step": 31950
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.102510166877016e-05,
+      "loss": 0.1925,
+      "step": 32000
+    },
+    {
+      "epoch": 0.9,
+      "eval_bleu": 95.8275,
+      "eval_gen_len": 64.1743,
+      "eval_loss": 0.2557416260242462,
+      "eval_rouge1": 93.6863,
+      "eval_rouge2": 89.9362,
+      "eval_rougeL": 93.482,
+      "eval_rougeLsum": 93.4648,
+      "eval_runtime": 3095.3083,
+      "eval_samples_per_second": 0.969,
+      "eval_steps_per_second": 0.242,
+      "step": 32000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1011078390127613e-05,
+      "loss": 0.1858,
+      "step": 32050
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0997055111485067e-05,
+      "loss": 0.1588,
+      "step": 32100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.098303183284252e-05,
+      "loss": 0.1797,
+      "step": 32150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.096900855419997e-05,
+      "loss": 0.1846,
+      "step": 32200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0954985275557427e-05,
+      "loss": 0.185,
+      "step": 32250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.094096199691488e-05,
+      "loss": 0.1565,
+      "step": 32300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.092693871827233e-05,
+      "loss": 0.2177,
+      "step": 32350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0912915439629787e-05,
+      "loss": 0.1861,
+      "step": 32400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.089889216098724e-05,
+      "loss": 0.1667,
+      "step": 32450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0884868882344693e-05,
+      "loss": 0.1788,
+      "step": 32500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0870845603702147e-05,
+      "loss": 0.2086,
+      "step": 32550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0856822325059597e-05,
+      "loss": 0.1607,
+      "step": 32600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0842799046417053e-05,
+      "loss": 0.1655,
+      "step": 32650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0828775767774507e-05,
+      "loss": 0.1798,
+      "step": 32700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0814752489131957e-05,
+      "loss": 0.1953,
+      "step": 32750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0800729210489414e-05,
+      "loss": 0.166,
+      "step": 32800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0786705931846867e-05,
+      "loss": 0.1718,
+      "step": 32850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.077268265320432e-05,
+      "loss": 0.1888,
+      "step": 32900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0758659374561774e-05,
+      "loss": 0.1788,
+      "step": 32950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0744636095919224e-05,
+      "loss": 0.1853,
+      "step": 33000
+    },
+    {
+      "epoch": 0.93,
+      "eval_bleu": 95.8702,
+      "eval_gen_len": 64.183,
+      "eval_loss": 0.2551117241382599,
+      "eval_rouge1": 93.7191,
+      "eval_rouge2": 90.0474,
+      "eval_rougeL": 93.5286,
+      "eval_rougeLsum": 93.516,
+      "eval_runtime": 3134.3467,
+      "eval_samples_per_second": 0.957,
+      "eval_steps_per_second": 0.239,
+      "step": 33000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.073061281727668e-05,
+      "loss": 0.1964,
+      "step": 33050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0716589538634134e-05,
+      "loss": 0.178,
+      "step": 33100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0702566259991587e-05,
+      "loss": 0.1911,
+      "step": 33150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.068854298134904e-05,
+      "loss": 0.1705,
+      "step": 33200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0674519702706494e-05,
+      "loss": 0.1835,
+      "step": 33250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0660496424063947e-05,
+      "loss": 0.1617,
+      "step": 33300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.06464731454214e-05,
+      "loss": 0.169,
+      "step": 33350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.063244986677885e-05,
+      "loss": 0.1738,
+      "step": 33400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0618426588136307e-05,
+      "loss": 0.1648,
+      "step": 33450
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.060440330949376e-05,
+      "loss": 0.1822,
+      "step": 33500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0590380030851214e-05,
+      "loss": 0.1719,
+      "step": 33550
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0576356752208667e-05,
+      "loss": 0.1899,
+      "step": 33600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.056233347356612e-05,
+      "loss": 0.1754,
+      "step": 33650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0548310194923574e-05,
+      "loss": 0.1693,
+      "step": 33700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0534286916281027e-05,
+      "loss": 0.1578,
+      "step": 33750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0520263637638477e-05,
+      "loss": 0.1757,
+      "step": 33800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0506240358995934e-05,
+      "loss": 0.1735,
+      "step": 33850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0492217080353387e-05,
+      "loss": 0.1732,
+      "step": 33900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.047819380171084e-05,
+      "loss": 0.1645,
+      "step": 33950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0464170523068294e-05,
+      "loss": 0.166,
+      "step": 34000
+    },
+    {
+      "epoch": 0.95,
+      "eval_bleu": 95.8966,
+      "eval_gen_len": 64.227,
+      "eval_loss": 0.2542245090007782,
+      "eval_rouge1": 93.6823,
+      "eval_rouge2": 89.9822,
+      "eval_rougeL": 93.4675,
+      "eval_rougeLsum": 93.4534,
+      "eval_runtime": 3105.0348,
+      "eval_samples_per_second": 0.966,
+      "eval_steps_per_second": 0.242,
+      "step": 34000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0450147244425747e-05,
+      "loss": 0.1736,
+      "step": 34050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.04361239657832e-05,
+      "loss": 0.1888,
+      "step": 34100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0422100687140654e-05,
+      "loss": 0.1701,
+      "step": 34150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0408077408498108e-05,
+      "loss": 0.151,
+      "step": 34200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.039405412985556e-05,
+      "loss": 0.1776,
+      "step": 34250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0380030851213014e-05,
+      "loss": 0.1928,
+      "step": 34300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0366007572570468e-05,
+      "loss": 0.1762,
+      "step": 34350
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.035198429392792e-05,
+      "loss": 0.1631,
+      "step": 34400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0337961015285374e-05,
+      "loss": 0.1913,
+      "step": 34450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0323937736642828e-05,
+      "loss": 0.1662,
+      "step": 34500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.030991445800028e-05,
+      "loss": 0.1691,
+      "step": 34550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0295891179357734e-05,
+      "loss": 0.1664,
+      "step": 34600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0281867900715188e-05,
+      "loss": 0.1526,
+      "step": 34650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.026784462207264e-05,
+      "loss": 0.1816,
+      "step": 34700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0253821343430095e-05,
+      "loss": 0.1892,
+      "step": 34750
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0239798064787548e-05,
+      "loss": 0.1632,
+      "step": 34800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0225774786145e-05,
+      "loss": 0.1542,
+      "step": 34850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0211751507502455e-05,
+      "loss": 0.1529,
+      "step": 34900
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0197728228859908e-05,
+      "loss": 0.1597,
+      "step": 34950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018370495021736e-05,
+      "loss": 0.1795,
+      "step": 35000
+    },
+    {
+      "epoch": 0.98,
+      "eval_bleu": 95.9192,
+      "eval_gen_len": 64.2247,
+      "eval_loss": 0.2535455822944641,
+      "eval_rouge1": 93.7504,
+      "eval_rouge2": 90.1079,
+      "eval_rougeL": 93.5482,
+      "eval_rougeLsum": 93.5289,
+      "eval_runtime": 3098.1952,
+      "eval_samples_per_second": 0.968,
+      "eval_steps_per_second": 0.242,
+      "step": 35000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0169681671574815e-05,
+      "loss": 0.1634,
+      "step": 35050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0155658392932268e-05,
+      "loss": 0.1723,
+      "step": 35100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.014163511428972e-05,
+      "loss": 0.1624,
+      "step": 35150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0127611835647175e-05,
+      "loss": 0.1437,
+      "step": 35200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0113588557004628e-05,
+      "loss": 0.1638,
+      "step": 35250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009956527836208e-05,
+      "loss": 0.153,
+      "step": 35300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0085541999719535e-05,
+      "loss": 0.1749,
+      "step": 35350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0071518721076988e-05,
+      "loss": 0.1802,
+      "step": 35400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005749544243444e-05,
+      "loss": 0.1794,
+      "step": 35450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0043472163791895e-05,
+      "loss": 0.1778,
+      "step": 35500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0029448885149348e-05,
+      "loss": 0.1608,
+      "step": 35550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00154256065068e-05,
+      "loss": 0.1722,
+      "step": 35600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001402327864255e-05,
+      "loss": 0.1674,
+      "step": 35650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9987379049221708e-05,
+      "loss": 0.1786,
+      "step": 35700
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.997335577057916e-05,
+      "loss": 0.1597,
+      "step": 35750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9959332491936615e-05,
+      "loss": 0.1684,
+      "step": 35800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.994530921329407e-05,
+      "loss": 0.1526,
+      "step": 35850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9931285934651522e-05,
+      "loss": 0.1523,
+      "step": 35900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9917262656008975e-05,
+      "loss": 0.1743,
+      "step": 35950
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.990323937736643e-05,
+      "loss": 0.152,
+      "step": 36000
+    },
+    {
+      "epoch": 1.01,
+      "eval_bleu": 95.9457,
+      "eval_gen_len": 64.2747,
+      "eval_loss": 0.2530899941921234,
+      "eval_rouge1": 93.7477,
+      "eval_rouge2": 90.0914,
+      "eval_rougeL": 93.5409,
+      "eval_rougeLsum": 93.5214,
+      "eval_runtime": 3128.4817,
+      "eval_samples_per_second": 0.959,
+      "eval_steps_per_second": 0.24,
+      "step": 36000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9889216098723885e-05,
+      "loss": 0.1547,
+      "step": 36050
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9875192820081335e-05,
+      "loss": 0.1586,
+      "step": 36100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.986116954143879e-05,
+      "loss": 0.1778,
+      "step": 36150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9847146262796242e-05,
+      "loss": 0.1466,
+      "step": 36200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9833122984153695e-05,
+      "loss": 0.1561,
+      "step": 36250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.981909970551115e-05,
+      "loss": 0.1611,
+      "step": 36300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9805076426868602e-05,
+      "loss": 0.1547,
+      "step": 36350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9791053148226055e-05,
+      "loss": 0.1413,
+      "step": 36400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9777029869583512e-05,
+      "loss": 0.1641,
+      "step": 36450
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9763006590940962e-05,
+      "loss": 0.1591,
+      "step": 36500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9748983312298415e-05,
+      "loss": 0.1604,
+      "step": 36550
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.973496003365587e-05,
+      "loss": 0.163,
+      "step": 36600
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9720936755013322e-05,
+      "loss": 0.1517,
+      "step": 36650
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9706913476370775e-05,
+      "loss": 0.1675,
+      "step": 36700
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.969289019772823e-05,
+      "loss": 0.1729,
+      "step": 36750
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9678866919085682e-05,
+      "loss": 0.1564,
+      "step": 36800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.966484364044314e-05,
+      "loss": 0.1807,
+      "step": 36850
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.965082036180059e-05,
+      "loss": 0.161,
+      "step": 36900
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9636797083158042e-05,
+      "loss": 0.1514,
+      "step": 36950
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9622773804515496e-05,
+      "loss": 0.1548,
+      "step": 37000
+    },
+    {
+      "epoch": 1.04,
+      "eval_bleu": 95.9217,
+      "eval_gen_len": 64.223,
+      "eval_loss": 0.25345203280448914,
+      "eval_rouge1": 93.7839,
+      "eval_rouge2": 90.1221,
+      "eval_rougeL": 93.5635,
+      "eval_rougeLsum": 93.554,
+      "eval_runtime": 3095.5847,
+      "eval_samples_per_second": 0.969,
+      "eval_steps_per_second": 0.242,
+      "step": 37000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.960875052587295e-05,
+      "loss": 0.1647,
+      "step": 37050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9594727247230402e-05,
+      "loss": 0.1629,
+      "step": 37100
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9580703968587856e-05,
+      "loss": 0.1805,
+      "step": 37150
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.956668068994531e-05,
+      "loss": 0.1727,
+      "step": 37200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9552657411302766e-05,
+      "loss": 0.1967,
+      "step": 37250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9538634132660216e-05,
+      "loss": 0.1551,
+      "step": 37300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.952461085401767e-05,
+      "loss": 0.1554,
+      "step": 37350
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9510587575375122e-05,
+      "loss": 0.1491,
+      "step": 37400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9496564296732576e-05,
+      "loss": 0.1572,
+      "step": 37450
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9482541018090033e-05,
+      "loss": 0.1706,
+      "step": 37500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9468517739447483e-05,
+      "loss": 0.1632,
+      "step": 37550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9454494460804936e-05,
+      "loss": 0.1609,
+      "step": 37600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9440471182162393e-05,
+      "loss": 0.1672,
+      "step": 37650
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9426447903519843e-05,
+      "loss": 0.1535,
+      "step": 37700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9412424624877296e-05,
+      "loss": 0.1545,
+      "step": 37750
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.939840134623475e-05,
+      "loss": 0.1606,
+      "step": 37800
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9384378067592203e-05,
+      "loss": 0.1475,
+      "step": 37850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.937035478894966e-05,
+      "loss": 0.1663,
+      "step": 37900
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.935633151030711e-05,
+      "loss": 0.1698,
+      "step": 37950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9342308231664563e-05,
+      "loss": 0.1565,
+      "step": 38000
+    },
+    {
+      "epoch": 1.07,
+      "eval_bleu": 95.9359,
+      "eval_gen_len": 64.2767,
+      "eval_loss": 0.25277698040008545,
+      "eval_rouge1": 93.7483,
+      "eval_rouge2": 90.0899,
+      "eval_rougeL": 93.5433,
+      "eval_rougeLsum": 93.5171,
+      "eval_runtime": 3125.498,
+      "eval_samples_per_second": 0.96,
+      "eval_steps_per_second": 0.24,
+      "step": 38000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.932828495302202e-05,
+      "loss": 0.1686,
+      "step": 38050
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.931426167437947e-05,
+      "loss": 0.1466,
+      "step": 38100
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9300238395736923e-05,
+      "loss": 0.1817,
+      "step": 38150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9286215117094376e-05,
+      "loss": 0.1765,
+      "step": 38200
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.927219183845183e-05,
+      "loss": 0.1642,
+      "step": 38250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9258168559809286e-05,
+      "loss": 0.1732,
+      "step": 38300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9244145281166736e-05,
+      "loss": 0.1462,
+      "step": 38350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.923012200252419e-05,
+      "loss": 0.167,
+      "step": 38400
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9216098723881646e-05,
+      "loss": 0.1484,
+      "step": 38450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9202075445239096e-05,
+      "loss": 0.1661,
+      "step": 38500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.918805216659655e-05,
+      "loss": 0.1457,
+      "step": 38550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9174028887954003e-05,
+      "loss": 0.1445,
+      "step": 38600
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9160005609311456e-05,
+      "loss": 0.1499,
+      "step": 38650
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9145982330668913e-05,
+      "loss": 0.1495,
+      "step": 38700
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9131959052026363e-05,
+      "loss": 0.1752,
+      "step": 38750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9117935773383816e-05,
+      "loss": 0.174,
+      "step": 38800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9103912494741273e-05,
+      "loss": 0.1513,
+      "step": 38850
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.9089889216098723e-05,
+      "loss": 0.1727,
+      "step": 38900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.907586593745618e-05,
+      "loss": 0.1738,
+      "step": 38950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.906184265881363e-05,
+      "loss": 0.1695,
+      "step": 39000
+    },
+    {
+      "epoch": 1.09,
+      "eval_bleu": 95.9169,
+      "eval_gen_len": 64.224,
+      "eval_loss": 0.2530539035797119,
+      "eval_rouge1": 93.7543,
+      "eval_rouge2": 90.0728,
+      "eval_rougeL": 93.5572,
+      "eval_rougeLsum": 93.5471,
+      "eval_runtime": 3118.1985,
+      "eval_samples_per_second": 0.962,
+      "eval_steps_per_second": 0.241,
+      "step": 39000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9047819380171083e-05,
+      "loss": 0.1575,
+      "step": 39050
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.903379610152854e-05,
+      "loss": 0.1706,
+      "step": 39100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.901977282288599e-05,
+      "loss": 0.1483,
+      "step": 39150
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9005749544243443e-05,
+      "loss": 0.1535,
+      "step": 39200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.89917262656009e-05,
+      "loss": 0.1454,
+      "step": 39250
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.897770298695835e-05,
+      "loss": 0.1614,
+      "step": 39300
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8963679708315807e-05,
+      "loss": 0.1584,
+      "step": 39350
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8949656429673257e-05,
+      "loss": 0.1733,
+      "step": 39400
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.893563315103071e-05,
+      "loss": 0.1391,
+      "step": 39450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8921609872388167e-05,
+      "loss": 0.1517,
+      "step": 39500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8907586593745617e-05,
+      "loss": 0.152,
+      "step": 39550
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.889356331510307e-05,
+      "loss": 0.1458,
+      "step": 39600
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8879540036460527e-05,
+      "loss": 0.1675,
+      "step": 39650
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8865516757817977e-05,
+      "loss": 0.1554,
+      "step": 39700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8851493479175434e-05,
+      "loss": 0.169,
+      "step": 39750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8837470200532884e-05,
+      "loss": 0.1841,
+      "step": 39800
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8823446921890337e-05,
+      "loss": 0.1717,
+      "step": 39850
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8809423643247794e-05,
+      "loss": 0.1386,
+      "step": 39900
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8795400364605244e-05,
+      "loss": 0.1642,
+      "step": 39950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.87813770859627e-05,
+      "loss": 0.1804,
+      "step": 40000
+    },
+    {
+      "epoch": 1.12,
+      "eval_bleu": 95.9129,
+      "eval_gen_len": 64.2003,
+      "eval_loss": 0.2509741187095642,
+      "eval_rouge1": 93.8069,
+      "eval_rouge2": 90.1198,
+      "eval_rougeL": 93.5952,
+      "eval_rougeLsum": 93.5824,
+      "eval_runtime": 3104.9942,
+      "eval_samples_per_second": 0.966,
+      "eval_steps_per_second": 0.242,
+      "step": 40000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8767353807320154e-05,
+      "loss": 0.1632,
+      "step": 40050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8753330528677604e-05,
+      "loss": 0.1432,
+      "step": 40100
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.873930725003506e-05,
+      "loss": 0.1485,
+      "step": 40150
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.872528397139251e-05,
+      "loss": 0.1397,
+      "step": 40200
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8711260692749964e-05,
+      "loss": 0.1595,
+      "step": 40250
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.869723741410742e-05,
+      "loss": 0.1539,
+      "step": 40300
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.868321413546487e-05,
+      "loss": 0.1635,
+      "step": 40350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8669190856822327e-05,
+      "loss": 0.1709,
+      "step": 40400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.865516757817978e-05,
+      "loss": 0.1554,
+      "step": 40450
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.864114429953723e-05,
+      "loss": 0.1597,
+      "step": 40500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8627121020894687e-05,
+      "loss": 0.1652,
+      "step": 40550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8613097742252137e-05,
+      "loss": 0.1794,
+      "step": 40600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.859907446360959e-05,
+      "loss": 0.159,
+      "step": 40650
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8585051184967047e-05,
+      "loss": 0.1611,
+      "step": 40700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8571027906324497e-05,
+      "loss": 0.1545,
+      "step": 40750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8557004627681954e-05,
+      "loss": 0.1458,
+      "step": 40800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8542981349039408e-05,
+      "loss": 0.1745,
+      "step": 40850
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8528958070396858e-05,
+      "loss": 0.1612,
+      "step": 40900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8514934791754314e-05,
+      "loss": 0.1589,
+      "step": 40950
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8500911513111764e-05,
+      "loss": 0.1421,
+      "step": 41000
+    },
+    {
+      "epoch": 1.15,
+      "eval_bleu": 95.9507,
+      "eval_gen_len": 64.2133,
+      "eval_loss": 0.2519792318344116,
+      "eval_rouge1": 93.805,
+      "eval_rouge2": 90.1779,
+      "eval_rougeL": 93.6222,
+      "eval_rougeLsum": 93.6155,
+      "eval_runtime": 3133.4577,
+      "eval_samples_per_second": 0.957,
+      "eval_steps_per_second": 0.239,
+      "step": 41000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8486888234469218e-05,
+      "loss": 0.1424,
+      "step": 41050
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8472864955826674e-05,
+      "loss": 0.1578,
+      "step": 41100
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.8458841677184124e-05,
+      "loss": 0.1553,
+      "step": 41150
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.844481839854158e-05,
+      "loss": 0.1969,
+      "step": 41200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8430795119899034e-05,
+      "loss": 0.1674,
+      "step": 41250
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8416771841256484e-05,
+      "loss": 0.1763,
+      "step": 41300
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.840274856261394e-05,
+      "loss": 0.1751,
+      "step": 41350
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.838872528397139e-05,
+      "loss": 0.1588,
+      "step": 41400
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.8374702005328848e-05,
+      "loss": 0.173,
+      "step": 41450
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.83606787266863e-05,
+      "loss": 0.1759,
+      "step": 41500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.834665544804375e-05,
+      "loss": 0.1614,
+      "step": 41550
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8332632169401208e-05,
+      "loss": 0.1463,
+      "step": 41600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.831860889075866e-05,
+      "loss": 0.1825,
+      "step": 41650
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.830458561211611e-05,
+      "loss": 0.1483,
+      "step": 41700
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8290562333473568e-05,
+      "loss": 0.1651,
+      "step": 41750
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8276539054831018e-05,
+      "loss": 0.1602,
+      "step": 41800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.8262515776188475e-05,
+      "loss": 0.1462,
+      "step": 41850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8248492497545928e-05,
+      "loss": 0.1686,
+      "step": 41900
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8234469218903378e-05,
+      "loss": 0.1474,
+      "step": 41950
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8220445940260835e-05,
+      "loss": 0.1594,
+      "step": 42000
+    },
+    {
+      "epoch": 1.18,
+      "eval_bleu": 95.9473,
+      "eval_gen_len": 64.235,
+      "eval_loss": 0.2515351176261902,
+      "eval_rouge1": 93.7671,
+      "eval_rouge2": 90.1013,
+      "eval_rougeL": 93.575,
+      "eval_rougeLsum": 93.5369,
+      "eval_runtime": 3108.2504,
+      "eval_samples_per_second": 0.965,
+      "eval_steps_per_second": 0.241,
+      "step": 42000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8206422661618288e-05,
+      "loss": 0.1723,
+      "step": 42050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8192399382975738e-05,
+      "loss": 0.1553,
+      "step": 42100
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8178376104333195e-05,
+      "loss": 0.1709,
+      "step": 42150
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.8164352825690645e-05,
+      "loss": 0.1861,
+      "step": 42200
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.81503295470481e-05,
+      "loss": 0.1631,
+      "step": 42250
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8136306268405555e-05,
+      "loss": 0.1657,
+      "step": 42300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8122282989763005e-05,
+      "loss": 0.1516,
+      "step": 42350
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.810825971112046e-05,
+      "loss": 0.1532,
+      "step": 42400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8094236432477915e-05,
+      "loss": 0.1428,
+      "step": 42450
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.8080213153835365e-05,
+      "loss": 0.1493,
+      "step": 42500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.806618987519282e-05,
+      "loss": 0.1378,
+      "step": 42550
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.805216659655027e-05,
+      "loss": 0.144,
+      "step": 42600
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.803814331790773e-05,
+      "loss": 0.1685,
+      "step": 42650
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8024120039265182e-05,
+      "loss": 0.1612,
+      "step": 42700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.8010096760622632e-05,
+      "loss": 0.1678,
+      "step": 42750
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.799607348198009e-05,
+      "loss": 0.1843,
+      "step": 42800
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7982050203337542e-05,
+      "loss": 0.1592,
+      "step": 42850
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7968026924694995e-05,
+      "loss": 0.1376,
+      "step": 42900
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.795400364605245e-05,
+      "loss": 0.1777,
+      "step": 42950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.79399803674099e-05,
+      "loss": 0.1343,
+      "step": 43000
+    },
+    {
+      "epoch": 1.21,
+      "eval_bleu": 95.9595,
+      "eval_gen_len": 64.2177,
+      "eval_loss": 0.25027555227279663,
+      "eval_rouge1": 93.7701,
+      "eval_rouge2": 90.1732,
+      "eval_rougeL": 93.607,
+      "eval_rougeLsum": 93.6046,
+      "eval_runtime": 3107.9422,
+      "eval_samples_per_second": 0.965,
+      "eval_steps_per_second": 0.241,
+      "step": 43000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7925957088767355e-05,
+      "loss": 0.1539,
+      "step": 43050
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.791193381012481e-05,
+      "loss": 0.1527,
+      "step": 43100
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.789791053148226e-05,
+      "loss": 0.1821,
+      "step": 43150
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7883887252839715e-05,
+      "loss": 0.1645,
+      "step": 43200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.786986397419717e-05,
+      "loss": 0.1596,
+      "step": 43250
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7855840695554622e-05,
+      "loss": 0.1394,
+      "step": 43300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7841817416912075e-05,
+      "loss": 0.1526,
+      "step": 43350
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7827794138269525e-05,
+      "loss": 0.1529,
+      "step": 43400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7813770859626982e-05,
+      "loss": 0.1662,
+      "step": 43450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7799747580984435e-05,
+      "loss": 0.1572,
+      "step": 43500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7785724302341885e-05,
+      "loss": 0.168,
+      "step": 43550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7771701023699342e-05,
+      "loss": 0.1442,
+      "step": 43600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7757677745056796e-05,
+      "loss": 0.15,
+      "step": 43650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.774365446641425e-05,
+      "loss": 0.1666,
+      "step": 43700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7729631187771702e-05,
+      "loss": 0.1609,
+      "step": 43750
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7715607909129152e-05,
+      "loss": 0.1732,
+      "step": 43800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.770158463048661e-05,
+      "loss": 0.163,
+      "step": 43850
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7687561351844062e-05,
+      "loss": 0.1482,
+      "step": 43900
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7673538073201512e-05,
+      "loss": 0.1489,
+      "step": 43950
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765951479455897e-05,
+      "loss": 0.1435,
+      "step": 44000
+    },
+    {
+      "epoch": 1.23,
+      "eval_bleu": 95.9319,
+      "eval_gen_len": 64.194,
+      "eval_loss": 0.2497410923242569,
+      "eval_rouge1": 93.7912,
+      "eval_rouge2": 90.1359,
+      "eval_rougeL": 93.5982,
+      "eval_rougeLsum": 93.5886,
+      "eval_runtime": 3092.2489,
+      "eval_samples_per_second": 0.97,
+      "eval_steps_per_second": 0.243,
+      "step": 44000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7645491515916422e-05,
+      "loss": 0.1686,
+      "step": 44050
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7631468237273876e-05,
+      "loss": 0.1699,
+      "step": 44100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.761744495863133e-05,
+      "loss": 0.1564,
+      "step": 44150
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760342167998878e-05,
+      "loss": 0.1613,
+      "step": 44200
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7589398401346236e-05,
+      "loss": 0.1526,
+      "step": 44250
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.757537512270369e-05,
+      "loss": 0.1549,
+      "step": 44300
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7561351844061143e-05,
+      "loss": 0.181,
+      "step": 44350
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7547328565418596e-05,
+      "loss": 0.1665,
+      "step": 44400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.753330528677605e-05,
+      "loss": 0.1805,
+      "step": 44450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7519282008133503e-05,
+      "loss": 0.1414,
+      "step": 44500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7505258729490956e-05,
+      "loss": 0.1581,
+      "step": 44550
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7491235450848406e-05,
+      "loss": 0.15,
+      "step": 44600
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7477212172205863e-05,
+      "loss": 0.1401,
+      "step": 44650
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7463188893563316e-05,
+      "loss": 0.144,
+      "step": 44700
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.744916561492077e-05,
+      "loss": 0.1459,
+      "step": 44750
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7435142336278223e-05,
+      "loss": 0.164,
+      "step": 44800
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7421119057635676e-05,
+      "loss": 0.1428,
+      "step": 44850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.740709577899313e-05,
+      "loss": 0.1529,
+      "step": 44900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7393072500350583e-05,
+      "loss": 0.1631,
+      "step": 44950
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7379049221708033e-05,
+      "loss": 0.1613,
+      "step": 45000
+    },
+    {
+      "epoch": 1.26,
+      "eval_bleu": 95.9702,
+      "eval_gen_len": 64.2347,
+      "eval_loss": 0.24982018768787384,
+      "eval_rouge1": 93.7924,
+      "eval_rouge2": 90.1761,
+      "eval_rougeL": 93.6063,
+      "eval_rougeLsum": 93.5896,
+      "eval_runtime": 3101.0672,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 45000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.736502594306549e-05,
+      "loss": 0.165,
+      "step": 45050
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7351002664422943e-05,
+      "loss": 0.1467,
+      "step": 45100
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7336979385780396e-05,
+      "loss": 0.1935,
+      "step": 45150
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.732295610713785e-05,
+      "loss": 0.1451,
+      "step": 45200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7308932828495303e-05,
+      "loss": 0.1428,
+      "step": 45250
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7294909549852756e-05,
+      "loss": 0.1494,
+      "step": 45300
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.728088627121021e-05,
+      "loss": 0.1626,
+      "step": 45350
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.726686299256766e-05,
+      "loss": 0.1591,
+      "step": 45400
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7252839713925116e-05,
+      "loss": 0.1611,
+      "step": 45450
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.723881643528257e-05,
+      "loss": 0.1549,
+      "step": 45500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7224793156640023e-05,
+      "loss": 0.1818,
+      "step": 45550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7210769877997477e-05,
+      "loss": 0.1738,
+      "step": 45600
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.719674659935493e-05,
+      "loss": 0.1516,
+      "step": 45650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7182723320712383e-05,
+      "loss": 0.1508,
+      "step": 45700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7168700042069837e-05,
+      "loss": 0.1578,
+      "step": 45750
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.715467676342729e-05,
+      "loss": 0.1601,
+      "step": 45800
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7140653484784743e-05,
+      "loss": 0.1603,
+      "step": 45850
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7126630206142197e-05,
+      "loss": 0.1698,
+      "step": 45900
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.711260692749965e-05,
+      "loss": 0.1406,
+      "step": 45950
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7098583648857103e-05,
+      "loss": 0.1636,
+      "step": 46000
+    },
+    {
+      "epoch": 1.29,
+      "eval_bleu": 95.9319,
+      "eval_gen_len": 64.244,
+      "eval_loss": 0.25013232231140137,
+      "eval_rouge1": 93.7318,
+      "eval_rouge2": 90.1123,
+      "eval_rougeL": 93.5419,
+      "eval_rougeLsum": 93.5341,
+      "eval_runtime": 3080.401,
+      "eval_samples_per_second": 0.974,
+      "eval_steps_per_second": 0.243,
+      "step": 46000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7084560370214557e-05,
+      "loss": 0.1589,
+      "step": 46050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.707053709157201e-05,
+      "loss": 0.1574,
+      "step": 46100
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7056513812929463e-05,
+      "loss": 0.1478,
+      "step": 46150
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7042490534286917e-05,
+      "loss": 0.1528,
+      "step": 46200
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.702846725564437e-05,
+      "loss": 0.1469,
+      "step": 46250
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7014443977001824e-05,
+      "loss": 0.1489,
+      "step": 46300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7000420698359277e-05,
+      "loss": 0.1464,
+      "step": 46350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.698639741971673e-05,
+      "loss": 0.144,
+      "step": 46400
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6972374141074184e-05,
+      "loss": 0.1694,
+      "step": 46450
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.6958350862431637e-05,
+      "loss": 0.1569,
+      "step": 46500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.694432758378909e-05,
+      "loss": 0.1658,
+      "step": 46550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6930304305146544e-05,
+      "loss": 0.166,
+      "step": 46600
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6916281026503997e-05,
+      "loss": 0.1837,
+      "step": 46650
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.690225774786145e-05,
+      "loss": 0.1858,
+      "step": 46700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6888234469218904e-05,
+      "loss": 0.149,
+      "step": 46750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.6874211190576357e-05,
+      "loss": 0.1679,
+      "step": 46800
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.686018791193381e-05,
+      "loss": 0.1571,
+      "step": 46850
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.6846164633291264e-05,
+      "loss": 0.1439,
+      "step": 46900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.6832141354648717e-05,
+      "loss": 0.1661,
+      "step": 46950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.681811807600617e-05,
+      "loss": 0.1661,
+      "step": 47000
+    },
+    {
+      "epoch": 1.32,
+      "eval_bleu": 95.9777,
+      "eval_gen_len": 64.2717,
+      "eval_loss": 0.24935156106948853,
+      "eval_rouge1": 93.831,
+      "eval_rouge2": 90.2688,
+      "eval_rougeL": 93.6413,
+      "eval_rougeLsum": 93.6338,
+      "eval_runtime": 3079.0307,
+      "eval_samples_per_second": 0.974,
+      "eval_steps_per_second": 0.244,
+      "step": 47000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.6804094797363624e-05,
+      "loss": 0.1605,
+      "step": 47050
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.6790071518721077e-05,
+      "loss": 0.1488,
+      "step": 47100
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.677604824007853e-05,
+      "loss": 0.1515,
+      "step": 47150
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.6762024961435984e-05,
+      "loss": 0.1651,
+      "step": 47200
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.674800168279344e-05,
+      "loss": 0.166,
+      "step": 47250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.673397840415089e-05,
+      "loss": 0.1507,
+      "step": 47300
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.6719955125508344e-05,
+      "loss": 0.1538,
+      "step": 47350
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.6705931846865797e-05,
+      "loss": 0.1574,
+      "step": 47400
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.669190856822325e-05,
+      "loss": 0.1534,
+      "step": 47450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.6677885289580704e-05,
+      "loss": 0.1617,
+      "step": 47500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.6663862010938157e-05,
+      "loss": 0.1506,
+      "step": 47550
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.664983873229561e-05,
+      "loss": 0.1554,
+      "step": 47600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.6635815453653068e-05,
+      "loss": 0.1405,
+      "step": 47650
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.6621792175010518e-05,
+      "loss": 0.162,
+      "step": 47700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.660776889636797e-05,
+      "loss": 0.1456,
+      "step": 47750
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.6593745617725424e-05,
+      "loss": 0.1664,
+      "step": 47800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.6579722339082878e-05,
+      "loss": 0.1626,
+      "step": 47850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.656569906044033e-05,
+      "loss": 0.1568,
+      "step": 47900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.6551675781797784e-05,
+      "loss": 0.154,
+      "step": 47950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.6537652503155238e-05,
+      "loss": 0.1484,
+      "step": 48000
+    },
+    {
+      "epoch": 1.35,
+      "eval_bleu": 95.9573,
+      "eval_gen_len": 64.25,
+      "eval_loss": 0.24856652319431305,
+      "eval_rouge1": 93.8022,
+      "eval_rouge2": 90.2009,
+      "eval_rougeL": 93.6171,
+      "eval_rougeLsum": 93.5939,
+      "eval_runtime": 3131.2312,
+      "eval_samples_per_second": 0.958,
+      "eval_steps_per_second": 0.24,
+      "step": 48000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.6523629224512694e-05,
+      "loss": 0.1558,
+      "step": 48050
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.6509605945870144e-05,
+      "loss": 0.144,
+      "step": 48100
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.6495582667227598e-05,
+      "loss": 0.16,
+      "step": 48150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.648155938858505e-05,
+      "loss": 0.1812,
+      "step": 48200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.6467536109942504e-05,
+      "loss": 0.1516,
+      "step": 48250
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.6453512831299958e-05,
+      "loss": 0.1506,
+      "step": 48300
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.643948955265741e-05,
+      "loss": 0.1515,
+      "step": 48350
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6425466274014865e-05,
+      "loss": 0.1418,
+      "step": 48400
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.641144299537232e-05,
+      "loss": 0.162,
+      "step": 48450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.639741971672977e-05,
+      "loss": 0.1552,
+      "step": 48500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6383396438087225e-05,
+      "loss": 0.1613,
+      "step": 48550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.6369373159444678e-05,
+      "loss": 0.1621,
+      "step": 48600
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.635534988080213e-05,
+      "loss": 0.1594,
+      "step": 48650
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6341326602159588e-05,
+      "loss": 0.1447,
+      "step": 48700
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6327303323517038e-05,
+      "loss": 0.1508,
+      "step": 48750
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.631328004487449e-05,
+      "loss": 0.1663,
+      "step": 48800
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6299256766231948e-05,
+      "loss": 0.1546,
+      "step": 48850
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6285233487589398e-05,
+      "loss": 0.1435,
+      "step": 48900
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.627121020894685e-05,
+      "loss": 0.1645,
+      "step": 48950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.6257186930304305e-05,
+      "loss": 0.1532,
+      "step": 49000
+    },
+    {
+      "epoch": 1.37,
+      "eval_bleu": 95.9496,
+      "eval_gen_len": 64.2453,
+      "eval_loss": 0.24891996383666992,
+      "eval_rouge1": 93.7774,
+      "eval_rouge2": 90.1343,
+      "eval_rougeL": 93.5686,
+      "eval_rougeLsum": 93.5586,
+      "eval_runtime": 3102.2649,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 49000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6243163651661758e-05,
+      "loss": 0.1465,
+      "step": 49050
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6229140373019215e-05,
+      "loss": 0.1652,
+      "step": 49100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6215117094376665e-05,
+      "loss": 0.1538,
+      "step": 49150
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6201093815734118e-05,
+      "loss": 0.1608,
+      "step": 49200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6187070537091575e-05,
+      "loss": 0.1735,
+      "step": 49250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6173047258449025e-05,
+      "loss": 0.1752,
+      "step": 49300
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.615902397980648e-05,
+      "loss": 0.1812,
+      "step": 49350
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.614500070116393e-05,
+      "loss": 0.1568,
+      "step": 49400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6130977422521385e-05,
+      "loss": 0.1543,
+      "step": 49450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6116954143878842e-05,
+      "loss": 0.1625,
+      "step": 49500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6102930865236292e-05,
+      "loss": 0.1547,
+      "step": 49550
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6088907586593745e-05,
+      "loss": 0.1455,
+      "step": 49600
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6074884307951202e-05,
+      "loss": 0.1623,
+      "step": 49650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6060861029308652e-05,
+      "loss": 0.1532,
+      "step": 49700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6046837750666105e-05,
+      "loss": 0.1539,
+      "step": 49750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.603281447202356e-05,
+      "loss": 0.1541,
+      "step": 49800
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6018791193381012e-05,
+      "loss": 0.1538,
+      "step": 49850
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.600476791473847e-05,
+      "loss": 0.1656,
+      "step": 49900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.599074463609592e-05,
+      "loss": 0.1598,
+      "step": 49950
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5976721357453372e-05,
+      "loss": 0.1523,
+      "step": 50000
+    },
+    {
+      "epoch": 1.4,
+      "eval_bleu": 95.9721,
+      "eval_gen_len": 64.2117,
+      "eval_loss": 0.2481638789176941,
+      "eval_rouge1": 93.8515,
+      "eval_rouge2": 90.2059,
+      "eval_rougeL": 93.6666,
+      "eval_rougeLsum": 93.6523,
+      "eval_runtime": 3096.96,
+      "eval_samples_per_second": 0.969,
+      "eval_steps_per_second": 0.242,
+      "step": 50000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.596269807881083e-05,
+      "loss": 0.1373,
+      "step": 50050
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.594867480016828e-05,
+      "loss": 0.1535,
+      "step": 50100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5934651521525735e-05,
+      "loss": 0.1483,
+      "step": 50150
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5920628242883185e-05,
+      "loss": 0.1558,
+      "step": 50200
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.590660496424064e-05,
+      "loss": 0.1669,
+      "step": 50250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5892581685598096e-05,
+      "loss": 0.1654,
+      "step": 50300
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5878558406955545e-05,
+      "loss": 0.178,
+      "step": 50350
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5864535128313e-05,
+      "loss": 0.1556,
+      "step": 50400
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5850511849670456e-05,
+      "loss": 0.1705,
+      "step": 50450
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5836488571027906e-05,
+      "loss": 0.1585,
+      "step": 50500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5822465292385362e-05,
+      "loss": 0.1307,
+      "step": 50550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5808442013742812e-05,
+      "loss": 0.1471,
+      "step": 50600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5794418735100266e-05,
+      "loss": 0.1732,
+      "step": 50650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5780395456457722e-05,
+      "loss": 0.1465,
+      "step": 50700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5766372177815172e-05,
+      "loss": 0.1387,
+      "step": 50750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.5752348899172626e-05,
+      "loss": 0.1711,
+      "step": 50800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.5738325620530082e-05,
+      "loss": 0.1361,
+      "step": 50850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.5724302341887532e-05,
+      "loss": 0.1503,
+      "step": 50900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.571027906324499e-05,
+      "loss": 0.1615,
+      "step": 50950
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.569625578460244e-05,
+      "loss": 0.1597,
+      "step": 51000
+    },
+    {
+      "epoch": 1.43,
+      "eval_bleu": 96.0124,
+      "eval_gen_len": 64.2523,
+      "eval_loss": 0.24883191287517548,
+      "eval_rouge1": 93.8575,
+      "eval_rouge2": 90.2696,
+      "eval_rougeL": 93.6678,
+      "eval_rougeLsum": 93.6672,
+      "eval_runtime": 3113.1853,
+      "eval_samples_per_second": 0.964,
+      "eval_steps_per_second": 0.241,
+      "step": 51000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.5682232505959892e-05,
+      "loss": 0.1595,
+      "step": 51050
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.566820922731735e-05,
+      "loss": 0.1589,
+      "step": 51100
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.56541859486748e-05,
+      "loss": 0.1436,
+      "step": 51150
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5640162670032253e-05,
+      "loss": 0.1417,
+      "step": 51200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.562613939138971e-05,
+      "loss": 0.1727,
+      "step": 51250
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.561211611274716e-05,
+      "loss": 0.1563,
+      "step": 51300
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5598092834104616e-05,
+      "loss": 0.1521,
+      "step": 51350
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5584069555462066e-05,
+      "loss": 0.1583,
+      "step": 51400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.557004627681952e-05,
+      "loss": 0.1627,
+      "step": 51450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.5556022998176976e-05,
+      "loss": 0.142,
+      "step": 51500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5541999719534426e-05,
+      "loss": 0.165,
+      "step": 51550
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5527976440891883e-05,
+      "loss": 0.1666,
+      "step": 51600
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5513953162249336e-05,
+      "loss": 0.1522,
+      "step": 51650
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5499929883606786e-05,
+      "loss": 0.1576,
+      "step": 51700
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5485906604964243e-05,
+      "loss": 0.1435,
+      "step": 51750
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5471883326321693e-05,
+      "loss": 0.1809,
+      "step": 51800
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.5457860047679146e-05,
+      "loss": 0.1621,
+      "step": 51850
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5443836769036603e-05,
+      "loss": 0.1644,
+      "step": 51900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5429813490394053e-05,
+      "loss": 0.1629,
+      "step": 51950
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.541579021175151e-05,
+      "loss": 0.1564,
+      "step": 52000
+    },
+    {
+      "epoch": 1.46,
+      "eval_bleu": 96.0177,
+      "eval_gen_len": 64.2583,
+      "eval_loss": 0.24838809669017792,
+      "eval_rouge1": 93.8642,
+      "eval_rouge2": 90.2554,
+      "eval_rougeL": 93.6654,
+      "eval_rougeLsum": 93.6673,
+      "eval_runtime": 3134.0792,
+      "eval_samples_per_second": 0.957,
+      "eval_steps_per_second": 0.239,
+      "step": 52000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5401766933108963e-05,
+      "loss": 0.1697,
+      "step": 52050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.5387743654466413e-05,
+      "loss": 0.1629,
+      "step": 52100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.537372037582387e-05,
+      "loss": 0.1769,
+      "step": 52150
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.535969709718132e-05,
+      "loss": 0.1569,
+      "step": 52200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5345673818538773e-05,
+      "loss": 0.158,
+      "step": 52250
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.533165053989623e-05,
+      "loss": 0.1486,
+      "step": 52300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.531762726125368e-05,
+      "loss": 0.1517,
+      "step": 52350
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5303603982611137e-05,
+      "loss": 0.1496,
+      "step": 52400
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.528958070396859e-05,
+      "loss": 0.1511,
+      "step": 52450
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.527555742532604e-05,
+      "loss": 0.1435,
+      "step": 52500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.5261534146683497e-05,
+      "loss": 0.1342,
+      "step": 52550
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5247510868040947e-05,
+      "loss": 0.1369,
+      "step": 52600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5233487589398402e-05,
+      "loss": 0.1393,
+      "step": 52650
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5219464310755857e-05,
+      "loss": 0.1537,
+      "step": 52700
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5205441032113307e-05,
+      "loss": 0.1514,
+      "step": 52750
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5191417753470762e-05,
+      "loss": 0.1517,
+      "step": 52800
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5177394474828217e-05,
+      "loss": 0.1507,
+      "step": 52850
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.5163371196185668e-05,
+      "loss": 0.1845,
+      "step": 52900
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5149347917543123e-05,
+      "loss": 0.1449,
+      "step": 52950
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5135324638900573e-05,
+      "loss": 0.1488,
+      "step": 53000
+    },
+    {
+      "epoch": 1.49,
+      "eval_bleu": 96.0359,
+      "eval_gen_len": 64.297,
+      "eval_loss": 0.2471029907464981,
+      "eval_rouge1": 93.848,
+      "eval_rouge2": 90.2745,
+      "eval_rougeL": 93.6537,
+      "eval_rougeLsum": 93.6498,
+      "eval_runtime": 3110.5388,
+      "eval_samples_per_second": 0.964,
+      "eval_steps_per_second": 0.241,
+      "step": 53000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5121301360258028e-05,
+      "loss": 0.1479,
+      "step": 53050
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5107278081615484e-05,
+      "loss": 0.1713,
+      "step": 53100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5093254802972935e-05,
+      "loss": 0.1583,
+      "step": 53150
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5079231524330389e-05,
+      "loss": 0.159,
+      "step": 53200
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5065208245687844e-05,
+      "loss": 0.1429,
+      "step": 53250
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.5051184967045295e-05,
+      "loss": 0.1513,
+      "step": 53300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.503716168840275e-05,
+      "loss": 0.1605,
+      "step": 53350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.50231384097602e-05,
+      "loss": 0.1537,
+      "step": 53400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.5009115131117655e-05,
+      "loss": 0.1573,
+      "step": 53450
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4995091852475109e-05,
+      "loss": 0.1678,
+      "step": 53500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4981068573832564e-05,
+      "loss": 0.1602,
+      "step": 53550
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4967045295190015e-05,
+      "loss": 0.1313,
+      "step": 53600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4953022016547469e-05,
+      "loss": 0.1377,
+      "step": 53650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4938998737904922e-05,
+      "loss": 0.1509,
+      "step": 53700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4924975459262377e-05,
+      "loss": 0.1717,
+      "step": 53750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4910952180619829e-05,
+      "loss": 0.1518,
+      "step": 53800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4896928901977282e-05,
+      "loss": 0.1576,
+      "step": 53850
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4882905623334736e-05,
+      "loss": 0.1741,
+      "step": 53900
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.486888234469219e-05,
+      "loss": 0.1457,
+      "step": 53950
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4854859066049642e-05,
+      "loss": 0.1285,
+      "step": 54000
+    },
+    {
+      "epoch": 1.51,
+      "eval_bleu": 96.0361,
+      "eval_gen_len": 64.2367,
+      "eval_loss": 0.2483721524477005,
+      "eval_rouge1": 93.8768,
+      "eval_rouge2": 90.3035,
+      "eval_rougeL": 93.698,
+      "eval_rougeLsum": 93.6976,
+      "eval_runtime": 3128.3254,
+      "eval_samples_per_second": 0.959,
+      "eval_steps_per_second": 0.24,
+      "step": 54000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4840835787407096e-05,
+      "loss": 0.1327,
+      "step": 54050
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4826812508764549e-05,
+      "loss": 0.1455,
+      "step": 54100
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4812789230122004e-05,
+      "loss": 0.1336,
+      "step": 54150
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4798765951479456e-05,
+      "loss": 0.1497,
+      "step": 54200
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4784742672836909e-05,
+      "loss": 0.1545,
+      "step": 54250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4770719394194362e-05,
+      "loss": 0.1488,
+      "step": 54300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4756696115551817e-05,
+      "loss": 0.1591,
+      "step": 54350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4742672836909271e-05,
+      "loss": 0.1515,
+      "step": 54400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4728649558266723e-05,
+      "loss": 0.1538,
+      "step": 54450
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4714626279624176e-05,
+      "loss": 0.1652,
+      "step": 54500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4700603000981631e-05,
+      "loss": 0.1561,
+      "step": 54550
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4686579722339084e-05,
+      "loss": 0.1644,
+      "step": 54600
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4672556443696536e-05,
+      "loss": 0.1596,
+      "step": 54650
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.465853316505399e-05,
+      "loss": 0.1585,
+      "step": 54700
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4644509886411444e-05,
+      "loss": 0.1557,
+      "step": 54750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4630486607768898e-05,
+      "loss": 0.137,
+      "step": 54800
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.461646332912635e-05,
+      "loss": 0.1783,
+      "step": 54850
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4602440050483803e-05,
+      "loss": 0.1731,
+      "step": 54900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4588416771841258e-05,
+      "loss": 0.1552,
+      "step": 54950
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4574393493198711e-05,
+      "loss": 0.1628,
+      "step": 55000
+    },
+    {
+      "epoch": 1.54,
+      "eval_bleu": 96.033,
+      "eval_gen_len": 64.2217,
+      "eval_loss": 0.2469196319580078,
+      "eval_rouge1": 93.9233,
+      "eval_rouge2": 90.3508,
+      "eval_rougeL": 93.7332,
+      "eval_rougeLsum": 93.7184,
+      "eval_runtime": 3083.8837,
+      "eval_samples_per_second": 0.973,
+      "eval_steps_per_second": 0.243,
+      "step": 55000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4560370214556163e-05,
+      "loss": 0.1606,
+      "step": 55050
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4546346935913616e-05,
+      "loss": 0.1534,
+      "step": 55100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4532323657271071e-05,
+      "loss": 0.1516,
+      "step": 55150
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4518300378628525e-05,
+      "loss": 0.1674,
+      "step": 55200
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4504277099985976e-05,
+      "loss": 0.1562,
+      "step": 55250
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.449025382134343e-05,
+      "loss": 0.1509,
+      "step": 55300
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4476230542700885e-05,
+      "loss": 0.1476,
+      "step": 55350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4462207264058338e-05,
+      "loss": 0.1514,
+      "step": 55400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.444818398541579e-05,
+      "loss": 0.1497,
+      "step": 55450
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4434160706773243e-05,
+      "loss": 0.1728,
+      "step": 55500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4420137428130698e-05,
+      "loss": 0.1499,
+      "step": 55550
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4406114149488151e-05,
+      "loss": 0.1588,
+      "step": 55600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4392090870845603e-05,
+      "loss": 0.1589,
+      "step": 55650
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4378067592203056e-05,
+      "loss": 0.1332,
+      "step": 55700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4364044313560512e-05,
+      "loss": 0.15,
+      "step": 55750
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.4350021034917965e-05,
+      "loss": 0.1731,
+      "step": 55800
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4335997756275418e-05,
+      "loss": 0.1468,
+      "step": 55850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.432197447763287e-05,
+      "loss": 0.1464,
+      "step": 55900
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4307951198990325e-05,
+      "loss": 0.1597,
+      "step": 55950
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4293927920347778e-05,
+      "loss": 0.155,
+      "step": 56000
+    },
+    {
+      "epoch": 1.57,
+      "eval_bleu": 96.0287,
+      "eval_gen_len": 64.226,
+      "eval_loss": 0.24680346250534058,
+      "eval_rouge1": 93.8476,
+      "eval_rouge2": 90.2618,
+      "eval_rougeL": 93.6663,
+      "eval_rougeLsum": 93.6463,
+      "eval_runtime": 3086.3631,
+      "eval_samples_per_second": 0.972,
+      "eval_steps_per_second": 0.243,
+      "step": 56000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4279904641705232e-05,
+      "loss": 0.1392,
+      "step": 56050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4265881363062683e-05,
+      "loss": 0.1546,
+      "step": 56100
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.4251858084420138e-05,
+      "loss": 0.1596,
+      "step": 56150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4237834805777592e-05,
+      "loss": 0.1574,
+      "step": 56200
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4223811527135045e-05,
+      "loss": 0.1593,
+      "step": 56250
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4209788248492497e-05,
+      "loss": 0.1497,
+      "step": 56300
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4195764969849952e-05,
+      "loss": 0.1447,
+      "step": 56350
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4181741691207405e-05,
+      "loss": 0.1399,
+      "step": 56400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.4167718412564859e-05,
+      "loss": 0.1392,
+      "step": 56450
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.415369513392231e-05,
+      "loss": 0.1587,
+      "step": 56500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4139671855279765e-05,
+      "loss": 0.1763,
+      "step": 56550
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4125648576637219e-05,
+      "loss": 0.1542,
+      "step": 56600
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4111625297994672e-05,
+      "loss": 0.1422,
+      "step": 56650
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4097602019352124e-05,
+      "loss": 0.158,
+      "step": 56700
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4083578740709579e-05,
+      "loss": 0.1511,
+      "step": 56750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4069555462067032e-05,
+      "loss": 0.1572,
+      "step": 56800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.4055532183424485e-05,
+      "loss": 0.1677,
+      "step": 56850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4041508904781937e-05,
+      "loss": 0.1425,
+      "step": 56900
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4027485626139392e-05,
+      "loss": 0.1744,
+      "step": 56950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.4013462347496845e-05,
+      "loss": 0.1514,
+      "step": 57000
+    },
+    {
+      "epoch": 1.6,
+      "eval_bleu": 96.0359,
+      "eval_gen_len": 64.2267,
+      "eval_loss": 0.24612654745578766,
+      "eval_rouge1": 93.9162,
+      "eval_rouge2": 90.3455,
+      "eval_rougeL": 93.7454,
+      "eval_rougeLsum": 93.7264,
+      "eval_runtime": 3113.4814,
+      "eval_samples_per_second": 0.964,
+      "eval_steps_per_second": 0.241,
+      "step": 57000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3999439068854299e-05,
+      "loss": 0.149,
+      "step": 57050
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3985415790211752e-05,
+      "loss": 0.1457,
+      "step": 57100
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3971392511569206e-05,
+      "loss": 0.1426,
+      "step": 57150
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3957369232926659e-05,
+      "loss": 0.1574,
+      "step": 57200
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3943345954284112e-05,
+      "loss": 0.1566,
+      "step": 57250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3929322675641566e-05,
+      "loss": 0.1459,
+      "step": 57300
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3915299396999019e-05,
+      "loss": 0.1431,
+      "step": 57350
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3901276118356472e-05,
+      "loss": 0.154,
+      "step": 57400
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3887252839713926e-05,
+      "loss": 0.1513,
+      "step": 57450
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3873229561071379e-05,
+      "loss": 0.1571,
+      "step": 57500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3859206282428832e-05,
+      "loss": 0.1499,
+      "step": 57550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3845183003786286e-05,
+      "loss": 0.1563,
+      "step": 57600
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3831159725143739e-05,
+      "loss": 0.1719,
+      "step": 57650
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3817136446501192e-05,
+      "loss": 0.1507,
+      "step": 57700
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3803113167858646e-05,
+      "loss": 0.1387,
+      "step": 57750
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.37890898892161e-05,
+      "loss": 0.1439,
+      "step": 57800
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3775066610573553e-05,
+      "loss": 0.1434,
+      "step": 57850
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3761043331931006e-05,
+      "loss": 0.1719,
+      "step": 57900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.374702005328846e-05,
+      "loss": 0.1824,
+      "step": 57950
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3732996774645913e-05,
+      "loss": 0.1393,
+      "step": 58000
+    },
+    {
+      "epoch": 1.63,
+      "eval_bleu": 96.0528,
+      "eval_gen_len": 64.256,
+      "eval_loss": 0.24644368886947632,
+      "eval_rouge1": 93.9108,
+      "eval_rouge2": 90.3273,
+      "eval_rougeL": 93.7332,
+      "eval_rougeLsum": 93.7113,
+      "eval_runtime": 3121.7481,
+      "eval_samples_per_second": 0.961,
+      "eval_steps_per_second": 0.24,
+      "step": 58000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3718973496003366e-05,
+      "loss": 0.1601,
+      "step": 58050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.370495021736082e-05,
+      "loss": 0.1494,
+      "step": 58100
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3690926938718273e-05,
+      "loss": 0.1413,
+      "step": 58150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3676903660075726e-05,
+      "loss": 0.1609,
+      "step": 58200
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.366288038143318e-05,
+      "loss": 0.1683,
+      "step": 58250
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3648857102790633e-05,
+      "loss": 0.1556,
+      "step": 58300
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3634833824148086e-05,
+      "loss": 0.1479,
+      "step": 58350
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.362081054550554e-05,
+      "loss": 0.169,
+      "step": 58400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3606787266862993e-05,
+      "loss": 0.1521,
+      "step": 58450
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3592763988220446e-05,
+      "loss": 0.16,
+      "step": 58500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.35787407095779e-05,
+      "loss": 0.1453,
+      "step": 58550
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3564717430935353e-05,
+      "loss": 0.1439,
+      "step": 58600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3550694152292806e-05,
+      "loss": 0.1546,
+      "step": 58650
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.353667087365026e-05,
+      "loss": 0.157,
+      "step": 58700
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3522647595007715e-05,
+      "loss": 0.1769,
+      "step": 58750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3508624316365166e-05,
+      "loss": 0.1591,
+      "step": 58800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.349460103772262e-05,
+      "loss": 0.1492,
+      "step": 58850
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3480577759080073e-05,
+      "loss": 0.1474,
+      "step": 58900
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3466554480437528e-05,
+      "loss": 0.1622,
+      "step": 58950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.345253120179498e-05,
+      "loss": 0.1691,
+      "step": 59000
+    },
+    {
+      "epoch": 1.65,
+      "eval_bleu": 96.0335,
+      "eval_gen_len": 64.2307,
+      "eval_loss": 0.24554915726184845,
+      "eval_rouge1": 93.8976,
+      "eval_rouge2": 90.3009,
+      "eval_rougeL": 93.7085,
+      "eval_rougeLsum": 93.6965,
+      "eval_runtime": 3154.4266,
+      "eval_samples_per_second": 0.951,
+      "eval_steps_per_second": 0.238,
+      "step": 59000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3438507923152433e-05,
+      "loss": 0.157,
+      "step": 59050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3424484644509886e-05,
+      "loss": 0.1341,
+      "step": 59100
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3410461365867342e-05,
+      "loss": 0.1638,
+      "step": 59150
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3396438087224793e-05,
+      "loss": 0.1523,
+      "step": 59200
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3382414808582247e-05,
+      "loss": 0.1301,
+      "step": 59250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.33683915299397e-05,
+      "loss": 0.1456,
+      "step": 59300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3354368251297155e-05,
+      "loss": 0.1592,
+      "step": 59350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3340344972654607e-05,
+      "loss": 0.1464,
+      "step": 59400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.332632169401206e-05,
+      "loss": 0.1383,
+      "step": 59450
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3312298415369513e-05,
+      "loss": 0.1372,
+      "step": 59500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3298275136726968e-05,
+      "loss": 0.1515,
+      "step": 59550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.328425185808442e-05,
+      "loss": 0.1714,
+      "step": 59600
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3270228579441873e-05,
+      "loss": 0.1572,
+      "step": 59650
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3256205300799327e-05,
+      "loss": 0.1468,
+      "step": 59700
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3242182022156782e-05,
+      "loss": 0.16,
+      "step": 59750
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3228158743514233e-05,
+      "loss": 0.1402,
+      "step": 59800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3214135464871687e-05,
+      "loss": 0.1427,
+      "step": 59850
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.320011218622914e-05,
+      "loss": 0.1508,
+      "step": 59900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3186088907586595e-05,
+      "loss": 0.1613,
+      "step": 59950
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3172065628944049e-05,
+      "loss": 0.1623,
+      "step": 60000
+    },
+    {
+      "epoch": 1.68,
+      "eval_bleu": 96.0534,
+      "eval_gen_len": 64.217,
+      "eval_loss": 0.245611310005188,
+      "eval_rouge1": 93.9428,
+      "eval_rouge2": 90.3851,
+      "eval_rougeL": 93.7507,
+      "eval_rougeLsum": 93.731,
+      "eval_runtime": 3078.7009,
+      "eval_samples_per_second": 0.974,
+      "eval_steps_per_second": 0.244,
+      "step": 60000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.31580423503015e-05,
+      "loss": 0.1554,
+      "step": 60050
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3144019071658954e-05,
+      "loss": 0.1618,
+      "step": 60100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3129995793016409e-05,
+      "loss": 0.164,
+      "step": 60150
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3115972514373862e-05,
+      "loss": 0.1601,
+      "step": 60200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3101949235731314e-05,
+      "loss": 0.1497,
+      "step": 60250
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3087925957088767e-05,
+      "loss": 0.1414,
+      "step": 60300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3073902678446222e-05,
+      "loss": 0.1665,
+      "step": 60350
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3059879399803675e-05,
+      "loss": 0.1615,
+      "step": 60400
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3045856121161127e-05,
+      "loss": 0.1411,
+      "step": 60450
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.303183284251858e-05,
+      "loss": 0.1527,
+      "step": 60500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3017809563876036e-05,
+      "loss": 0.147,
+      "step": 60550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3003786285233489e-05,
+      "loss": 0.1346,
+      "step": 60600
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.298976300659094e-05,
+      "loss": 0.1512,
+      "step": 60650
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2975739727948394e-05,
+      "loss": 0.1363,
+      "step": 60700
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2961716449305849e-05,
+      "loss": 0.1592,
+      "step": 60750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2947693170663302e-05,
+      "loss": 0.1357,
+      "step": 60800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2933669892020754e-05,
+      "loss": 0.1617,
+      "step": 60850
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2919646613378207e-05,
+      "loss": 0.1425,
+      "step": 60900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2905623334735662e-05,
+      "loss": 0.1334,
+      "step": 60950
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2891600056093116e-05,
+      "loss": 0.1577,
+      "step": 61000
+    },
+    {
+      "epoch": 1.71,
+      "eval_bleu": 96.0539,
+      "eval_gen_len": 64.2317,
+      "eval_loss": 0.24573621153831482,
+      "eval_rouge1": 93.8833,
+      "eval_rouge2": 90.3011,
+      "eval_rougeL": 93.7003,
+      "eval_rougeLsum": 93.6749,
+      "eval_runtime": 3094.1804,
+      "eval_samples_per_second": 0.97,
+      "eval_steps_per_second": 0.242,
+      "step": 61000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2877576777450567e-05,
+      "loss": 0.1561,
+      "step": 61050
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.286355349880802e-05,
+      "loss": 0.1526,
+      "step": 61100
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2849530220165476e-05,
+      "loss": 0.1516,
+      "step": 61150
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.283550694152293e-05,
+      "loss": 0.1604,
+      "step": 61200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2821483662880381e-05,
+      "loss": 0.1393,
+      "step": 61250
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2807460384237834e-05,
+      "loss": 0.1522,
+      "step": 61300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.279343710559529e-05,
+      "loss": 0.1605,
+      "step": 61350
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2779413826952743e-05,
+      "loss": 0.1483,
+      "step": 61400
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2765390548310196e-05,
+      "loss": 0.1866,
+      "step": 61450
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2751367269667648e-05,
+      "loss": 0.1714,
+      "step": 61500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2737343991025103e-05,
+      "loss": 0.1616,
+      "step": 61550
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2723320712382556e-05,
+      "loss": 0.1377,
+      "step": 61600
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.270929743374001e-05,
+      "loss": 0.1322,
+      "step": 61650
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2695274155097461e-05,
+      "loss": 0.1655,
+      "step": 61700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2681250876454916e-05,
+      "loss": 0.1752,
+      "step": 61750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.266722759781237e-05,
+      "loss": 0.1452,
+      "step": 61800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2653204319169823e-05,
+      "loss": 0.1592,
+      "step": 61850
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2639181040527274e-05,
+      "loss": 0.1676,
+      "step": 61900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.262515776188473e-05,
+      "loss": 0.1578,
+      "step": 61950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2611134483242183e-05,
+      "loss": 0.1543,
+      "step": 62000
+    },
+    {
+      "epoch": 1.74,
+      "eval_bleu": 96.0497,
+      "eval_gen_len": 64.2223,
+      "eval_loss": 0.2444319725036621,
+      "eval_rouge1": 93.9499,
+      "eval_rouge2": 90.3622,
+      "eval_rougeL": 93.7659,
+      "eval_rougeLsum": 93.7507,
+      "eval_runtime": 3090.9521,
+      "eval_samples_per_second": 0.971,
+      "eval_steps_per_second": 0.243,
+      "step": 62000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2597111204599636e-05,
+      "loss": 0.1429,
+      "step": 62050
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2583087925957088e-05,
+      "loss": 0.1425,
+      "step": 62100
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2569064647314543e-05,
+      "loss": 0.1501,
+      "step": 62150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2555041368671996e-05,
+      "loss": 0.1417,
+      "step": 62200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.254101809002945e-05,
+      "loss": 0.1495,
+      "step": 62250
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2526994811386901e-05,
+      "loss": 0.1493,
+      "step": 62300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2512971532744356e-05,
+      "loss": 0.1437,
+      "step": 62350
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.249894825410181e-05,
+      "loss": 0.1497,
+      "step": 62400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2484924975459263e-05,
+      "loss": 0.1659,
+      "step": 62450
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2470901696816715e-05,
+      "loss": 0.1702,
+      "step": 62500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.245687841817417e-05,
+      "loss": 0.1591,
+      "step": 62550
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2442855139531623e-05,
+      "loss": 0.1472,
+      "step": 62600
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2428831860889077e-05,
+      "loss": 0.1638,
+      "step": 62650
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2414808582246528e-05,
+      "loss": 0.1669,
+      "step": 62700
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2400785303603983e-05,
+      "loss": 0.1435,
+      "step": 62750
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2386762024961437e-05,
+      "loss": 0.1513,
+      "step": 62800
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.237273874631889e-05,
+      "loss": 0.1455,
+      "step": 62850
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2358715467676343e-05,
+      "loss": 0.1578,
+      "step": 62900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2344692189033797e-05,
+      "loss": 0.1554,
+      "step": 62950
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.233066891039125e-05,
+      "loss": 0.1544,
+      "step": 63000
+    },
+    {
+      "epoch": 1.77,
+      "eval_bleu": 96.0681,
+      "eval_gen_len": 64.241,
+      "eval_loss": 0.245001420378685,
+      "eval_rouge1": 93.9489,
+      "eval_rouge2": 90.383,
+      "eval_rougeL": 93.7697,
+      "eval_rougeLsum": 93.7483,
+      "eval_runtime": 3098.5549,
+      "eval_samples_per_second": 0.968,
+      "eval_steps_per_second": 0.242,
+      "step": 63000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2316645631748703e-05,
+      "loss": 0.1269,
+      "step": 63050
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2302622353106157e-05,
+      "loss": 0.1689,
+      "step": 63100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.228859907446361e-05,
+      "loss": 0.1512,
+      "step": 63150
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2274575795821063e-05,
+      "loss": 0.1584,
+      "step": 63200
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2260552517178517e-05,
+      "loss": 0.142,
+      "step": 63250
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.224652923853597e-05,
+      "loss": 0.1424,
+      "step": 63300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2232505959893424e-05,
+      "loss": 0.144,
+      "step": 63350
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2218482681250877e-05,
+      "loss": 0.1608,
+      "step": 63400
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.220445940260833e-05,
+      "loss": 0.1376,
+      "step": 63450
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2190436123965784e-05,
+      "loss": 0.17,
+      "step": 63500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2176412845323237e-05,
+      "loss": 0.1587,
+      "step": 63550
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.216238956668069e-05,
+      "loss": 0.1572,
+      "step": 63600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2148366288038144e-05,
+      "loss": 0.1501,
+      "step": 63650
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2134343009395597e-05,
+      "loss": 0.1514,
+      "step": 63700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.212031973075305e-05,
+      "loss": 0.1656,
+      "step": 63750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2106296452110504e-05,
+      "loss": 0.1684,
+      "step": 63800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2092273173467957e-05,
+      "loss": 0.1481,
+      "step": 63850
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.207824989482541e-05,
+      "loss": 0.1528,
+      "step": 63900
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2064226616182864e-05,
+      "loss": 0.1433,
+      "step": 63950
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2050203337540317e-05,
+      "loss": 0.148,
+      "step": 64000
+    },
+    {
+      "epoch": 1.79,
+      "eval_bleu": 96.1034,
+      "eval_gen_len": 64.2707,
+      "eval_loss": 0.24430668354034424,
+      "eval_rouge1": 93.969,
+      "eval_rouge2": 90.4279,
+      "eval_rougeL": 93.7797,
+      "eval_rougeLsum": 93.7478,
+      "eval_runtime": 3170.7735,
+      "eval_samples_per_second": 0.946,
+      "eval_steps_per_second": 0.237,
+      "step": 64000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.203618005889777e-05,
+      "loss": 0.1517,
+      "step": 64050
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2022156780255224e-05,
+      "loss": 0.1383,
+      "step": 64100
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2008133501612677e-05,
+      "loss": 0.1335,
+      "step": 64150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.199411022297013e-05,
+      "loss": 0.1427,
+      "step": 64200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1980086944327584e-05,
+      "loss": 0.1617,
+      "step": 64250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1966063665685037e-05,
+      "loss": 0.1247,
+      "step": 64300
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.1952040387042492e-05,
+      "loss": 0.1574,
+      "step": 64350
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1938017108399944e-05,
+      "loss": 0.1431,
+      "step": 64400
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1923993829757397e-05,
+      "loss": 0.1423,
+      "step": 64450
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.190997055111485e-05,
+      "loss": 0.1508,
+      "step": 64500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1895947272472306e-05,
+      "loss": 0.1516,
+      "step": 64550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1881923993829758e-05,
+      "loss": 0.1392,
+      "step": 64600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1867900715187211e-05,
+      "loss": 0.1487,
+      "step": 64650
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1853877436544664e-05,
+      "loss": 0.1772,
+      "step": 64700
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.183985415790212e-05,
+      "loss": 0.1416,
+      "step": 64750
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1825830879259571e-05,
+      "loss": 0.1644,
+      "step": 64800
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1811807600617024e-05,
+      "loss": 0.1524,
+      "step": 64850
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1797784321974478e-05,
+      "loss": 0.1573,
+      "step": 64900
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1783761043331933e-05,
+      "loss": 0.1497,
+      "step": 64950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1769737764689384e-05,
+      "loss": 0.1676,
+      "step": 65000
+    },
+    {
+      "epoch": 1.82,
+      "eval_bleu": 96.0602,
+      "eval_gen_len": 64.2243,
+      "eval_loss": 0.24451076984405518,
+      "eval_rouge1": 93.9544,
+      "eval_rouge2": 90.3539,
+      "eval_rougeL": 93.7721,
+      "eval_rougeLsum": 93.7482,
+      "eval_runtime": 3116.5293,
+      "eval_samples_per_second": 0.963,
+      "eval_steps_per_second": 0.241,
+      "step": 65000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1755714486046838e-05,
+      "loss": 0.1516,
+      "step": 65050
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1741691207404291e-05,
+      "loss": 0.1279,
+      "step": 65100
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1727667928761746e-05,
+      "loss": 0.1641,
+      "step": 65150
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1713644650119198e-05,
+      "loss": 0.1234,
+      "step": 65200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1699621371476651e-05,
+      "loss": 0.1391,
+      "step": 65250
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1685598092834105e-05,
+      "loss": 0.1653,
+      "step": 65300
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.167157481419156e-05,
+      "loss": 0.1587,
+      "step": 65350
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1657551535549011e-05,
+      "loss": 0.1459,
+      "step": 65400
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1643528256906465e-05,
+      "loss": 0.1583,
+      "step": 65450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1629504978263918e-05,
+      "loss": 0.1477,
+      "step": 65500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1615481699621373e-05,
+      "loss": 0.1549,
+      "step": 65550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1601458420978825e-05,
+      "loss": 0.1531,
+      "step": 65600
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1587435142336278e-05,
+      "loss": 0.1594,
+      "step": 65650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1573411863693731e-05,
+      "loss": 0.1635,
+      "step": 65700
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1559388585051186e-05,
+      "loss": 0.1508,
+      "step": 65750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.154536530640864e-05,
+      "loss": 0.1314,
+      "step": 65800
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1531342027766091e-05,
+      "loss": 0.1398,
+      "step": 65850
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1517318749123545e-05,
+      "loss": 0.1427,
+      "step": 65900
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1503295470481e-05,
+      "loss": 0.1338,
+      "step": 65950
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1489272191838453e-05,
+      "loss": 0.1542,
+      "step": 66000
+    },
+    {
+      "epoch": 1.85,
+      "eval_bleu": 96.0707,
+      "eval_gen_len": 64.2427,
+      "eval_loss": 0.24484632909297943,
+      "eval_rouge1": 93.9242,
+      "eval_rouge2": 90.3577,
+      "eval_rougeL": 93.7378,
+      "eval_rougeLsum": 93.7199,
+      "eval_runtime": 3063.6568,
+      "eval_samples_per_second": 0.979,
+      "eval_steps_per_second": 0.245,
+      "step": 66000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1475248913195905e-05,
+      "loss": 0.1571,
+      "step": 66050
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1461225634553358e-05,
+      "loss": 0.1345,
+      "step": 66100
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1447202355910813e-05,
+      "loss": 0.1669,
+      "step": 66150
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1433179077268267e-05,
+      "loss": 0.1534,
+      "step": 66200
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1419155798625718e-05,
+      "loss": 0.1729,
+      "step": 66250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1405132519983172e-05,
+      "loss": 0.1338,
+      "step": 66300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1391109241340627e-05,
+      "loss": 0.1464,
+      "step": 66350
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.137708596269808e-05,
+      "loss": 0.1445,
+      "step": 66400
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1363062684055532e-05,
+      "loss": 0.1372,
+      "step": 66450
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1349039405412985e-05,
+      "loss": 0.1819,
+      "step": 66500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.133501612677044e-05,
+      "loss": 0.1483,
+      "step": 66550
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1320992848127894e-05,
+      "loss": 0.1391,
+      "step": 66600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1306969569485345e-05,
+      "loss": 0.1555,
+      "step": 66650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1292946290842799e-05,
+      "loss": 0.1748,
+      "step": 66700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1278923012200254e-05,
+      "loss": 0.1608,
+      "step": 66750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1264899733557707e-05,
+      "loss": 0.1399,
+      "step": 66800
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1250876454915159e-05,
+      "loss": 0.1399,
+      "step": 66850
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1236853176272612e-05,
+      "loss": 0.1753,
+      "step": 66900
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1222829897630067e-05,
+      "loss": 0.1494,
+      "step": 66950
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.120880661898752e-05,
+      "loss": 0.1507,
+      "step": 67000
+    },
+    {
+      "epoch": 1.88,
+      "eval_bleu": 96.0481,
+      "eval_gen_len": 64.2317,
+      "eval_loss": 0.24457640945911407,
+      "eval_rouge1": 93.9262,
+      "eval_rouge2": 90.3319,
+      "eval_rougeL": 93.7352,
+      "eval_rougeLsum": 93.7164,
+      "eval_runtime": 3094.0136,
+      "eval_samples_per_second": 0.97,
+      "eval_steps_per_second": 0.242,
+      "step": 67000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1194783340344972e-05,
+      "loss": 0.1775,
+      "step": 67050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1180760061702425e-05,
+      "loss": 0.1619,
+      "step": 67100
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.116673678305988e-05,
+      "loss": 0.1589,
+      "step": 67150
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1152713504417334e-05,
+      "loss": 0.1598,
+      "step": 67200
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1138690225774787e-05,
+      "loss": 0.1652,
+      "step": 67250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1124666947132239e-05,
+      "loss": 0.1336,
+      "step": 67300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1110643668489694e-05,
+      "loss": 0.1574,
+      "step": 67350
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1096620389847147e-05,
+      "loss": 0.1644,
+      "step": 67400
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.10825971112046e-05,
+      "loss": 0.1637,
+      "step": 67450
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1068573832562052e-05,
+      "loss": 0.1396,
+      "step": 67500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1054550553919507e-05,
+      "loss": 0.1304,
+      "step": 67550
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.104052727527696e-05,
+      "loss": 0.153,
+      "step": 67600
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1026503996634414e-05,
+      "loss": 0.1324,
+      "step": 67650
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1012480717991866e-05,
+      "loss": 0.1581,
+      "step": 67700
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.099845743934932e-05,
+      "loss": 0.1593,
+      "step": 67750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0984434160706774e-05,
+      "loss": 0.1422,
+      "step": 67800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0970410882064227e-05,
+      "loss": 0.1491,
+      "step": 67850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.0956387603421679e-05,
+      "loss": 0.1727,
+      "step": 67900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0942364324779134e-05,
+      "loss": 0.1576,
+      "step": 67950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0928341046136588e-05,
+      "loss": 0.1701,
+      "step": 68000
+    },
+    {
+      "epoch": 1.91,
+      "eval_bleu": 96.0842,
+      "eval_gen_len": 64.2387,
+      "eval_loss": 0.24398915469646454,
+      "eval_rouge1": 93.9618,
+      "eval_rouge2": 90.4097,
+      "eval_rougeL": 93.7775,
+      "eval_rougeLsum": 93.7665,
+      "eval_runtime": 3135.9742,
+      "eval_samples_per_second": 0.957,
+      "eval_steps_per_second": 0.239,
+      "step": 68000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0914317767494041e-05,
+      "loss": 0.1359,
+      "step": 68050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0900294488851493e-05,
+      "loss": 0.1762,
+      "step": 68100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0886271210208948e-05,
+      "loss": 0.1596,
+      "step": 68150
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0872247931566401e-05,
+      "loss": 0.1574,
+      "step": 68200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.0858224652923854e-05,
+      "loss": 0.1334,
+      "step": 68250
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0844201374281306e-05,
+      "loss": 0.1456,
+      "step": 68300
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0830178095638761e-05,
+      "loss": 0.1625,
+      "step": 68350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0816154816996214e-05,
+      "loss": 0.1715,
+      "step": 68400
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0802131538353668e-05,
+      "loss": 0.1444,
+      "step": 68450
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.078810825971112e-05,
+      "loss": 0.1748,
+      "step": 68500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0774084981068574e-05,
+      "loss": 0.1522,
+      "step": 68550
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.0760061702426028e-05,
+      "loss": 0.1364,
+      "step": 68600
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0746038423783481e-05,
+      "loss": 0.1557,
+      "step": 68650
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0732015145140935e-05,
+      "loss": 0.1399,
+      "step": 68700
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0717991866498388e-05,
+      "loss": 0.1478,
+      "step": 68750
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0703968587855841e-05,
+      "loss": 0.1444,
+      "step": 68800
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0689945309213295e-05,
+      "loss": 0.1419,
+      "step": 68850
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0675922030570748e-05,
+      "loss": 0.1642,
+      "step": 68900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0661898751928201e-05,
+      "loss": 0.1646,
+      "step": 68950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0647875473285655e-05,
+      "loss": 0.161,
+      "step": 69000
+    },
+    {
+      "epoch": 1.94,
+      "eval_bleu": 96.1059,
+      "eval_gen_len": 64.2643,
+      "eval_loss": 0.24390248954296112,
+      "eval_rouge1": 93.9311,
+      "eval_rouge2": 90.376,
+      "eval_rougeL": 93.746,
+      "eval_rougeLsum": 93.7269,
+      "eval_runtime": 3190.0005,
+      "eval_samples_per_second": 0.94,
+      "eval_steps_per_second": 0.235,
+      "step": 69000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0633852194643108e-05,
+      "loss": 0.1595,
+      "step": 69050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0619828916000561e-05,
+      "loss": 0.1468,
+      "step": 69100
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0605805637358015e-05,
+      "loss": 0.1417,
+      "step": 69150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0591782358715468e-05,
+      "loss": 0.1555,
+      "step": 69200
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0577759080072921e-05,
+      "loss": 0.1611,
+      "step": 69250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0563735801430375e-05,
+      "loss": 0.1587,
+      "step": 69300
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0549712522787828e-05,
+      "loss": 0.1521,
+      "step": 69350
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0535689244145282e-05,
+      "loss": 0.1356,
+      "step": 69400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0521665965502735e-05,
+      "loss": 0.1494,
+      "step": 69450
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0507642686860188e-05,
+      "loss": 0.1391,
+      "step": 69500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0493619408217642e-05,
+      "loss": 0.1631,
+      "step": 69550
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0479596129575095e-05,
+      "loss": 0.1526,
+      "step": 69600
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0465572850932548e-05,
+      "loss": 0.1431,
+      "step": 69650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0451549572290002e-05,
+      "loss": 0.161,
+      "step": 69700
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0437526293647455e-05,
+      "loss": 0.1628,
+      "step": 69750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0423503015004908e-05,
+      "loss": 0.1529,
+      "step": 69800
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0409479736362362e-05,
+      "loss": 0.1452,
+      "step": 69850
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0395456457719815e-05,
+      "loss": 0.1459,
+      "step": 69900
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0381433179077268e-05,
+      "loss": 0.1444,
+      "step": 69950
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0367409900434722e-05,
+      "loss": 0.1316,
+      "step": 70000
+    },
+    {
+      "epoch": 1.96,
+      "eval_bleu": 96.0919,
+      "eval_gen_len": 64.248,
+      "eval_loss": 0.24350972473621368,
+      "eval_rouge1": 93.9235,
+      "eval_rouge2": 90.3844,
+      "eval_rougeL": 93.751,
+      "eval_rougeLsum": 93.7278,
+      "eval_runtime": 3089.9712,
+      "eval_samples_per_second": 0.971,
+      "eval_steps_per_second": 0.243,
+      "step": 70000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0353386621792175e-05,
+      "loss": 0.1485,
+      "step": 70050
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0339363343149629e-05,
+      "loss": 0.1559,
+      "step": 70100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0325340064507084e-05,
+      "loss": 0.1483,
+      "step": 70150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0311316785864535e-05,
+      "loss": 0.1845,
+      "step": 70200
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0297293507221989e-05,
+      "loss": 0.1417,
+      "step": 70250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0283270228579442e-05,
+      "loss": 0.1341,
+      "step": 70300
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0269246949936897e-05,
+      "loss": 0.1462,
+      "step": 70350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0255223671294349e-05,
+      "loss": 0.1721,
+      "step": 70400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0241200392651802e-05,
+      "loss": 0.1549,
+      "step": 70450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0227177114009255e-05,
+      "loss": 0.16,
+      "step": 70500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.021315383536671e-05,
+      "loss": 0.1455,
+      "step": 70550
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0199130556724162e-05,
+      "loss": 0.1312,
+      "step": 70600
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0185107278081615e-05,
+      "loss": 0.1607,
+      "step": 70650
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0171083999439069e-05,
+      "loss": 0.1483,
+      "step": 70700
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0157060720796524e-05,
+      "loss": 0.1725,
+      "step": 70750
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0143037442153976e-05,
+      "loss": 0.1598,
+      "step": 70800
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0129014163511429e-05,
+      "loss": 0.1384,
+      "step": 70850
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0114990884868882e-05,
+      "loss": 0.1448,
+      "step": 70900
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0100967606226337e-05,
+      "loss": 0.1534,
+      "step": 70950
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0086944327583789e-05,
+      "loss": 0.136,
+      "step": 71000
+    },
+    {
+      "epoch": 1.99,
+      "eval_bleu": 96.0747,
+      "eval_gen_len": 64.2193,
+      "eval_loss": 0.24425600469112396,
+      "eval_rouge1": 93.9676,
+      "eval_rouge2": 90.4404,
+      "eval_rougeL": 93.7902,
+      "eval_rougeLsum": 93.7617,
+      "eval_runtime": 3086.9294,
+      "eval_samples_per_second": 0.972,
+      "eval_steps_per_second": 0.243,
+      "step": 71000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0072921048941242e-05,
+      "loss": 0.163,
+      "step": 71050
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0058897770298696e-05,
+      "loss": 0.1373,
+      "step": 71100
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.004487449165615e-05,
+      "loss": 0.1452,
+      "step": 71150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0030851213013602e-05,
+      "loss": 0.1583,
+      "step": 71200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0016827934371056e-05,
+      "loss": 0.1396,
+      "step": 71250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0002804655728509e-05,
+      "loss": 0.1309,
+      "step": 71300
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.988781377085964e-06,
+      "loss": 0.1628,
+      "step": 71350
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.974758098443416e-06,
+      "loss": 0.1354,
+      "step": 71400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96073481980087e-06,
+      "loss": 0.1565,
+      "step": 71450
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.946711541158323e-06,
+      "loss": 0.1411,
+      "step": 71500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.932688262515778e-06,
+      "loss": 0.1681,
+      "step": 71550
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.918664983873231e-06,
+      "loss": 0.1406,
+      "step": 71600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.904641705230683e-06,
+      "loss": 0.1308,
+      "step": 71650
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.890618426588136e-06,
+      "loss": 0.1221,
+      "step": 71700
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.876595147945591e-06,
+      "loss": 0.1325,
+      "step": 71750
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.862571869303044e-06,
+      "loss": 0.1377,
+      "step": 71800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.848548590660496e-06,
+      "loss": 0.1437,
+      "step": 71850
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.83452531201795e-06,
+      "loss": 0.153,
+      "step": 71900
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.820502033375404e-06,
+      "loss": 0.1462,
+      "step": 71950
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.806478754732858e-06,
+      "loss": 0.1651,
+      "step": 72000
+    },
+    {
+      "epoch": 2.02,
+      "eval_bleu": 96.1009,
+      "eval_gen_len": 64.2463,
+      "eval_loss": 0.24310755729675293,
+      "eval_rouge1": 94.0005,
+      "eval_rouge2": 90.4979,
+      "eval_rougeL": 93.8136,
+      "eval_rougeLsum": 93.7931,
+      "eval_runtime": 3096.8678,
+      "eval_samples_per_second": 0.969,
+      "eval_steps_per_second": 0.242,
+      "step": 72000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.79245547609031e-06,
+      "loss": 0.1675,
+      "step": 72050
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.778432197447763e-06,
+      "loss": 0.15,
+      "step": 72100
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.764408918805218e-06,
+      "loss": 0.1351,
+      "step": 72150
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.750385640162671e-06,
+      "loss": 0.1424,
+      "step": 72200
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.736362361520123e-06,
+      "loss": 0.1351,
+      "step": 72250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.722339082877576e-06,
+      "loss": 0.1265,
+      "step": 72300
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.708315804235031e-06,
+      "loss": 0.1696,
+      "step": 72350
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.694292525592485e-06,
+      "loss": 0.1451,
+      "step": 72400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.680269246949936e-06,
+      "loss": 0.1417,
+      "step": 72450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.66624596830739e-06,
+      "loss": 0.1383,
+      "step": 72500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.652222689664845e-06,
+      "loss": 0.1393,
+      "step": 72550
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.638199411022298e-06,
+      "loss": 0.1378,
+      "step": 72600
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.62417613237975e-06,
+      "loss": 0.1632,
+      "step": 72650
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.610152853737203e-06,
+      "loss": 0.152,
+      "step": 72700
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.596129575094658e-06,
+      "loss": 0.1392,
+      "step": 72750
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.582106296452112e-06,
+      "loss": 0.1381,
+      "step": 72800
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.568083017809565e-06,
+      "loss": 0.1341,
+      "step": 72850
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.554059739167017e-06,
+      "loss": 0.1372,
+      "step": 72900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.540036460524472e-06,
+      "loss": 0.1521,
+      "step": 72950
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.526013181881925e-06,
+      "loss": 0.1552,
+      "step": 73000
+    },
+    {
+      "epoch": 2.05,
+      "eval_bleu": 96.0944,
+      "eval_gen_len": 64.2553,
+      "eval_loss": 0.2440294772386551,
+      "eval_rouge1": 93.9391,
+      "eval_rouge2": 90.4451,
+      "eval_rougeL": 93.7575,
+      "eval_rougeLsum": 93.7421,
+      "eval_runtime": 3104.3345,
+      "eval_samples_per_second": 0.966,
+      "eval_steps_per_second": 0.242,
+      "step": 73000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.511989903239378e-06,
+      "loss": 0.1374,
+      "step": 73050
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.49796662459683e-06,
+      "loss": 0.1416,
+      "step": 73100
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.483943345954285e-06,
+      "loss": 0.1511,
+      "step": 73150
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.469920067311738e-06,
+      "loss": 0.1287,
+      "step": 73200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.455896788669192e-06,
+      "loss": 0.154,
+      "step": 73250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.441873510026643e-06,
+      "loss": 0.1399,
+      "step": 73300
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.427850231384098e-06,
+      "loss": 0.1371,
+      "step": 73350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.413826952741552e-06,
+      "loss": 0.1424,
+      "step": 73400
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.399803674099005e-06,
+      "loss": 0.1441,
+      "step": 73450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.385780395456457e-06,
+      "loss": 0.1276,
+      "step": 73500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.371757116813912e-06,
+      "loss": 0.1268,
+      "step": 73550
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.357733838171365e-06,
+      "loss": 0.1587,
+      "step": 73600
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.343710559528819e-06,
+      "loss": 0.1475,
+      "step": 73650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.32968728088627e-06,
+      "loss": 0.1288,
+      "step": 73700
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.315664002243725e-06,
+      "loss": 0.1432,
+      "step": 73750
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.301640723601179e-06,
+      "loss": 0.148,
+      "step": 73800
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.287617444958632e-06,
+      "loss": 0.1506,
+      "step": 73850
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.273594166316084e-06,
+      "loss": 0.1345,
+      "step": 73900
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.259570887673539e-06,
+      "loss": 0.1276,
+      "step": 73950
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.245547609030992e-06,
+      "loss": 0.1491,
+      "step": 74000
+    },
+    {
+      "epoch": 2.08,
+      "eval_bleu": 96.07,
+      "eval_gen_len": 64.2557,
+      "eval_loss": 0.24399851262569427,
+      "eval_rouge1": 93.9143,
+      "eval_rouge2": 90.3682,
+      "eval_rougeL": 93.7204,
+      "eval_rougeLsum": 93.712,
+      "eval_runtime": 3103.8893,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 74000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.231524330388445e-06,
+      "loss": 0.1277,
+      "step": 74050
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.217501051745897e-06,
+      "loss": 0.1306,
+      "step": 74100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.203477773103352e-06,
+      "loss": 0.1371,
+      "step": 74150
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.189454494460806e-06,
+      "loss": 0.1384,
+      "step": 74200
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.175431215818259e-06,
+      "loss": 0.1405,
+      "step": 74250
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.161407937175712e-06,
+      "loss": 0.1373,
+      "step": 74300
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.147384658533166e-06,
+      "loss": 0.1418,
+      "step": 74350
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.133361379890619e-06,
+      "loss": 0.126,
+      "step": 74400
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.119338101248072e-06,
+      "loss": 0.1738,
+      "step": 74450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.105314822605526e-06,
+      "loss": 0.1466,
+      "step": 74500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.091291543962979e-06,
+      "loss": 0.1381,
+      "step": 74550
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.077268265320432e-06,
+      "loss": 0.1384,
+      "step": 74600
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.063244986677886e-06,
+      "loss": 0.1442,
+      "step": 74650
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.049221708035339e-06,
+      "loss": 0.1371,
+      "step": 74700
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.035198429392793e-06,
+      "loss": 0.1331,
+      "step": 74750
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.021175150750246e-06,
+      "loss": 0.136,
+      "step": 74800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.0071518721077e-06,
+      "loss": 0.1422,
+      "step": 74850
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.993128593465153e-06,
+      "loss": 0.1386,
+      "step": 74900
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.979105314822606e-06,
+      "loss": 0.1247,
+      "step": 74950
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.96508203618006e-06,
+      "loss": 0.1567,
+      "step": 75000
+    },
+    {
+      "epoch": 2.1,
+      "eval_bleu": 96.1043,
+      "eval_gen_len": 64.258,
+      "eval_loss": 0.24418920278549194,
+      "eval_rouge1": 93.9448,
+      "eval_rouge2": 90.3671,
+      "eval_rougeL": 93.7538,
+      "eval_rougeLsum": 93.7435,
+      "eval_runtime": 3102.9976,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 75000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 8.951058757537513e-06,
+      "loss": 0.1308,
+      "step": 75050
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.937035478894966e-06,
+      "loss": 0.1387,
+      "step": 75100
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.92301220025242e-06,
+      "loss": 0.1356,
+      "step": 75150
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.908988921609873e-06,
+      "loss": 0.1397,
+      "step": 75200
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.894965642967326e-06,
+      "loss": 0.1276,
+      "step": 75250
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.88094236432478e-06,
+      "loss": 0.1354,
+      "step": 75300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.866919085682233e-06,
+      "loss": 0.1537,
+      "step": 75350
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 8.852895807039686e-06,
+      "loss": 0.1508,
+      "step": 75400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.83887252839714e-06,
+      "loss": 0.1659,
+      "step": 75450
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.824849249754593e-06,
+      "loss": 0.1433,
+      "step": 75500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.810825971112046e-06,
+      "loss": 0.1315,
+      "step": 75550
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.7968026924695e-06,
+      "loss": 0.1257,
+      "step": 75600
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.782779413826953e-06,
+      "loss": 0.1226,
+      "step": 75650
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.768756135184406e-06,
+      "loss": 0.1397,
+      "step": 75700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 8.754732856541861e-06,
+      "loss": 0.1245,
+      "step": 75750
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.740709577899313e-06,
+      "loss": 0.1508,
+      "step": 75800
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.726686299256766e-06,
+      "loss": 0.152,
+      "step": 75850
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.71266302061422e-06,
+      "loss": 0.1475,
+      "step": 75900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.698639741971675e-06,
+      "loss": 0.1448,
+      "step": 75950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.684616463329126e-06,
+      "loss": 0.1444,
+      "step": 76000
+    },
+    {
+      "epoch": 2.13,
+      "eval_bleu": 96.0981,
+      "eval_gen_len": 64.2277,
+      "eval_loss": 0.24386686086654663,
+      "eval_rouge1": 93.9613,
+      "eval_rouge2": 90.4186,
+      "eval_rougeL": 93.7791,
+      "eval_rougeLsum": 93.7578,
+      "eval_runtime": 3114.9073,
+      "eval_samples_per_second": 0.963,
+      "eval_steps_per_second": 0.241,
+      "step": 76000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.67059318468658e-06,
+      "loss": 0.1552,
+      "step": 76050
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 8.656569906044033e-06,
+      "loss": 0.151,
+      "step": 76100
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.642546627401488e-06,
+      "loss": 0.1493,
+      "step": 76150
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.62852334875894e-06,
+      "loss": 0.1279,
+      "step": 76200
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.614500070116393e-06,
+      "loss": 0.1521,
+      "step": 76250
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.600476791473847e-06,
+      "loss": 0.1565,
+      "step": 76300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.586453512831302e-06,
+      "loss": 0.1412,
+      "step": 76350
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.572430234188753e-06,
+      "loss": 0.138,
+      "step": 76400
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 8.558406955546207e-06,
+      "loss": 0.1302,
+      "step": 76450
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.54438367690366e-06,
+      "loss": 0.1542,
+      "step": 76500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.530360398261115e-06,
+      "loss": 0.1341,
+      "step": 76550
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.516337119618567e-06,
+      "loss": 0.1643,
+      "step": 76600
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.50231384097602e-06,
+      "loss": 0.1574,
+      "step": 76650
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.488290562333473e-06,
+      "loss": 0.1382,
+      "step": 76700
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.474267283690929e-06,
+      "loss": 0.1355,
+      "step": 76750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 8.46024400504838e-06,
+      "loss": 0.1396,
+      "step": 76800
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.446220726405834e-06,
+      "loss": 0.1244,
+      "step": 76850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.432197447763287e-06,
+      "loss": 0.1291,
+      "step": 76900
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.418174169120742e-06,
+      "loss": 0.1402,
+      "step": 76950
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.404150890478194e-06,
+      "loss": 0.151,
+      "step": 77000
+    },
+    {
+      "epoch": 2.16,
+      "eval_bleu": 96.1028,
+      "eval_gen_len": 64.2047,
+      "eval_loss": 0.243386372923851,
+      "eval_rouge1": 93.9323,
+      "eval_rouge2": 90.4104,
+      "eval_rougeL": 93.7353,
+      "eval_rougeLsum": 93.7303,
+      "eval_runtime": 3082.1904,
+      "eval_samples_per_second": 0.973,
+      "eval_steps_per_second": 0.243,
+      "step": 77000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.390127611835647e-06,
+      "loss": 0.1471,
+      "step": 77050
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.3761043331931e-06,
+      "loss": 0.1363,
+      "step": 77100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 8.362081054550555e-06,
+      "loss": 0.1174,
+      "step": 77150
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.348057775908009e-06,
+      "loss": 0.1249,
+      "step": 77200
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.33403449726546e-06,
+      "loss": 0.1323,
+      "step": 77250
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.320011218622914e-06,
+      "loss": 0.1441,
+      "step": 77300
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.305987939980369e-06,
+      "loss": 0.1354,
+      "step": 77350
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.291964661337822e-06,
+      "loss": 0.1299,
+      "step": 77400
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.277941382695274e-06,
+      "loss": 0.16,
+      "step": 77450
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 8.263918104052727e-06,
+      "loss": 0.1471,
+      "step": 77500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.249894825410182e-06,
+      "loss": 0.1507,
+      "step": 77550
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.235871546767636e-06,
+      "loss": 0.1345,
+      "step": 77600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.221848268125087e-06,
+      "loss": 0.1597,
+      "step": 77650
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.20782498948254e-06,
+      "loss": 0.1361,
+      "step": 77700
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.193801710839996e-06,
+      "loss": 0.1437,
+      "step": 77750
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.179778432197449e-06,
+      "loss": 0.1438,
+      "step": 77800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.1657551535549e-06,
+      "loss": 0.1442,
+      "step": 77850
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.151731874912354e-06,
+      "loss": 0.1451,
+      "step": 77900
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.137708596269809e-06,
+      "loss": 0.1455,
+      "step": 77950
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.123685317627262e-06,
+      "loss": 0.1331,
+      "step": 78000
+    },
+    {
+      "epoch": 2.19,
+      "eval_bleu": 96.0923,
+      "eval_gen_len": 64.236,
+      "eval_loss": 0.24420137703418732,
+      "eval_rouge1": 93.9515,
+      "eval_rouge2": 90.4414,
+      "eval_rougeL": 93.7733,
+      "eval_rougeLsum": 93.7553,
+      "eval_runtime": 3102.8715,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 78000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.109662038984714e-06,
+      "loss": 0.1332,
+      "step": 78050
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.095638760342167e-06,
+      "loss": 0.1361,
+      "step": 78100
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.081615481699623e-06,
+      "loss": 0.1394,
+      "step": 78150
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.067592203057076e-06,
+      "loss": 0.1387,
+      "step": 78200
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.053568924414528e-06,
+      "loss": 0.1264,
+      "step": 78250
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.039545645771981e-06,
+      "loss": 0.1464,
+      "step": 78300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.025522367129436e-06,
+      "loss": 0.1574,
+      "step": 78350
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.01149908848689e-06,
+      "loss": 0.1449,
+      "step": 78400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.997475809844341e-06,
+      "loss": 0.1361,
+      "step": 78450
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.983452531201794e-06,
+      "loss": 0.1254,
+      "step": 78500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.96942925255925e-06,
+      "loss": 0.1396,
+      "step": 78550
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 7.955405973916703e-06,
+      "loss": 0.1437,
+      "step": 78600
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.941382695274156e-06,
+      "loss": 0.1434,
+      "step": 78650
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.927359416631608e-06,
+      "loss": 0.1273,
+      "step": 78700
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.913336137989063e-06,
+      "loss": 0.1393,
+      "step": 78750
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.899312859346516e-06,
+      "loss": 0.1356,
+      "step": 78800
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.88528958070397e-06,
+      "loss": 0.1554,
+      "step": 78850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.871266302061421e-06,
+      "loss": 0.1372,
+      "step": 78900
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 7.857243023418876e-06,
+      "loss": 0.1407,
+      "step": 78950
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.84321974477633e-06,
+      "loss": 0.1658,
+      "step": 79000
+    },
+    {
+      "epoch": 2.22,
+      "eval_bleu": 96.0964,
+      "eval_gen_len": 64.256,
+      "eval_loss": 0.24366088211536407,
+      "eval_rouge1": 93.9674,
+      "eval_rouge2": 90.4223,
+      "eval_rougeL": 93.7734,
+      "eval_rougeLsum": 93.7712,
+      "eval_runtime": 3096.7975,
+      "eval_samples_per_second": 0.969,
+      "eval_steps_per_second": 0.242,
+      "step": 79000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.829196466133783e-06,
+      "loss": 0.1379,
+      "step": 79050
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.815173187491235e-06,
+      "loss": 0.1474,
+      "step": 79100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.80114990884869e-06,
+      "loss": 0.1297,
+      "step": 79150
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.787126630206143e-06,
+      "loss": 0.145,
+      "step": 79200
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.773103351563596e-06,
+      "loss": 0.1223,
+      "step": 79250
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 7.759080072921048e-06,
+      "loss": 0.1425,
+      "step": 79300
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.745056794278503e-06,
+      "loss": 0.1358,
+      "step": 79350
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.731033515635956e-06,
+      "loss": 0.1391,
+      "step": 79400
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.71701023699341e-06,
+      "loss": 0.1581,
+      "step": 79450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.702986958350861e-06,
+      "loss": 0.1425,
+      "step": 79500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.688963679708317e-06,
+      "loss": 0.15,
+      "step": 79550
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.67494040106577e-06,
+      "loss": 0.1496,
+      "step": 79600
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 7.660917122423223e-06,
+      "loss": 0.1609,
+      "step": 79650
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.646893843780675e-06,
+      "loss": 0.1502,
+      "step": 79700
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.63287056513813e-06,
+      "loss": 0.1345,
+      "step": 79750
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.618847286495583e-06,
+      "loss": 0.1429,
+      "step": 79800
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.604824007853036e-06,
+      "loss": 0.1408,
+      "step": 79850
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.590800729210489e-06,
+      "loss": 0.1613,
+      "step": 79900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.576777450567943e-06,
+      "loss": 0.1605,
+      "step": 79950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 7.562754171925397e-06,
+      "loss": 0.1396,
+      "step": 80000
+    },
+    {
+      "epoch": 2.24,
+      "eval_bleu": 96.1134,
+      "eval_gen_len": 64.2447,
+      "eval_loss": 0.2430116832256317,
+      "eval_rouge1": 93.9967,
+      "eval_rouge2": 90.463,
+      "eval_rougeL": 93.817,
+      "eval_rougeLsum": 93.8046,
+      "eval_runtime": 3075.4099,
+      "eval_samples_per_second": 0.975,
+      "eval_steps_per_second": 0.244,
+      "step": 80000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.54873089328285e-06,
+      "loss": 0.147,
+      "step": 80050
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.534707614640303e-06,
+      "loss": 0.1501,
+      "step": 80100
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.520684335997758e-06,
+      "loss": 0.1215,
+      "step": 80150
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.50666105735521e-06,
+      "loss": 0.1323,
+      "step": 80200
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.4926377787126635e-06,
+      "loss": 0.1307,
+      "step": 80250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.478614500070117e-06,
+      "loss": 0.1369,
+      "step": 80300
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.46459122142757e-06,
+      "loss": 0.1502,
+      "step": 80350
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 7.450567942785024e-06,
+      "loss": 0.14,
+      "step": 80400
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.436544664142477e-06,
+      "loss": 0.1411,
+      "step": 80450
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.42252138549993e-06,
+      "loss": 0.1401,
+      "step": 80500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.408498106857384e-06,
+      "loss": 0.1713,
+      "step": 80550
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.394474828214837e-06,
+      "loss": 0.1678,
+      "step": 80600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.38045154957229e-06,
+      "loss": 0.1203,
+      "step": 80650
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.366428270929744e-06,
+      "loss": 0.1635,
+      "step": 80700
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 7.352404992287197e-06,
+      "loss": 0.1318,
+      "step": 80750
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.3383817136446505e-06,
+      "loss": 0.1292,
+      "step": 80800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.324358435002104e-06,
+      "loss": 0.1433,
+      "step": 80850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.310335156359557e-06,
+      "loss": 0.1572,
+      "step": 80900
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.2963118777170105e-06,
+      "loss": 0.1338,
+      "step": 80950
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.282288599074464e-06,
+      "loss": 0.147,
+      "step": 81000
+    },
+    {
+      "epoch": 2.27,
+      "eval_bleu": 96.1115,
+      "eval_gen_len": 64.2613,
+      "eval_loss": 0.24394148588180542,
+      "eval_rouge1": 93.9741,
+      "eval_rouge2": 90.4331,
+      "eval_rougeL": 93.775,
+      "eval_rougeLsum": 93.7668,
+      "eval_runtime": 3119.4921,
+      "eval_samples_per_second": 0.962,
+      "eval_steps_per_second": 0.24,
+      "step": 81000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.268265320431917e-06,
+      "loss": 0.1329,
+      "step": 81050
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 7.254242041789371e-06,
+      "loss": 0.1416,
+      "step": 81100
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.240218763146824e-06,
+      "loss": 0.1443,
+      "step": 81150
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.226195484504277e-06,
+      "loss": 0.1519,
+      "step": 81200
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.212172205861731e-06,
+      "loss": 0.1501,
+      "step": 81250
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.198148927219184e-06,
+      "loss": 0.1465,
+      "step": 81300
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.184125648576637e-06,
+      "loss": 0.1639,
+      "step": 81350
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.170102369934091e-06,
+      "loss": 0.1335,
+      "step": 81400
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 7.156079091291544e-06,
+      "loss": 0.1588,
+      "step": 81450
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.142055812648998e-06,
+      "loss": 0.1513,
+      "step": 81500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.128032534006451e-06,
+      "loss": 0.1523,
+      "step": 81550
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.114009255363905e-06,
+      "loss": 0.1477,
+      "step": 81600
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.0999859767213576e-06,
+      "loss": 0.1269,
+      "step": 81650
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.085962698078812e-06,
+      "loss": 0.1501,
+      "step": 81700
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.071939419436264e-06,
+      "loss": 0.1421,
+      "step": 81750
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 7.0579161407937185e-06,
+      "loss": 0.151,
+      "step": 81800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.043892862151171e-06,
+      "loss": 0.1383,
+      "step": 81850
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.029869583508625e-06,
+      "loss": 0.142,
+      "step": 81900
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.015846304866078e-06,
+      "loss": 0.1574,
+      "step": 81950
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 7.001823026223532e-06,
+      "loss": 0.1601,
+      "step": 82000
+    },
+    {
+      "epoch": 2.3,
+      "eval_bleu": 96.0955,
+      "eval_gen_len": 64.251,
+      "eval_loss": 0.24331849813461304,
+      "eval_rouge1": 93.9796,
+      "eval_rouge2": 90.4277,
+      "eval_rougeL": 93.7849,
+      "eval_rougeLsum": 93.7712,
+      "eval_runtime": 3157.1975,
+      "eval_samples_per_second": 0.95,
+      "eval_steps_per_second": 0.238,
+      "step": 82000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.987799747580984e-06,
+      "loss": 0.1371,
+      "step": 82050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.973776468938439e-06,
+      "loss": 0.1392,
+      "step": 82100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.959753190295891e-06,
+      "loss": 0.1309,
+      "step": 82150
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.945729911653345e-06,
+      "loss": 0.1614,
+      "step": 82200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.931706633010798e-06,
+      "loss": 0.1377,
+      "step": 82250
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.917683354368251e-06,
+      "loss": 0.1473,
+      "step": 82300
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.9036600757257046e-06,
+      "loss": 0.145,
+      "step": 82350
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.889636797083158e-06,
+      "loss": 0.1601,
+      "step": 82400
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.875613518440611e-06,
+      "loss": 0.1367,
+      "step": 82450
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.861590239798065e-06,
+      "loss": 0.1492,
+      "step": 82500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.847566961155518e-06,
+      "loss": 0.1403,
+      "step": 82550
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.833543682512971e-06,
+      "loss": 0.1419,
+      "step": 82600
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.819520403870425e-06,
+      "loss": 0.1287,
+      "step": 82650
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.805497125227878e-06,
+      "loss": 0.14,
+      "step": 82700
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.7914738465853314e-06,
+      "loss": 0.1348,
+      "step": 82750
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.777450567942785e-06,
+      "loss": 0.1375,
+      "step": 82800
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.763427289300239e-06,
+      "loss": 0.1324,
+      "step": 82850
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.7494040106576915e-06,
+      "loss": 0.1366,
+      "step": 82900
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.735380732015146e-06,
+      "loss": 0.1333,
+      "step": 82950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.721357453372598e-06,
+      "loss": 0.1389,
+      "step": 83000
+    },
+    {
+      "epoch": 2.33,
+      "eval_bleu": 96.1354,
+      "eval_gen_len": 64.263,
+      "eval_loss": 0.24297164380550385,
+      "eval_rouge1": 93.9976,
+      "eval_rouge2": 90.4522,
+      "eval_rougeL": 93.8079,
+      "eval_rougeLsum": 93.7903,
+      "eval_runtime": 3153.994,
+      "eval_samples_per_second": 0.951,
+      "eval_steps_per_second": 0.238,
+      "step": 83000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.707334174730052e-06,
+      "loss": 0.1571,
+      "step": 83050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.693310896087505e-06,
+      "loss": 0.1383,
+      "step": 83100
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.679287617444959e-06,
+      "loss": 0.1524,
+      "step": 83150
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.665264338802412e-06,
+      "loss": 0.1568,
+      "step": 83200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.651241060159866e-06,
+      "loss": 0.1226,
+      "step": 83250
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.637217781517318e-06,
+      "loss": 0.1259,
+      "step": 83300
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.6231945028747726e-06,
+      "loss": 0.1466,
+      "step": 83350
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.609171224232225e-06,
+      "loss": 0.1464,
+      "step": 83400
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.595147945589679e-06,
+      "loss": 0.1355,
+      "step": 83450
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.581124666947132e-06,
+      "loss": 0.1387,
+      "step": 83500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.567101388304586e-06,
+      "loss": 0.1477,
+      "step": 83550
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 6.5530781096620385e-06,
+      "loss": 0.1507,
+      "step": 83600
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.539054831019493e-06,
+      "loss": 0.1477,
+      "step": 83650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.525031552376945e-06,
+      "loss": 0.1483,
+      "step": 83700
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.5110082737343994e-06,
+      "loss": 0.1524,
+      "step": 83750
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.496984995091852e-06,
+      "loss": 0.1522,
+      "step": 83800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.482961716449306e-06,
+      "loss": 0.1347,
+      "step": 83850
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.468938437806759e-06,
+      "loss": 0.1562,
+      "step": 83900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 6.454915159164213e-06,
+      "loss": 0.1554,
+      "step": 83950
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.440891880521665e-06,
+      "loss": 0.1416,
+      "step": 84000
+    },
+    {
+      "epoch": 2.36,
+      "eval_bleu": 96.1053,
+      "eval_gen_len": 64.255,
+      "eval_loss": 0.2427954524755478,
+      "eval_rouge1": 93.9492,
+      "eval_rouge2": 90.4232,
+      "eval_rougeL": 93.7656,
+      "eval_rougeLsum": 93.7482,
+      "eval_runtime": 3082.0924,
+      "eval_samples_per_second": 0.973,
+      "eval_steps_per_second": 0.243,
+      "step": 84000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.4268686018791196e-06,
+      "loss": 0.1348,
+      "step": 84050
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.412845323236572e-06,
+      "loss": 0.1689,
+      "step": 84100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.398822044594026e-06,
+      "loss": 0.139,
+      "step": 84150
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.384798765951479e-06,
+      "loss": 0.1409,
+      "step": 84200
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.370775487308933e-06,
+      "loss": 0.1619,
+      "step": 84250
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 6.356752208666386e-06,
+      "loss": 0.1342,
+      "step": 84300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.34272893002384e-06,
+      "loss": 0.1418,
+      "step": 84350
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.328705651381293e-06,
+      "loss": 0.1504,
+      "step": 84400
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.3146823727387464e-06,
+      "loss": 0.1529,
+      "step": 84450
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.3006590940962e-06,
+      "loss": 0.1363,
+      "step": 84500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.286635815453653e-06,
+      "loss": 0.1557,
+      "step": 84550
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.2726125368111065e-06,
+      "loss": 0.14,
+      "step": 84600
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 6.25858925816856e-06,
+      "loss": 0.139,
+      "step": 84650
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.244565979526013e-06,
+      "loss": 0.1357,
+      "step": 84700
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.230542700883467e-06,
+      "loss": 0.154,
+      "step": 84750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.21651942224092e-06,
+      "loss": 0.1294,
+      "step": 84800
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.202496143598373e-06,
+      "loss": 0.1392,
+      "step": 84850
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.188472864955827e-06,
+      "loss": 0.1495,
+      "step": 84900
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.17444958631328e-06,
+      "loss": 0.147,
+      "step": 84950
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 6.160426307670733e-06,
+      "loss": 0.1411,
+      "step": 85000
+    },
+    {
+      "epoch": 2.38,
+      "eval_bleu": 96.1115,
+      "eval_gen_len": 64.2543,
+      "eval_loss": 0.24300958216190338,
+      "eval_rouge1": 93.9688,
+      "eval_rouge2": 90.4201,
+      "eval_rougeL": 93.7772,
+      "eval_rougeLsum": 93.7569,
+      "eval_runtime": 3128.8052,
+      "eval_samples_per_second": 0.959,
+      "eval_steps_per_second": 0.24,
+      "step": 85000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.146403029028187e-06,
+      "loss": 0.1596,
+      "step": 85050
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.13237975038564e-06,
+      "loss": 0.1376,
+      "step": 85100
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.1183564717430934e-06,
+      "loss": 0.1409,
+      "step": 85150
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.104333193100547e-06,
+      "loss": 0.1576,
+      "step": 85200
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.090309914458e-06,
+      "loss": 0.1442,
+      "step": 85250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.0762866358154535e-06,
+      "loss": 0.1338,
+      "step": 85300
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 6.062263357172907e-06,
+      "loss": 0.1561,
+      "step": 85350
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.04824007853036e-06,
+      "loss": 0.1302,
+      "step": 85400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.034216799887814e-06,
+      "loss": 0.1793,
+      "step": 85450
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.020193521245267e-06,
+      "loss": 0.1602,
+      "step": 85500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 6.00617024260272e-06,
+      "loss": 0.143,
+      "step": 85550
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.992146963960174e-06,
+      "loss": 0.1253,
+      "step": 85600
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.978123685317627e-06,
+      "loss": 0.141,
+      "step": 85650
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.96410040667508e-06,
+      "loss": 0.1298,
+      "step": 85700
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.950077128032535e-06,
+      "loss": 0.1495,
+      "step": 85750
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.936053849389987e-06,
+      "loss": 0.1397,
+      "step": 85800
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.922030570747441e-06,
+      "loss": 0.118,
+      "step": 85850
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.908007292104894e-06,
+      "loss": 0.1418,
+      "step": 85900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.893984013462348e-06,
+      "loss": 0.1633,
+      "step": 85950
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.8799607348198005e-06,
+      "loss": 0.1351,
+      "step": 86000
+    },
+    {
+      "epoch": 2.41,
+      "eval_bleu": 96.1109,
+      "eval_gen_len": 64.2517,
+      "eval_loss": 0.24320924282073975,
+      "eval_rouge1": 93.9905,
+      "eval_rouge2": 90.4711,
+      "eval_rougeL": 93.7884,
+      "eval_rougeLsum": 93.7735,
+      "eval_runtime": 3131.4139,
+      "eval_samples_per_second": 0.958,
+      "eval_steps_per_second": 0.24,
+      "step": 86000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.865937456177255e-06,
+      "loss": 0.1365,
+      "step": 86050
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.851914177534707e-06,
+      "loss": 0.1397,
+      "step": 86100
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.8378908988921614e-06,
+      "loss": 0.1393,
+      "step": 86150
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.823867620249614e-06,
+      "loss": 0.1381,
+      "step": 86200
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.809844341607068e-06,
+      "loss": 0.1503,
+      "step": 86250
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.795821062964521e-06,
+      "loss": 0.1352,
+      "step": 86300
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.781797784321975e-06,
+      "loss": 0.1371,
+      "step": 86350
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.767774505679427e-06,
+      "loss": 0.1468,
+      "step": 86400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.753751227036882e-06,
+      "loss": 0.1402,
+      "step": 86450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.739727948394334e-06,
+      "loss": 0.1472,
+      "step": 86500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.725704669751788e-06,
+      "loss": 0.1517,
+      "step": 86550
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.711681391109241e-06,
+      "loss": 0.1518,
+      "step": 86600
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.697658112466695e-06,
+      "loss": 0.1366,
+      "step": 86650
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.6836348338241475e-06,
+      "loss": 0.1325,
+      "step": 86700
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.669611555181602e-06,
+      "loss": 0.1415,
+      "step": 86750
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.655588276539054e-06,
+      "loss": 0.155,
+      "step": 86800
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.6415649978965084e-06,
+      "loss": 0.1388,
+      "step": 86850
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.627541719253961e-06,
+      "loss": 0.1539,
+      "step": 86900
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.613518440611415e-06,
+      "loss": 0.1373,
+      "step": 86950
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.599495161968868e-06,
+      "loss": 0.1272,
+      "step": 87000
+    },
+    {
+      "epoch": 2.44,
+      "eval_bleu": 96.107,
+      "eval_gen_len": 64.2437,
+      "eval_loss": 0.2423429638147354,
+      "eval_rouge1": 93.9715,
+      "eval_rouge2": 90.4552,
+      "eval_rougeL": 93.783,
+      "eval_rougeLsum": 93.7701,
+      "eval_runtime": 3108.876,
+      "eval_samples_per_second": 0.965,
+      "eval_steps_per_second": 0.241,
+      "step": 87000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.585471883326322e-06,
+      "loss": 0.1392,
+      "step": 87050
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.571448604683775e-06,
+      "loss": 0.1521,
+      "step": 87100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.557425326041229e-06,
+      "loss": 0.1498,
+      "step": 87150
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.543402047398682e-06,
+      "loss": 0.1395,
+      "step": 87200
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.529378768756135e-06,
+      "loss": 0.1411,
+      "step": 87250
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.515355490113589e-06,
+      "loss": 0.141,
+      "step": 87300
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.501332211471042e-06,
+      "loss": 0.1477,
+      "step": 87350
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.487308932828495e-06,
+      "loss": 0.1756,
+      "step": 87400
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.473285654185949e-06,
+      "loss": 0.1696,
+      "step": 87450
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.459262375543402e-06,
+      "loss": 0.146,
+      "step": 87500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.4452390969008555e-06,
+      "loss": 0.1241,
+      "step": 87550
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.431215818258309e-06,
+      "loss": 0.147,
+      "step": 87600
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.417192539615762e-06,
+      "loss": 0.1288,
+      "step": 87650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.4031692609732155e-06,
+      "loss": 0.14,
+      "step": 87700
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.389145982330669e-06,
+      "loss": 0.1577,
+      "step": 87750
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.375122703688122e-06,
+      "loss": 0.1302,
+      "step": 87800
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.361099425045576e-06,
+      "loss": 0.1343,
+      "step": 87850
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.347076146403029e-06,
+      "loss": 0.1218,
+      "step": 87900
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.333052867760482e-06,
+      "loss": 0.1592,
+      "step": 87950
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.319029589117936e-06,
+      "loss": 0.1501,
+      "step": 88000
+    },
+    {
+      "epoch": 2.47,
+      "eval_bleu": 96.0986,
+      "eval_gen_len": 64.2647,
+      "eval_loss": 0.2428046017885208,
+      "eval_rouge1": 93.9542,
+      "eval_rouge2": 90.3776,
+      "eval_rougeL": 93.7639,
+      "eval_rougeLsum": 93.7476,
+      "eval_runtime": 3098.6862,
+      "eval_samples_per_second": 0.968,
+      "eval_steps_per_second": 0.242,
+      "step": 88000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.305006310475389e-06,
+      "loss": 0.1199,
+      "step": 88050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.290983031832842e-06,
+      "loss": 0.1412,
+      "step": 88100
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.276959753190296e-06,
+      "loss": 0.1489,
+      "step": 88150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.262936474547749e-06,
+      "loss": 0.1319,
+      "step": 88200
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.2489131959052025e-06,
+      "loss": 0.1308,
+      "step": 88250
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.234889917262656e-06,
+      "loss": 0.116,
+      "step": 88300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.220866638620109e-06,
+      "loss": 0.1496,
+      "step": 88350
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.2068433599775625e-06,
+      "loss": 0.1404,
+      "step": 88400
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.192820081335016e-06,
+      "loss": 0.1378,
+      "step": 88450
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.178796802692469e-06,
+      "loss": 0.1391,
+      "step": 88500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.1647735240499235e-06,
+      "loss": 0.1307,
+      "step": 88550
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.150750245407376e-06,
+      "loss": 0.1454,
+      "step": 88600
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.13672696676483e-06,
+      "loss": 0.1421,
+      "step": 88650
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.122703688122283e-06,
+      "loss": 0.1483,
+      "step": 88700
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.108680409479737e-06,
+      "loss": 0.1311,
+      "step": 88750
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.094657130837189e-06,
+      "loss": 0.1403,
+      "step": 88800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.080633852194644e-06,
+      "loss": 0.137,
+      "step": 88850
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.066610573552096e-06,
+      "loss": 0.1562,
+      "step": 88900
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.05258729490955e-06,
+      "loss": 0.1512,
+      "step": 88950
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.038564016267003e-06,
+      "loss": 0.1537,
+      "step": 89000
+    },
+    {
+      "epoch": 2.5,
+      "eval_bleu": 96.0983,
+      "eval_gen_len": 64.2437,
+      "eval_loss": 0.2420242875814438,
+      "eval_rouge1": 93.9781,
+      "eval_rouge2": 90.4393,
+      "eval_rougeL": 93.7904,
+      "eval_rougeLsum": 93.78,
+      "eval_runtime": 3102.7126,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 89000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.024540737624457e-06,
+      "loss": 0.1406,
+      "step": 89050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.0105174589819095e-06,
+      "loss": 0.1533,
+      "step": 89100
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.996494180339364e-06,
+      "loss": 0.1394,
+      "step": 89150
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.982470901696816e-06,
+      "loss": 0.1335,
+      "step": 89200
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.9684476230542705e-06,
+      "loss": 0.126,
+      "step": 89250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.954424344411723e-06,
+      "loss": 0.1383,
+      "step": 89300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.940401065769177e-06,
+      "loss": 0.1488,
+      "step": 89350
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.92637778712663e-06,
+      "loss": 0.151,
+      "step": 89400
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.912354508484084e-06,
+      "loss": 0.137,
+      "step": 89450
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.898331229841536e-06,
+      "loss": 0.1622,
+      "step": 89500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.884307951198991e-06,
+      "loss": 0.1332,
+      "step": 89550
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.870284672556443e-06,
+      "loss": 0.159,
+      "step": 89600
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 4.856261393913897e-06,
+      "loss": 0.1437,
+      "step": 89650
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.84223811527135e-06,
+      "loss": 0.1521,
+      "step": 89700
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.828214836628804e-06,
+      "loss": 0.1497,
+      "step": 89750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.8141915579862566e-06,
+      "loss": 0.1333,
+      "step": 89800
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.800168279343711e-06,
+      "loss": 0.1464,
+      "step": 89850
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.786145000701163e-06,
+      "loss": 0.1362,
+      "step": 89900
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.7721217220586175e-06,
+      "loss": 0.1379,
+      "step": 89950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.758098443416071e-06,
+      "loss": 0.1404,
+      "step": 90000
+    },
+    {
+      "epoch": 2.52,
+      "eval_bleu": 96.1002,
+      "eval_gen_len": 64.232,
+      "eval_loss": 0.2419024407863617,
+      "eval_rouge1": 94.0091,
+      "eval_rouge2": 90.4435,
+      "eval_rougeL": 93.8185,
+      "eval_rougeLsum": 93.8034,
+      "eval_runtime": 3103.5189,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 90000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.744075164773524e-06,
+      "loss": 0.1659,
+      "step": 90050
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.7300518861309775e-06,
+      "loss": 0.1377,
+      "step": 90100
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.716028607488431e-06,
+      "loss": 0.1398,
+      "step": 90150
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.702005328845884e-06,
+      "loss": 0.1551,
+      "step": 90200
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.687982050203338e-06,
+      "loss": 0.1371,
+      "step": 90250
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.673958771560791e-06,
+      "loss": 0.1332,
+      "step": 90300
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 4.659935492918244e-06,
+      "loss": 0.1401,
+      "step": 90350
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.645912214275698e-06,
+      "loss": 0.1191,
+      "step": 90400
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.631888935633151e-06,
+      "loss": 0.1421,
+      "step": 90450
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.617865656990604e-06,
+      "loss": 0.1362,
+      "step": 90500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.603842378348058e-06,
+      "loss": 0.1488,
+      "step": 90550
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.589819099705511e-06,
+      "loss": 0.1591,
+      "step": 90600
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.5757958210629645e-06,
+      "loss": 0.1331,
+      "step": 90650
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.561772542420418e-06,
+      "loss": 0.1337,
+      "step": 90700
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.547749263777871e-06,
+      "loss": 0.1487,
+      "step": 90750
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.5337259851353246e-06,
+      "loss": 0.1401,
+      "step": 90800
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.519702706492778e-06,
+      "loss": 0.1403,
+      "step": 90850
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.505679427850231e-06,
+      "loss": 0.141,
+      "step": 90900
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.491656149207685e-06,
+      "loss": 0.1422,
+      "step": 90950
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.477632870565138e-06,
+      "loss": 0.1415,
+      "step": 91000
+    },
+    {
+      "epoch": 2.55,
+      "eval_bleu": 96.1065,
+      "eval_gen_len": 64.2577,
+      "eval_loss": 0.24246443808078766,
+      "eval_rouge1": 94.0034,
+      "eval_rouge2": 90.453,
+      "eval_rougeL": 93.8071,
+      "eval_rougeLsum": 93.7875,
+      "eval_runtime": 3093.3898,
+      "eval_samples_per_second": 0.97,
+      "eval_steps_per_second": 0.242,
+      "step": 91000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 4.463609591922591e-06,
+      "loss": 0.1411,
+      "step": 91050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.449586313280045e-06,
+      "loss": 0.1617,
+      "step": 91100
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.435563034637498e-06,
+      "loss": 0.1657,
+      "step": 91150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.421539755994951e-06,
+      "loss": 0.1337,
+      "step": 91200
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.407516477352405e-06,
+      "loss": 0.1298,
+      "step": 91250
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.393493198709858e-06,
+      "loss": 0.1551,
+      "step": 91300
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.3794699200673115e-06,
+      "loss": 0.1551,
+      "step": 91350
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.365446641424765e-06,
+      "loss": 0.1484,
+      "step": 91400
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 4.351423362782219e-06,
+      "loss": 0.1345,
+      "step": 91450
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.3374000841396716e-06,
+      "loss": 0.1532,
+      "step": 91500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.323376805497126e-06,
+      "loss": 0.1451,
+      "step": 91550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.309353526854578e-06,
+      "loss": 0.1437,
+      "step": 91600
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.2953302482120325e-06,
+      "loss": 0.1572,
+      "step": 91650
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.281306969569485e-06,
+      "loss": 0.1351,
+      "step": 91700
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.267283690926939e-06,
+      "loss": 0.134,
+      "step": 91750
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.253260412284392e-06,
+      "loss": 0.1358,
+      "step": 91800
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.239237133641846e-06,
+      "loss": 0.138,
+      "step": 91850
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.225213854999298e-06,
+      "loss": 0.1377,
+      "step": 91900
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.211190576356753e-06,
+      "loss": 0.1473,
+      "step": 91950
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.197167297714205e-06,
+      "loss": 0.1288,
+      "step": 92000
+    },
+    {
+      "epoch": 2.58,
+      "eval_bleu": 96.1036,
+      "eval_gen_len": 64.2483,
+      "eval_loss": 0.24192480742931366,
+      "eval_rouge1": 93.9846,
+      "eval_rouge2": 90.4605,
+      "eval_rougeL": 93.8057,
+      "eval_rougeLsum": 93.7865,
+      "eval_runtime": 3083.7598,
+      "eval_samples_per_second": 0.973,
+      "eval_steps_per_second": 0.243,
+      "step": 92000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.183144019071659e-06,
+      "loss": 0.1448,
+      "step": 92050
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.169120740429112e-06,
+      "loss": 0.1436,
+      "step": 92100
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.155097461786566e-06,
+      "loss": 0.1567,
+      "step": 92150
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1410741831440186e-06,
+      "loss": 0.1441,
+      "step": 92200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.127050904501473e-06,
+      "loss": 0.1365,
+      "step": 92250
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.113027625858925e-06,
+      "loss": 0.1466,
+      "step": 92300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.0990043472163795e-06,
+      "loss": 0.1398,
+      "step": 92350
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.084981068573832e-06,
+      "loss": 0.1404,
+      "step": 92400
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.070957789931286e-06,
+      "loss": 0.149,
+      "step": 92450
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.056934511288739e-06,
+      "loss": 0.1322,
+      "step": 92500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.042911232646193e-06,
+      "loss": 0.1344,
+      "step": 92550
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.0288879540036454e-06,
+      "loss": 0.1471,
+      "step": 92600
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.0148646753611e-06,
+      "loss": 0.1535,
+      "step": 92650
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.000841396718552e-06,
+      "loss": 0.1403,
+      "step": 92700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.986818118076006e-06,
+      "loss": 0.1265,
+      "step": 92750
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.972794839433459e-06,
+      "loss": 0.1376,
+      "step": 92800
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.958771560790913e-06,
+      "loss": 0.1421,
+      "step": 92850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.944748282148366e-06,
+      "loss": 0.1379,
+      "step": 92900
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.93072500350582e-06,
+      "loss": 0.1493,
+      "step": 92950
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.916701724863273e-06,
+      "loss": 0.1392,
+      "step": 93000
+    },
+    {
+      "epoch": 2.61,
+      "eval_bleu": 96.1104,
+      "eval_gen_len": 64.2603,
+      "eval_loss": 0.24193215370178223,
+      "eval_rouge1": 94.0378,
+      "eval_rouge2": 90.4824,
+      "eval_rougeL": 93.8304,
+      "eval_rougeLsum": 93.8203,
+      "eval_runtime": 3103.7806,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 93000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.9026784462207265e-06,
+      "loss": 0.1626,
+      "step": 93050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.88865516757818e-06,
+      "loss": 0.1596,
+      "step": 93100
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.874631888935633e-06,
+      "loss": 0.1582,
+      "step": 93150
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.8606086102930866e-06,
+      "loss": 0.1369,
+      "step": 93200
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.84658533165054e-06,
+      "loss": 0.1522,
+      "step": 93250
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.832562053007993e-06,
+      "loss": 0.1698,
+      "step": 93300
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.818538774365447e-06,
+      "loss": 0.1352,
+      "step": 93350
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.8045154957228996e-06,
+      "loss": 0.1307,
+      "step": 93400
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7904922170803534e-06,
+      "loss": 0.1264,
+      "step": 93450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7764689384378063e-06,
+      "loss": 0.1512,
+      "step": 93500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.7624456597952605e-06,
+      "loss": 0.1284,
+      "step": 93550
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.748422381152714e-06,
+      "loss": 0.1413,
+      "step": 93600
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.734399102510167e-06,
+      "loss": 0.1374,
+      "step": 93650
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.7203758238676206e-06,
+      "loss": 0.144,
+      "step": 93700
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.706352545225074e-06,
+      "loss": 0.1633,
+      "step": 93750
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6923292665825273e-06,
+      "loss": 0.1542,
+      "step": 93800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6783059879399806e-06,
+      "loss": 0.1286,
+      "step": 93850
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.664282709297434e-06,
+      "loss": 0.1526,
+      "step": 93900
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 3.6502594306548874e-06,
+      "loss": 0.1431,
+      "step": 93950
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6362361520123407e-06,
+      "loss": 0.141,
+      "step": 94000
+    },
+    {
+      "epoch": 2.64,
+      "eval_bleu": 96.1146,
+      "eval_gen_len": 64.2413,
+      "eval_loss": 0.24184368550777435,
+      "eval_rouge1": 94.0178,
+      "eval_rouge2": 90.4805,
+      "eval_rougeL": 93.8232,
+      "eval_rougeLsum": 93.8125,
+      "eval_runtime": 3081.3777,
+      "eval_samples_per_second": 0.974,
+      "eval_steps_per_second": 0.243,
+      "step": 94000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.622212873369794e-06,
+      "loss": 0.1481,
+      "step": 94050
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.6081895947272474e-06,
+      "loss": 0.1359,
+      "step": 94100
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.5941663160847008e-06,
+      "loss": 0.1494,
+      "step": 94150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.580143037442154e-06,
+      "loss": 0.1372,
+      "step": 94200
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.5661197587996075e-06,
+      "loss": 0.1474,
+      "step": 94250
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.552096480157061e-06,
+      "loss": 0.1232,
+      "step": 94300
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5380732015145142e-06,
+      "loss": 0.1355,
+      "step": 94350
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.5240499228719676e-06,
+      "loss": 0.1243,
+      "step": 94400
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.510026644229421e-06,
+      "loss": 0.1414,
+      "step": 94450
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4960033655868743e-06,
+      "loss": 0.1344,
+      "step": 94500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4819800869443276e-06,
+      "loss": 0.1251,
+      "step": 94550
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.467956808301781e-06,
+      "loss": 0.1469,
+      "step": 94600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.4539335296592344e-06,
+      "loss": 0.1671,
+      "step": 94650
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4399102510166877e-06,
+      "loss": 0.1343,
+      "step": 94700
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.425886972374141e-06,
+      "loss": 0.1493,
+      "step": 94750
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.4118636937315944e-06,
+      "loss": 0.1354,
+      "step": 94800
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.397840415089048e-06,
+      "loss": 0.1397,
+      "step": 94850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.383817136446501e-06,
+      "loss": 0.1304,
+      "step": 94900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.3697938578039545e-06,
+      "loss": 0.1198,
+      "step": 94950
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.3557705791614083e-06,
+      "loss": 0.1372,
+      "step": 95000
+    },
+    {
+      "epoch": 2.66,
+      "eval_bleu": 96.1034,
+      "eval_gen_len": 64.2807,
+      "eval_loss": 0.24183472990989685,
+      "eval_rouge1": 93.9826,
+      "eval_rouge2": 90.409,
+      "eval_rougeL": 93.7782,
+      "eval_rougeLsum": 93.7723,
+      "eval_runtime": 3083.1749,
+      "eval_samples_per_second": 0.973,
+      "eval_steps_per_second": 0.243,
+      "step": 95000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3417473005188616e-06,
+      "loss": 0.1558,
+      "step": 95050
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.327724021876315e-06,
+      "loss": 0.1282,
+      "step": 95100
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.3137007432337684e-06,
+      "loss": 0.1595,
+      "step": 95150
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.2996774645912217e-06,
+      "loss": 0.1403,
+      "step": 95200
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.285654185948675e-06,
+      "loss": 0.1429,
+      "step": 95250
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.2716309073061284e-06,
+      "loss": 0.1364,
+      "step": 95300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.257607628663582e-06,
+      "loss": 0.1406,
+      "step": 95350
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.243584350021035e-06,
+      "loss": 0.1495,
+      "step": 95400
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2295610713784885e-06,
+      "loss": 0.1608,
+      "step": 95450
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.215537792735942e-06,
+      "loss": 0.1503,
+      "step": 95500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.2015145140933952e-06,
+      "loss": 0.1371,
+      "step": 95550
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.1874912354508486e-06,
+      "loss": 0.1465,
+      "step": 95600
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.173467956808302e-06,
+      "loss": 0.1321,
+      "step": 95650
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.1594446781657553e-06,
+      "loss": 0.1382,
+      "step": 95700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1454213995232087e-06,
+      "loss": 0.1441,
+      "step": 95750
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.131398120880662e-06,
+      "loss": 0.1558,
+      "step": 95800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1173748422381154e-06,
+      "loss": 0.1395,
+      "step": 95850
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.1033515635955687e-06,
+      "loss": 0.1308,
+      "step": 95900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.089328284953022e-06,
+      "loss": 0.1322,
+      "step": 95950
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.0753050063104754e-06,
+      "loss": 0.1299,
+      "step": 96000
+    },
+    {
+      "epoch": 2.69,
+      "eval_bleu": 96.1016,
+      "eval_gen_len": 64.2387,
+      "eval_loss": 0.24173256754875183,
+      "eval_rouge1": 93.9954,
+      "eval_rouge2": 90.4732,
+      "eval_rougeL": 93.8053,
+      "eval_rougeLsum": 93.7909,
+      "eval_runtime": 3082.3137,
+      "eval_samples_per_second": 0.973,
+      "eval_steps_per_second": 0.243,
+      "step": 96000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.061281727667929e-06,
+      "loss": 0.1424,
+      "step": 96050
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.047258449025382e-06,
+      "loss": 0.1332,
+      "step": 96100
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0332351703828355e-06,
+      "loss": 0.1298,
+      "step": 96150
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.019211891740289e-06,
+      "loss": 0.1383,
+      "step": 96200
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.0051886130977422e-06,
+      "loss": 0.1434,
+      "step": 96250
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.9911653344551956e-06,
+      "loss": 0.1406,
+      "step": 96300
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.977142055812649e-06,
+      "loss": 0.1343,
+      "step": 96350
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 2.9631187771701023e-06,
+      "loss": 0.1335,
+      "step": 96400
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.949095498527556e-06,
+      "loss": 0.1457,
+      "step": 96450
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.9350722198850094e-06,
+      "loss": 0.1458,
+      "step": 96500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.921048941242463e-06,
+      "loss": 0.1494,
+      "step": 96550
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.907025662599916e-06,
+      "loss": 0.1453,
+      "step": 96600
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8930023839573695e-06,
+      "loss": 0.1452,
+      "step": 96650
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.878979105314823e-06,
+      "loss": 0.1607,
+      "step": 96700
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8649558266722762e-06,
+      "loss": 0.149,
+      "step": 96750
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 2.8509325480297296e-06,
+      "loss": 0.1419,
+      "step": 96800
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.836909269387183e-06,
+      "loss": 0.1308,
+      "step": 96850
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8228859907446363e-06,
+      "loss": 0.143,
+      "step": 96900
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.8088627121020897e-06,
+      "loss": 0.134,
+      "step": 96950
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.794839433459543e-06,
+      "loss": 0.1375,
+      "step": 97000
+    },
+    {
+      "epoch": 2.72,
+      "eval_bleu": 96.1201,
+      "eval_gen_len": 64.2793,
+      "eval_loss": 0.24178442358970642,
+      "eval_rouge1": 93.9882,
+      "eval_rouge2": 90.4529,
+      "eval_rougeL": 93.7934,
+      "eval_rougeLsum": 93.7813,
+      "eval_runtime": 3072.8346,
+      "eval_samples_per_second": 0.976,
+      "eval_steps_per_second": 0.244,
+      "step": 97000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7808161548169964e-06,
+      "loss": 0.1406,
+      "step": 97050
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.7667928761744497e-06,
+      "loss": 0.1449,
+      "step": 97100
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 2.752769597531903e-06,
+      "loss": 0.137,
+      "step": 97150
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.7387463188893565e-06,
+      "loss": 0.1525,
+      "step": 97200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.72472304024681e-06,
+      "loss": 0.1456,
+      "step": 97250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.710699761604263e-06,
+      "loss": 0.1472,
+      "step": 97300
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.6966764829617165e-06,
+      "loss": 0.1447,
+      "step": 97350
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.68265320431917e-06,
+      "loss": 0.1359,
+      "step": 97400
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.6686299256766232e-06,
+      "loss": 0.1593,
+      "step": 97450
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 2.6546066470340766e-06,
+      "loss": 0.136,
+      "step": 97500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.64058336839153e-06,
+      "loss": 0.1496,
+      "step": 97550
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6265600897489833e-06,
+      "loss": 0.1401,
+      "step": 97600
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.6125368111064367e-06,
+      "loss": 0.1427,
+      "step": 97650
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.59851353246389e-06,
+      "loss": 0.1495,
+      "step": 97700
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5844902538213434e-06,
+      "loss": 0.1276,
+      "step": 97750
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5704669751787967e-06,
+      "loss": 0.1434,
+      "step": 97800
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 2.5564436965362505e-06,
+      "loss": 0.13,
+      "step": 97850
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.542420417893704e-06,
+      "loss": 0.1309,
+      "step": 97900
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5283971392511572e-06,
+      "loss": 0.1659,
+      "step": 97950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.5143738606086106e-06,
+      "loss": 0.1282,
+      "step": 98000
+    },
+    {
+      "epoch": 2.75,
+      "eval_bleu": 96.1244,
+      "eval_gen_len": 64.2753,
+      "eval_loss": 0.24166151881217957,
+      "eval_rouge1": 93.9974,
+      "eval_rouge2": 90.4493,
+      "eval_rougeL": 93.8008,
+      "eval_rougeLsum": 93.7908,
+      "eval_runtime": 3089.0272,
+      "eval_samples_per_second": 0.971,
+      "eval_steps_per_second": 0.243,
+      "step": 98000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.500350581966064e-06,
+      "loss": 0.1406,
+      "step": 98050
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4863273033235173e-06,
+      "loss": 0.1402,
+      "step": 98100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.4723040246809707e-06,
+      "loss": 0.1543,
+      "step": 98150
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 2.458280746038424e-06,
+      "loss": 0.1434,
+      "step": 98200
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4442574673958774e-06,
+      "loss": 0.1338,
+      "step": 98250
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4302341887533307e-06,
+      "loss": 0.1202,
+      "step": 98300
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.416210910110784e-06,
+      "loss": 0.1641,
+      "step": 98350
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.4021876314682375e-06,
+      "loss": 0.1365,
+      "step": 98400
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.388164352825691e-06,
+      "loss": 0.1366,
+      "step": 98450
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.374141074183144e-06,
+      "loss": 0.1485,
+      "step": 98500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 2.3601177955405975e-06,
+      "loss": 0.1366,
+      "step": 98550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.346094516898051e-06,
+      "loss": 0.1644,
+      "step": 98600
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3320712382555042e-06,
+      "loss": 0.1359,
+      "step": 98650
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.3180479596129576e-06,
+      "loss": 0.1452,
+      "step": 98700
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.304024680970411e-06,
+      "loss": 0.1391,
+      "step": 98750
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2900014023278643e-06,
+      "loss": 0.1375,
+      "step": 98800
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.2759781236853177e-06,
+      "loss": 0.1517,
+      "step": 98850
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.261954845042771e-06,
+      "loss": 0.149,
+      "step": 98900
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2479315664002244e-06,
+      "loss": 0.1239,
+      "step": 98950
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2339082877576778e-06,
+      "loss": 0.1374,
+      "step": 99000
+    },
+    {
+      "epoch": 2.78,
+      "eval_bleu": 96.1272,
+      "eval_gen_len": 64.262,
+      "eval_loss": 0.24160811305046082,
+      "eval_rouge1": 93.9828,
+      "eval_rouge2": 90.4481,
+      "eval_rougeL": 93.7941,
+      "eval_rougeLsum": 93.7877,
+      "eval_runtime": 3114.2761,
+      "eval_samples_per_second": 0.963,
+      "eval_steps_per_second": 0.241,
+      "step": 99000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.219885009115131e-06,
+      "loss": 0.1295,
+      "step": 99050
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.2058617304725845e-06,
+      "loss": 0.1544,
+      "step": 99100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.191838451830038e-06,
+      "loss": 0.1506,
+      "step": 99150
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.177815173187491e-06,
+      "loss": 0.1497,
+      "step": 99200
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.1637918945449445e-06,
+      "loss": 0.1591,
+      "step": 99250
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1497686159023983e-06,
+      "loss": 0.1467,
+      "step": 99300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1357453372598517e-06,
+      "loss": 0.131,
+      "step": 99350
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.121722058617305e-06,
+      "loss": 0.1436,
+      "step": 99400
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.1076987799747584e-06,
+      "loss": 0.1561,
+      "step": 99450
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.0936755013322118e-06,
+      "loss": 0.1469,
+      "step": 99500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.079652222689665e-06,
+      "loss": 0.1375,
+      "step": 99550
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.0656289440471185e-06,
+      "loss": 0.1539,
+      "step": 99600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.051605665404572e-06,
+      "loss": 0.1314,
+      "step": 99650
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.037582386762025e-06,
+      "loss": 0.1277,
+      "step": 99700
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.0235591081194785e-06,
+      "loss": 0.1627,
+      "step": 99750
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.009535829476932e-06,
+      "loss": 0.139,
+      "step": 99800
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9955125508343853e-06,
+      "loss": 0.137,
+      "step": 99850
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9814892721918386e-06,
+      "loss": 0.125,
+      "step": 99900
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.967465993549292e-06,
+      "loss": 0.1275,
+      "step": 99950
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.9534427149067453e-06,
+      "loss": 0.1327,
+      "step": 100000
+    },
+    {
+      "epoch": 2.8,
+      "eval_bleu": 96.1239,
+      "eval_gen_len": 64.253,
+      "eval_loss": 0.2416970431804657,
+      "eval_rouge1": 94.0048,
+      "eval_rouge2": 90.4634,
+      "eval_rougeL": 93.8114,
+      "eval_rougeLsum": 93.8025,
+      "eval_runtime": 3102.0331,
+      "eval_samples_per_second": 0.967,
+      "eval_steps_per_second": 0.242,
+      "step": 100000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9394194362641987e-06,
+      "loss": 0.1341,
+      "step": 100050
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.925396157621652e-06,
+      "loss": 0.13,
+      "step": 100100
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9113728789791054e-06,
+      "loss": 0.1335,
+      "step": 100150
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8973496003365588e-06,
+      "loss": 0.1611,
+      "step": 100200
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8833263216940121e-06,
+      "loss": 0.1226,
+      "step": 100250
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.8693030430514655e-06,
+      "loss": 0.1349,
+      "step": 100300
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.855279764408919e-06,
+      "loss": 0.1462,
+      "step": 100350
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8412564857663724e-06,
+      "loss": 0.1378,
+      "step": 100400
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8272332071238258e-06,
+      "loss": 0.1636,
+      "step": 100450
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8132099284812791e-06,
+      "loss": 0.1499,
+      "step": 100500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7991866498387325e-06,
+      "loss": 0.1489,
+      "step": 100550
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7851633711961858e-06,
+      "loss": 0.1582,
+      "step": 100600
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7711400925536392e-06,
+      "loss": 0.1305,
+      "step": 100650
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7571168139110925e-06,
+      "loss": 0.1596,
+      "step": 100700
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.743093535268546e-06,
+      "loss": 0.1275,
+      "step": 100750
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7290702566259993e-06,
+      "loss": 0.1398,
+      "step": 100800
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7150469779834526e-06,
+      "loss": 0.1348,
+      "step": 100850
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.701023699340906e-06,
+      "loss": 0.147,
+      "step": 100900
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6870004206983593e-06,
+      "loss": 0.1471,
+      "step": 100950
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6729771420558127e-06,
+      "loss": 0.1433,
+      "step": 101000
+    },
+    {
+      "epoch": 2.83,
+      "eval_bleu": 96.1231,
+      "eval_gen_len": 64.2653,
+      "eval_loss": 0.24147701263427734,
+      "eval_rouge1": 93.9973,
+      "eval_rouge2": 90.4389,
+      "eval_rougeL": 93.8007,
+      "eval_rougeLsum": 93.7953,
+      "eval_runtime": 3076.5825,
+      "eval_samples_per_second": 0.975,
+      "eval_steps_per_second": 0.244,
+      "step": 101000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6589538634132663e-06,
+      "loss": 0.1294,
+      "step": 101050
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6449305847707196e-06,
+      "loss": 0.1555,
+      "step": 101100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.630907306128173e-06,
+      "loss": 0.1633,
+      "step": 101150
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6168840274856263e-06,
+      "loss": 0.1383,
+      "step": 101200
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6028607488430797e-06,
+      "loss": 0.1397,
+      "step": 101250
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.588837470200533e-06,
+      "loss": 0.1473,
+      "step": 101300
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5748141915579864e-06,
+      "loss": 0.1379,
+      "step": 101350
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5607909129154398e-06,
+      "loss": 0.1373,
+      "step": 101400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5467676342728931e-06,
+      "loss": 0.1438,
+      "step": 101450
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5327443556303465e-06,
+      "loss": 0.146,
+      "step": 101500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5187210769877998e-06,
+      "loss": 0.1479,
+      "step": 101550
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.5046977983452532e-06,
+      "loss": 0.1447,
+      "step": 101600
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4906745197027066e-06,
+      "loss": 0.1461,
+      "step": 101650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.47665124106016e-06,
+      "loss": 0.1559,
+      "step": 101700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.4626279624176133e-06,
+      "loss": 0.162,
+      "step": 101750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4486046837750668e-06,
+      "loss": 0.1347,
+      "step": 101800
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4345814051325202e-06,
+      "loss": 0.1394,
+      "step": 101850
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.4205581264899736e-06,
+      "loss": 0.1429,
+      "step": 101900
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.406534847847427e-06,
+      "loss": 0.1298,
+      "step": 101950
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3925115692048803e-06,
+      "loss": 0.1688,
+      "step": 102000
+    },
+    {
+      "epoch": 2.86,
+      "eval_bleu": 96.1287,
+      "eval_gen_len": 64.2707,
+      "eval_loss": 0.24163706600666046,
+      "eval_rouge1": 94.0131,
+      "eval_rouge2": 90.4767,
+      "eval_rougeL": 93.8162,
+      "eval_rougeLsum": 93.8057,
+      "eval_runtime": 3096.7851,
+      "eval_samples_per_second": 0.969,
+      "eval_steps_per_second": 0.242,
+      "step": 102000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3784882905623336e-06,
+      "loss": 0.162,
+      "step": 102050
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.364465011919787e-06,
+      "loss": 0.1441,
+      "step": 102100
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.3504417332772403e-06,
+      "loss": 0.1396,
+      "step": 102150
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3364184546346937e-06,
+      "loss": 0.1423,
+      "step": 102200
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.322395175992147e-06,
+      "loss": 0.1324,
+      "step": 102250
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.3083718973496004e-06,
+      "loss": 0.16,
+      "step": 102300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.2943486187070538e-06,
+      "loss": 0.135,
+      "step": 102350
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.2803253400645071e-06,
+      "loss": 0.1298,
+      "step": 102400
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.2663020614219605e-06,
+      "loss": 0.1523,
+      "step": 102450
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.252278782779414e-06,
+      "loss": 0.1286,
+      "step": 102500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2382555041368674e-06,
+      "loss": 0.1439,
+      "step": 102550
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2242322254943208e-06,
+      "loss": 0.1283,
+      "step": 102600
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.2102089468517741e-06,
+      "loss": 0.1216,
+      "step": 102650
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1961856682092275e-06,
+      "loss": 0.1485,
+      "step": 102700
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1821623895666808e-06,
+      "loss": 0.1455,
+      "step": 102750
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1681391109241342e-06,
+      "loss": 0.1384,
+      "step": 102800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.1541158322815876e-06,
+      "loss": 0.14,
+      "step": 102850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1400925536390407e-06,
+      "loss": 0.1367,
+      "step": 102900
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.126069274996494e-06,
+      "loss": 0.1558,
+      "step": 102950
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.1120459963539474e-06,
+      "loss": 0.1377,
+      "step": 103000
+    },
+    {
+      "epoch": 2.89,
+      "eval_bleu": 96.1203,
+      "eval_gen_len": 64.258,
+      "eval_loss": 0.24138423800468445,
+      "eval_rouge1": 94.0261,
+      "eval_rouge2": 90.4778,
+      "eval_rougeL": 93.8237,
+      "eval_rougeLsum": 93.815,
+      "eval_runtime": 3149.2273,
+      "eval_samples_per_second": 0.953,
+      "eval_steps_per_second": 0.238,
+      "step": 103000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0980227177114008e-06,
+      "loss": 0.1283,
+      "step": 103050
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0839994390688541e-06,
+      "loss": 0.1487,
+      "step": 103100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.0699761604263075e-06,
+      "loss": 0.144,
+      "step": 103150
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.055952881783761e-06,
+      "loss": 0.1369,
+      "step": 103200
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0419296031412144e-06,
+      "loss": 0.1451,
+      "step": 103250
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0279063244986678e-06,
+      "loss": 0.1536,
+      "step": 103300
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.0138830458561211e-06,
+      "loss": 0.1434,
+      "step": 103350
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.998597672135745e-07,
+      "loss": 0.144,
+      "step": 103400
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.858364885710279e-07,
+      "loss": 0.142,
+      "step": 103450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.718132099284812e-07,
+      "loss": 0.1209,
+      "step": 103500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.577899312859346e-07,
+      "loss": 0.127,
+      "step": 103550
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.437666526433879e-07,
+      "loss": 0.1505,
+      "step": 103600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.297433740008414e-07,
+      "loss": 0.1377,
+      "step": 103650
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.157200953582947e-07,
+      "loss": 0.1458,
+      "step": 103700
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.016968167157482e-07,
+      "loss": 0.1445,
+      "step": 103750
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.876735380732016e-07,
+      "loss": 0.1363,
+      "step": 103800
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.736502594306549e-07,
+      "loss": 0.152,
+      "step": 103850
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.596269807881083e-07,
+      "loss": 0.1565,
+      "step": 103900
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.456037021455616e-07,
+      "loss": 0.142,
+      "step": 103950
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.31580423503015e-07,
+      "loss": 0.1482,
+      "step": 104000
+    },
+    {
+      "epoch": 2.92,
+      "eval_bleu": 96.1208,
+      "eval_gen_len": 64.2503,
+      "eval_loss": 0.24138183891773224,
+      "eval_rouge1": 94.0225,
+      "eval_rouge2": 90.488,
+      "eval_rougeL": 93.8289,
+      "eval_rougeLsum": 93.8131,
+      "eval_runtime": 3071.8283,
+      "eval_samples_per_second": 0.977,
+      "eval_steps_per_second": 0.244,
+      "step": 104000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.175571448604684e-07,
+      "loss": 0.1504,
+      "step": 104050
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.035338662179218e-07,
+      "loss": 0.1263,
+      "step": 104100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.895105875753752e-07,
+      "loss": 0.1346,
+      "step": 104150
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.754873089328285e-07,
+      "loss": 0.1471,
+      "step": 104200
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.614640302902819e-07,
+      "loss": 0.1356,
+      "step": 104250
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.474407516477353e-07,
+      "loss": 0.1457,
+      "step": 104300
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.334174730051886e-07,
+      "loss": 0.1458,
+      "step": 104350
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.19394194362642e-07,
+      "loss": 0.1373,
+      "step": 104400
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.053709157200954e-07,
+      "loss": 0.1362,
+      "step": 104450
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.913476370775488e-07,
+      "loss": 0.1548,
+      "step": 104500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.773243584350021e-07,
+      "loss": 0.1398,
+      "step": 104550
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 6.633010797924555e-07,
+      "loss": 0.1604,
+      "step": 104600
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.492778011499089e-07,
+      "loss": 0.146,
+      "step": 104650
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.352545225073622e-07,
+      "loss": 0.1272,
+      "step": 104700
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.212312438648156e-07,
+      "loss": 0.1495,
+      "step": 104750
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 6.07207965222269e-07,
+      "loss": 0.1408,
+      "step": 104800
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.931846865797224e-07,
+      "loss": 0.1502,
+      "step": 104850
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.791614079371758e-07,
+      "loss": 0.1471,
+      "step": 104900
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.651381292946291e-07,
+      "loss": 0.16,
+      "step": 104950
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.511148506520825e-07,
+      "loss": 0.1286,
+      "step": 105000
+    },
+    {
+      "epoch": 2.94,
+      "eval_bleu": 96.128,
+      "eval_gen_len": 64.2513,
+      "eval_loss": 0.24144940078258514,
+      "eval_rouge1": 94.0054,
+      "eval_rouge2": 90.4653,
+      "eval_rougeL": 93.8155,
+      "eval_rougeLsum": 93.8029,
+      "eval_runtime": 3085.2312,
+      "eval_samples_per_second": 0.972,
+      "eval_steps_per_second": 0.243,
+      "step": 105000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.370915720095358e-07,
+      "loss": 0.1509,
+      "step": 105050
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.230682933669892e-07,
+      "loss": 0.1342,
+      "step": 105100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.090450147244427e-07,
+      "loss": 0.144,
+      "step": 105150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.95021736081896e-07,
+      "loss": 0.166,
+      "step": 105200
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.809984574393494e-07,
+      "loss": 0.1405,
+      "step": 105250
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.669751787968027e-07,
+      "loss": 0.1551,
+      "step": 105300
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.529519001542561e-07,
+      "loss": 0.1417,
+      "step": 105350
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.3892862151170944e-07,
+      "loss": 0.1233,
+      "step": 105400
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.2490534286916285e-07,
+      "loss": 0.1406,
+      "step": 105450
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.108820642266162e-07,
+      "loss": 0.1347,
+      "step": 105500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.9685878558406956e-07,
+      "loss": 0.1533,
+      "step": 105550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.82835506941523e-07,
+      "loss": 0.1462,
+      "step": 105600
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.6881222829897633e-07,
+      "loss": 0.1404,
+      "step": 105650
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.547889496564297e-07,
+      "loss": 0.1323,
+      "step": 105700
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4076567101388305e-07,
+      "loss": 0.1564,
+      "step": 105750
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2674239237133646e-07,
+      "loss": 0.1508,
+      "step": 105800
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.127191137287898e-07,
+      "loss": 0.1467,
+      "step": 105850
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.986958350862432e-07,
+      "loss": 0.1574,
+      "step": 105900
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.846725564436966e-07,
+      "loss": 0.132,
+      "step": 105950
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.7064927780114994e-07,
+      "loss": 0.1392,
+      "step": 106000
+    },
+    {
+      "epoch": 2.97,
+      "eval_bleu": 96.1228,
+      "eval_gen_len": 64.2473,
+      "eval_loss": 0.2414444237947464,
+      "eval_rouge1": 94.0048,
+      "eval_rouge2": 90.4742,
+      "eval_rougeL": 93.8171,
+      "eval_rougeLsum": 93.8047,
+      "eval_runtime": 3077.7276,
+      "eval_samples_per_second": 0.975,
+      "eval_steps_per_second": 0.244,
+      "step": 106000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 2.566259991586033e-07,
+      "loss": 0.1544,
+      "step": 106050
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4260272051605666e-07,
+      "loss": 0.1498,
+      "step": 106100
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2857944187351002e-07,
+      "loss": 0.1291,
+      "step": 106150
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.145561632309634e-07,
+      "loss": 0.1411,
+      "step": 106200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0053288458841676e-07,
+      "loss": 0.1445,
+      "step": 106250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8650960594587014e-07,
+      "loss": 0.1499,
+      "step": 106300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7248632730332353e-07,
+      "loss": 0.1442,
+      "step": 106350
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5846304866077688e-07,
+      "loss": 0.1387,
+      "step": 106400
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4443977001823027e-07,
+      "loss": 0.138,
+      "step": 106450
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3041649137568363e-07,
+      "loss": 0.1559,
+      "step": 106500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1639321273313701e-07,
+      "loss": 0.1437,
+      "step": 106550
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.023699340905904e-07,
+      "loss": 0.1597,
+      "step": 106600
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.834665544804376e-08,
+      "loss": 0.1455,
+      "step": 106650
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.432337680549714e-08,
+      "loss": 0.1505,
+      "step": 106700
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.030009816295049e-08,
+      "loss": 0.1633,
+      "step": 106750
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.627681952040387e-08,
+      "loss": 0.135,
+      "step": 106800
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.225354087785724e-08,
+      "loss": 0.1483,
+      "step": 106850
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.8230262235310616e-08,
+      "loss": 0.1375,
+      "step": 106900
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.2069835927639876e-09,
+      "loss": 0.1489,
+      "step": 106950
+    },
+    {
+      "epoch": 3.0,
+      "step": 106965,
+      "total_flos": 7.416839339283579e+17,
+      "train_loss": 0.1832969891809344,
+      "train_runtime": 592063.0002,
+      "train_samples_per_second": 2.891,
+      "train_steps_per_second": 0.181
+    }
+  ],
+  "max_steps": 106965,
+  "num_train_epochs": 3,
+  "total_flos": 7.416839339283579e+17,
+  "trial_name": null,
+  "trial_params": null
+}