Training in progress, step 36, checkpoint

last-checkpoint/adapter_config.json

@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "q_proj",
-    "k_proj",
-    "v_proj",
     "up_proj",
+    "q_proj",
     "o_proj",
-    "down_proj",
-    "gate_proj"
+    "v_proj",
+    "gate_proj",
+    "k_proj",
+    "down_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c82944de0f083bae7d1682aba5cafb5573edb82f20e5eb1167f944265cd7f8e9
+oid sha256:cb5936b6074418059dc90977eb81043380c899f8c6835183721582101660a197
 size 83945296

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:652a7e6942ccec3253e1bef8b6f5941e88be96212599cb6956632b6f8bba0d9f
-size 42545748
+oid sha256:cd753746296b9024e407e8f0496372a9d49e89a57dfa709feed7f6b33fd398fe
+size 43706258

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:98ad479ae233e51f82f95d3a34340242e257af521ca087273075fd928399edb7
+oid sha256:b19b087c893f2cd0218cc55f24b97983f317f14527df7fe9f53cbfd09be0b1f2
 size 1064

last-checkpoint/trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.5708824466390571,
+  "epoch": 0.733991717107359,
   "eval_steps": 500,
-  "global_step": 28,
+  "global_step": 36,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -511,6 +511,150 @@
       "rewards/margins": 0.019869212061166763,
       "rewards/rejected": -0.09125552326440811,
       "step": 28
+    },
+    {
+      "epoch": 0.5912711054475948,
+      "grad_norm": 1.6056082248687744,
+      "learning_rate": 5.128205128205128e-05,
+      "log_odds_chosen": 0.5487304925918579,
+      "log_odds_ratio": -0.478320449590683,
+      "logits/chosen": -2.2418763637542725,
+      "logits/rejected": -2.2638566493988037,
+      "logps/chosen": -0.7001346349716187,
+      "logps/rejected": -1.0050172805786133,
+      "loss": 0.9538,
+      "nll_loss": 0.9059388041496277,
+      "rewards/accuracies": 0.921875,
+      "rewards/chosen": -0.07001346349716187,
+      "rewards/margins": 0.030488261952996254,
+      "rewards/rejected": -0.10050173103809357,
+      "step": 29
+    },
+    {
+      "epoch": 0.6116597642561326,
+      "grad_norm": 2.707125663757324,
+      "learning_rate": 4.871794871794872e-05,
+      "log_odds_chosen": 0.433153361082077,
+      "log_odds_ratio": -0.5276967883110046,
+      "logits/chosen": -2.2189157009124756,
+      "logits/rejected": -2.228149890899658,
+      "logps/chosen": -0.6423375606536865,
+      "logps/rejected": -0.8684220910072327,
+      "loss": 0.943,
+      "nll_loss": 0.89023357629776,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": -0.06423375755548477,
+      "rewards/margins": 0.022608455270528793,
+      "rewards/rejected": -0.08684220910072327,
+      "step": 30
+    },
+    {
+      "epoch": 0.6320484230646702,
+      "grad_norm": 3.1514434814453125,
+      "learning_rate": 4.615384615384616e-05,
+      "log_odds_chosen": 0.3669428527355194,
+      "log_odds_ratio": -0.5465207695960999,
+      "logits/chosen": -2.243276357650757,
+      "logits/rejected": -2.231745719909668,
+      "logps/chosen": -0.7233006954193115,
+      "logps/rejected": -0.9113064408302307,
+      "loss": 1.0016,
+      "nll_loss": 0.9469515681266785,
+      "rewards/accuracies": 0.8125,
+      "rewards/chosen": -0.07233007252216339,
+      "rewards/margins": 0.018800577148795128,
+      "rewards/rejected": -0.09113065153360367,
+      "step": 31
+    },
+    {
+      "epoch": 0.652437081873208,
+      "grad_norm": 2.5206167697906494,
+      "learning_rate": 4.358974358974359e-05,
+      "log_odds_chosen": 0.5872430205345154,
+      "log_odds_ratio": -0.46654394268989563,
+      "logits/chosen": -2.2469942569732666,
+      "logits/rejected": -2.269362211227417,
+      "logps/chosen": -0.6239035725593567,
+      "logps/rejected": -0.9376953840255737,
+      "loss": 0.9061,
+      "nll_loss": 0.8594872951507568,
+      "rewards/accuracies": 0.875,
+      "rewards/chosen": -0.06239035725593567,
+      "rewards/margins": 0.03137918934226036,
+      "rewards/rejected": -0.09376954287290573,
+      "step": 32
+    },
+    {
+      "epoch": 0.6728257406817458,
+      "grad_norm": 1.3284610509872437,
+      "learning_rate": 4.1025641025641023e-05,
+      "log_odds_chosen": 0.5279171466827393,
+      "log_odds_ratio": -0.49398207664489746,
+      "logits/chosen": -2.263728618621826,
+      "logits/rejected": -2.288696050643921,
+      "logps/chosen": -0.6505022644996643,
+      "logps/rejected": -0.9282124042510986,
+      "loss": 0.9288,
+      "nll_loss": 0.8793907761573792,
+      "rewards/accuracies": 0.890625,
+      "rewards/chosen": -0.06505022943019867,
+      "rewards/margins": 0.027771014720201492,
+      "rewards/rejected": -0.09282123297452927,
+      "step": 33
+    },
+    {
+      "epoch": 0.6932143994902835,
+      "grad_norm": 1.2634799480438232,
+      "learning_rate": 3.846153846153846e-05,
+      "log_odds_chosen": 0.5196709036827087,
+      "log_odds_ratio": -0.5028817653656006,
+      "logits/chosen": -2.3093760013580322,
+      "logits/rejected": -2.3152754306793213,
+      "logps/chosen": -0.6717790365219116,
+      "logps/rejected": -0.9456109404563904,
+      "loss": 0.9413,
+      "nll_loss": 0.8909698724746704,
+      "rewards/accuracies": 0.796875,
+      "rewards/chosen": -0.0671778991818428,
+      "rewards/margins": 0.027383197098970413,
+      "rewards/rejected": -0.09456109255552292,
+      "step": 34
+    },
+    {
+      "epoch": 0.7136030582988213,
+      "grad_norm": 1.2822017669677734,
+      "learning_rate": 3.58974358974359e-05,
+      "log_odds_chosen": 0.44004446268081665,
+      "log_odds_ratio": -0.53327876329422,
+      "logits/chosen": -2.3234097957611084,
+      "logits/rejected": -2.319546937942505,
+      "logps/chosen": -0.7039386630058289,
+      "logps/rejected": -0.9365054965019226,
+      "loss": 0.9674,
+      "nll_loss": 0.9140487909317017,
+      "rewards/accuracies": 0.796875,
+      "rewards/chosen": -0.070393867790699,
+      "rewards/margins": 0.023256685584783554,
+      "rewards/rejected": -0.09365054965019226,
+      "step": 35
+    },
+    {
+      "epoch": 0.733991717107359,
+      "grad_norm": 1.3868314027786255,
+      "learning_rate": 3.3333333333333335e-05,
+      "log_odds_chosen": 0.40167126059532166,
+      "log_odds_ratio": -0.5467191338539124,
+      "logits/chosen": -2.3325843811035156,
+      "logits/rejected": -2.3239898681640625,
+      "logps/chosen": -0.7655161619186401,
+      "logps/rejected": -0.9702532291412354,
+      "loss": 0.9815,
+      "nll_loss": 0.9267975091934204,
+      "rewards/accuracies": 0.765625,
+      "rewards/chosen": -0.07655161619186401,
+      "rewards/margins": 0.02047371119260788,
+      "rewards/rejected": -0.0970253199338913,
+      "step": 36
     }
   ],
   "logging_steps": 1,

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1e645a6afd0fb5de26677bef59ac2607c945856fb640ae342a9a3b8c2695f5c7
+oid sha256:c6bcc3f11131b07d173fb79ad8c31d1a50ed3843eadae9ec2e4213021d99bed1
 size 5560