Training in progress, step 48, checkpoint

last-checkpoint/adapter_config.json

@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "up_proj",
+    "down_proj",
     "q_proj",
     "o_proj",
-    "v_proj",
     "gate_proj",
     "k_proj",
-    "down_proj"
+    "v_proj",
+    "up_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 sha256:349c41609a51e6cfc6896a2a5d1a67153c68f9e2b7976427d560cec1ef2695d0
+oid sha256:9a587fe88328b0354f39625a46af633b6991ded003788a9b7ab0807b31298fe7
 size 83945296

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:66a5843d74545eb236796350e0ece8c200e91256593c92b31b5d53bd3686a8f4
+oid sha256:f770bb48ac6db34e078e1840377dec0256cb1b52bd176163f8b0b9c85ad8dc76
 size 43706258

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5f6daa0fed03c921127a64c255f462e66919bc76a6e6ba1e9093cf71d1068c6f
+oid sha256:c2fb311d43c732630c9c38144204d369cab84f4088b1165920818e91a4205052
 size 1064

last-checkpoint/trainer_state.json

@@ -1,9 +1,9 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.897100987575661,
+  "epoch": 0.978655622809812,
   "eval_steps": 500,
-  "global_step": 44,
+  "global_step": 48,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -799,6 +799,78 @@
       "rewards/margins": 0.0255814790725708,
       "rewards/rejected": -0.08916367590427399,
       "step": 44
+    },
+    {
+      "epoch": 0.9174896463841988,
+      "grad_norm": 1.5401713848114014,
+      "learning_rate": 1.0256410256410256e-05,
+      "log_odds_chosen": 0.5436590313911438,
+      "log_odds_ratio": -0.49072104692459106,
+      "logits/chosen": -2.357189655303955,
+      "logits/rejected": -2.368823528289795,
+      "logps/chosen": -0.6493522524833679,
+      "logps/rejected": -0.9346315860748291,
+      "loss": 0.9199,
+      "nll_loss": 0.8708474040031433,
+      "rewards/accuracies": 0.859375,
+      "rewards/chosen": -0.06493522971868515,
+      "rewards/margins": 0.02852793037891388,
+      "rewards/rejected": -0.09346316009759903,
+      "step": 45
+    },
+    {
+      "epoch": 0.9378783051927365,
+      "grad_norm": 1.42889404296875,
+      "learning_rate": 7.692307692307694e-06,
+      "log_odds_chosen": 0.49409711360931396,
+      "log_odds_ratio": -0.506333589553833,
+      "logits/chosen": -2.390031576156616,
+      "logits/rejected": -2.3763091564178467,
+      "logps/chosen": -0.6777428984642029,
+      "logps/rejected": -0.9434182643890381,
+      "loss": 0.9246,
+      "nll_loss": 0.8739202618598938,
+      "rewards/accuracies": 0.828125,
+      "rewards/chosen": -0.06777428835630417,
+      "rewards/margins": 0.02656753547489643,
+      "rewards/rejected": -0.09434183686971664,
+      "step": 46
+    },
+    {
+      "epoch": 0.9582669640012743,
+      "grad_norm": 1.4630545377731323,
+      "learning_rate": 5.128205128205128e-06,
+      "log_odds_chosen": 0.5089167356491089,
+      "log_odds_ratio": -0.49921250343322754,
+      "logits/chosen": -2.377633810043335,
+      "logits/rejected": -2.36307954788208,
+      "logps/chosen": -0.6796414256095886,
+      "logps/rejected": -0.9439955353736877,
+      "loss": 0.969,
+      "nll_loss": 0.9190601110458374,
+      "rewards/accuracies": 0.828125,
+      "rewards/chosen": -0.06796413660049438,
+      "rewards/margins": 0.02643541246652603,
+      "rewards/rejected": -0.09439954906702042,
+      "step": 47
+    },
+    {
+      "epoch": 0.978655622809812,
+      "grad_norm": 1.4280872344970703,
+      "learning_rate": 2.564102564102564e-06,
+      "log_odds_chosen": 0.5206915736198425,
+      "log_odds_ratio": -0.4926137626171112,
+      "logits/chosen": -2.381681442260742,
+      "logits/rejected": -2.3803200721740723,
+      "logps/chosen": -0.6376308798789978,
+      "logps/rejected": -0.9052294492721558,
+      "loss": 0.894,
+      "nll_loss": 0.8447284698486328,
+      "rewards/accuracies": 0.890625,
+      "rewards/chosen": -0.0637630894780159,
+      "rewards/margins": 0.026759855449199677,
+      "rewards/rejected": -0.09052293747663498,
+      "step": 48
     }
   ],
   "logging_steps": 1,

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c6bcc3f11131b07d173fb79ad8c31d1a50ed3843eadae9ec2e4213021d99bed1
+oid sha256:128f0cf3da8b1abd86530282a4bc926272323030313ebb6ed3dbec18e05701bf
 size 5560