End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 0.27013154824574787,
-    "train_runtime": 1786.8594,
-    "train_samples": 23,
-    "train_samples_per_second": 0.013,
-    "train_steps_per_second": 0.002
+    "train_loss": 0.06536803146203359,
+    "train_runtime": 3637.9329,
+    "train_samples": 30,
+    "train_samples_per_second": 0.007,
+    "train_steps_per_second": 0.001
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:042df9bab814993de332623fa5c40266568096ef8b903104ca3c676525f23d13
+oid sha256:77c40427fbbd03f4aa28c342daabd9193951114485d54cc99709a32af1c7b81a
 size 1976163472

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 0.27013154824574787,
-    "train_runtime": 1786.8594,
-    "train_samples": 23,
-    "train_samples_per_second": 0.013,
-    "train_steps_per_second": 0.002
+    "train_loss": 0.06536803146203359,
+    "train_runtime": 3637.9329,
+    "train_samples": 30,
+    "train_samples_per_second": 0.007,
+    "train_steps_per_second": 0.001
 }

trainer_state.json

@@ -2,7 +2,7 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.5217391304347826,
+  "epoch": 0.4,
   "eval_steps": 500,
   "global_step": 3,
   "is_hyper_param_search": false,
@@ -15,37 +15,37 @@
       "clip_ratio/low_mean": 0.0,
       "clip_ratio/low_min": 0.0,
       "clip_ratio/region_mean": 0.0,
-      "completions/clipped_ratio": 0.125,
-      "completions/max_length": 759.0,
-      "completions/max_terminated_length": 544.0,
-      "completions/mean_length": 344.3125,
-      "completions/mean_terminated_length": 246.29166412353516,
-      "completions/min_length": 64.5,
-      "completions/min_terminated_length": 64.5,
-      "epoch": 0.34782608695652173,
-      "grad_norm": 6.1356120109558105,
+      "completions/clipped_ratio": 0.0,
+      "completions/max_length": 605.5,
+      "completions/max_terminated_length": 605.5,
+      "completions/mean_length": 214.5625,
+      "completions/mean_terminated_length": 214.5625,
+      "completions/min_length": 9.0,
+      "completions/min_terminated_length": 9.0,
+      "epoch": 0.26666666666666666,
+      "grad_norm": 14.364421844482422,
       "kl": 0.0,
       "learning_rate": 5e-07,
-      "loss": 0.1231,
-      "num_tokens": 9605.0,
-      "reward": 0.0872948095202446,
-      "reward_std": 0.036231483216397464,
+      "loss": -0.0457,
+      "num_tokens": 7529.0,
+      "reward": 0.27101069688796997,
+      "reward_std": 0.3605286180973053,
       "rewards/concensus_correctness_reward_func/mean": 0.0,
       "rewards/concensus_correctness_reward_func/std": 0.0,
-      "rewards/consensus_reward_func/mean": 0.0,
-      "rewards/consensus_reward_func/std": 0.0,
+      "rewards/consensus_reward_func/mean": 0.125,
+      "rewards/consensus_reward_func/std": 0.3535533845424652,
       "rewards/cumulative_reward_2/mean": 0.0,
       "rewards/cumulative_reward_2/std": 0.0,
-      "rewards/final_correctness_reward_func/mean": 0.0,
-      "rewards/final_correctness_reward_func/std": 0.0,
-      "rewards/question_recreation_reward_func/mean": 0.0872948169708252,
-      "rewards/question_recreation_reward_func/std": 0.042022028006613255,
+      "rewards/final_correctness_reward_func/mean": 0.125,
+      "rewards/final_correctness_reward_func/std": 0.3535533845424652,
+      "rewards/question_recreation_reward_func/mean": 0.02001071721315384,
+      "rewards/question_recreation_reward_func/std": 0.009133741725236177,
       "rewards/soft_format_reward_func/mean": 0.0,
       "rewards/soft_format_reward_func/std": 0.0,
       "rewards/strict_format_reward_func/mean": 0.0,
       "rewards/strict_format_reward_func/std": 0.0,
-      "rewards/xmlcount_reward_func/mean": 0.0,
-      "rewards/xmlcount_reward_func/std": 0.0,
+      "rewards/xmlcount_reward_func/mean": 0.0010000000474974513,
+      "rewards/xmlcount_reward_func/std": 0.002828427357599139,
       "step": 2
     },
     {
@@ -55,27 +55,27 @@
       "clip_ratio/low_min": 0.0,
       "clip_ratio/region_mean": 0.0,
       "completions/clipped_ratio": 0.0,
-      "completions/max_length": 369.0,
-      "completions/max_terminated_length": 369.0,
-      "completions/mean_length": 126.25,
-      "completions/mean_terminated_length": 126.25,
-      "completions/min_length": 2.0,
-      "completions/min_terminated_length": 2.0,
-      "epoch": 0.5217391304347826,
-      "kl": 0.0009940905874827877,
-      "num_tokens": 12663.0,
-      "reward": 2.516864776611328,
-      "reward_std": 3.5382814407348633,
-      "rewards/concensus_correctness_reward_func/mean": 2.5,
-      "rewards/concensus_correctness_reward_func/std": 7.071068286895752,
+      "completions/max_length": 321.0,
+      "completions/max_terminated_length": 321.0,
+      "completions/mean_length": 167.625,
+      "completions/mean_terminated_length": 167.625,
+      "completions/min_length": 19.0,
+      "completions/min_terminated_length": 19.0,
+      "epoch": 0.4,
+      "kl": 0.0005115475723869167,
+      "num_tokens": 10918.0,
+      "reward": 0.16452272236347198,
+      "reward_std": 0.054086603224277496,
+      "rewards/concensus_correctness_reward_func/mean": 0.0,
+      "rewards/concensus_correctness_reward_func/std": 0.0,
       "rewards/consensus_reward_func/mean": 0.0,
       "rewards/consensus_reward_func/std": 0.0,
       "rewards/cumulative_reward_2/mean": 0.0,
       "rewards/cumulative_reward_2/std": 0.0,
       "rewards/final_correctness_reward_func/mean": 0.0,
       "rewards/final_correctness_reward_func/std": 0.0,
-      "rewards/question_recreation_reward_func/mean": 0.01686476171016693,
-      "rewards/question_recreation_reward_func/std": 0.01196382101625204,
+      "rewards/question_recreation_reward_func/mean": 0.16452273726463318,
+      "rewards/question_recreation_reward_func/std": 0.05140342935919762,
       "rewards/soft_format_reward_func/mean": 0.0,
       "rewards/soft_format_reward_func/std": 0.0,
       "rewards/strict_format_reward_func/mean": 0.0,
@@ -84,15 +84,15 @@
       "rewards/xmlcount_reward_func/std": 0.0,
       "step": 3,
       "total_flos": 0.0,
-      "train_loss": 0.27013154824574787,
-      "train_runtime": 1786.8594,
-      "train_samples_per_second": 0.013,
-      "train_steps_per_second": 0.002
+      "train_loss": 0.06536803146203359,
+      "train_runtime": 3637.9329,
+      "train_samples_per_second": 0.007,
+      "train_steps_per_second": 0.001
     }
   ],
   "logging_steps": 2,
   "max_steps": 3,
-  "num_input_tokens_seen": 12663,
+  "num_input_tokens_seen": 10918,
   "num_train_epochs": 1,
   "save_steps": 25,
   "stateful_callbacks": {

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6a71230f32e732eea5fcb1a0593b7bc74ff3864c9b7b760cda833ba20848a0ed
+oid sha256:0df0f354d248a121ff0bc1816ff97a871a1a1dfd3c63ddcf92aa745670c031c8
 size 6929