End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": -3.1709017093817236e-06,
-    "train_runtime": 590.6267,
+    "train_loss": 6.0011893765477e-06,
+    "train_runtime": 503.4756,
     "train_samples": 5,
-    "train_samples_per_second": 0.068,
-    "train_steps_per_second": 0.017
+    "train_samples_per_second": 0.079,
+    "train_steps_per_second": 0.02
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:49ca111340ed7c945725702a2578128d3a9247131a8c1fab02ed7fed9eb67b1f
+oid sha256:181eba8ad9555bd9e4d1b07177e72a31c44e53743dfae606b9b7876752d6de01
 size 1976163472

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": -3.1709017093817236e-06,
-    "train_runtime": 590.6267,
+    "train_loss": 6.0011893765477e-06,
+    "train_runtime": 503.4756,
     "train_samples": 5,
-    "train_samples_per_second": 0.068,
-    "train_steps_per_second": 0.017
+    "train_samples_per_second": 0.079,
+    "train_steps_per_second": 0.02
 }

trainer_state.json

@@ -10,95 +10,95 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "completion_length": 78.375,
+      "completion_length": 83.0,
       "epoch": 0.8,
-      "grad_norm": 5.627739906311035,
+      "grad_norm": 7.9807281494140625,
       "kl": 0.0,
       "learning_rate": 5e-07,
-      "loss": 0.0,
-      "reward": 0.7521542385220528,
-      "reward_std": 0.044539074537169654,
+      "loss": -0.0,
+      "reward": 1.7760924957692623,
+      "reward_std": 0.022786720073781908,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.5,
-      "rewards/question_recreation_reward_func": 0.15840421058237553,
+      "rewards/final_correctness_reward_func": 1.5,
+      "rewards/question_recreation_reward_func": 0.1510925330221653,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.09375,
+      "rewards/xmlcount_reward_func": 0.125,
       "step": 2
     },
     {
-      "completion_length": 49.666666666666664,
+      "completion_length": 56.833333333333336,
       "epoch": 1.4,
-      "grad_norm": 0.005242869723588228,
-      "kl": 0.016141654608266737,
+      "grad_norm": 10.24837589263916,
+      "kl": 0.0015244192351625923,
       "learning_rate": 4.415111107797445e-07,
       "loss": 0.0,
-      "reward": 1.9181493918100994,
-      "reward_std": 0.015638279418150585,
+      "reward": 2.550933281580607,
+      "reward_std": 0.07004699483513832,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 1.3333333333333333,
-      "rewards/question_recreation_reward_func": 0.4598161180814107,
+      "rewards/final_correctness_reward_func": 2.0,
+      "rewards/question_recreation_reward_func": 0.42593332131703693,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
       "rewards/xmlcount_reward_func": 0.125,
       "step": 4
     },
     {
-      "completion_length": 58.666666666666664,
+      "completion_length": 77.66666666666667,
       "epoch": 2.0,
-      "grad_norm": 7.197663307189941,
-      "kl": 0.010560240286091963,
+      "grad_norm": 6.0542120933532715,
+      "kl": 0.0016350069102675964,
       "learning_rate": 2.934120444167326e-07,
       "loss": 0.0,
-      "reward": 0.9169059892495474,
-      "reward_std": 0.0018884773987034957,
+      "reward": 1.5878510723511379,
+      "reward_std": 0.0029388506275912127,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.6666666666666666,
-      "rewards/question_recreation_reward_func": 0.16690599421660104,
+      "rewards/final_correctness_reward_func": 1.3333333333333333,
+      "rewards/question_recreation_reward_func": 0.1295176992813746,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.08333333333333333,
+      "rewards/xmlcount_reward_func": 0.125,
       "step": 6
     },
     {
-      "completion_length": 61.375,
+      "completion_length": 72.25,
       "epoch": 2.8,
-      "grad_norm": 2.7515034675598145,
-      "kl": 0.004465199788683094,
+      "grad_norm": 2.34584641456604,
+      "kl": 0.0029444374376907945,
       "learning_rate": 1.2500000000000005e-07,
-      "loss": 0.0,
-      "reward": 1.361022423952818,
-      "reward_std": 0.00136809196555987,
+      "loss": -0.0,
+      "reward": 1.8496044538915157,
+      "reward_std": 0.0011594545212574303,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 1.0,
-      "rewards/question_recreation_reward_func": 0.26727242209017277,
+      "rewards/final_correctness_reward_func": 1.5,
+      "rewards/question_recreation_reward_func": 0.22460449486970901,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.09375,
+      "rewards/xmlcount_reward_func": 0.125,
       "step": 8
     },
     {
-      "completion_length": 70.0,
+      "completion_length": 75.83333333333333,
       "epoch": 3.4,
-      "grad_norm": 6.803511619567871,
-      "kl": 0.01901205163449049,
+      "grad_norm": 5.32132625579834,
+      "kl": 0.007413267778853576,
       "learning_rate": 1.507684480352292e-08,
-      "loss": -0.0,
-      "reward": 0.9308561086654663,
-      "reward_std": 0.005250229577844341,
+      "loss": 0.0,
+      "reward": 1.6325677633285522,
+      "reward_std": 0.042186157782756105,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.6666666666666666,
-      "rewards/question_recreation_reward_func": 0.1391894891858101,
+      "rewards/final_correctness_reward_func": 1.3333333333333333,
+      "rewards/question_recreation_reward_func": 0.17423443992932638,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
       "rewards/xmlcount_reward_func": 0.125,
@@ -108,10 +108,10 @@
       "epoch": 3.4,
       "step": 10,
       "total_flos": 0.0,
-      "train_loss": -3.1709017093817236e-06,
-      "train_runtime": 590.6267,
-      "train_samples_per_second": 0.068,
-      "train_steps_per_second": 0.017
+      "train_loss": 6.0011893765477e-06,
+      "train_runtime": 503.4756,
+      "train_samples_per_second": 0.079,
+      "train_steps_per_second": 0.02
     }
   ],
   "logging_steps": 2,