End of training

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 1.2114644050598144e-06,
-    "train_runtime": 644.7074,
+    "train_loss": -3.1709017093817236e-06,
+    "train_runtime": 590.6267,
     "train_samples": 5,
-    "train_samples_per_second": 0.062,
-    "train_steps_per_second": 0.016
+    "train_samples_per_second": 0.068,
+    "train_steps_per_second": 0.017
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0e76fc2b921e395ae8a19c2d083531d8c42be442e8b6d796f2c0d01da8ee153b
+oid sha256:49ca111340ed7c945725702a2578128d3a9247131a8c1fab02ed7fed9eb67b1f
 size 1976163472

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 1.2114644050598144e-06,
-    "train_runtime": 644.7074,
+    "train_loss": -3.1709017093817236e-06,
+    "train_runtime": 590.6267,
     "train_samples": 5,
-    "train_samples_per_second": 0.062,
-    "train_steps_per_second": 0.016
+    "train_samples_per_second": 0.068,
+    "train_steps_per_second": 0.017
 }

trainer_state.json

@@ -10,95 +10,95 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "completion_length": 82.375,
+      "completion_length": 78.375,
       "epoch": 0.8,
-      "grad_norm": 6.81792688369751,
+      "grad_norm": 5.627739906311035,
       "kl": 0.0,
       "learning_rate": 5e-07,
-      "loss": -0.0,
-      "reward": 1.0713742524385452,
-      "reward_std": 0.35633096733363345,
+      "loss": 0.0,
+      "reward": 0.7521542385220528,
+      "reward_std": 0.044539074537169654,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.75,
-      "rewards/question_recreation_reward_func": 0.19637423381209373,
+      "rewards/final_correctness_reward_func": 0.5,
+      "rewards/question_recreation_reward_func": 0.15840421058237553,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.125,
+      "rewards/xmlcount_reward_func": 0.09375,
       "step": 2
     },
     {
-      "completion_length": 63.833333333333336,
+      "completion_length": 49.666666666666664,
       "epoch": 1.4,
-      "grad_norm": 6.891765117645264,
-      "kl": 0.0008907011342671467,
+      "grad_norm": 0.005242869723588228,
+      "kl": 0.016141654608266737,
       "learning_rate": 4.415111107797445e-07,
       "loss": 0.0,
-      "reward": 1.8277830282847087,
-      "reward_std": 0.01574741117656231,
+      "reward": 1.9181493918100994,
+      "reward_std": 0.015638279418150585,
       "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": 0.3694497048854828,
+      "rewards/question_recreation_reward_func": 0.4598161180814107,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
       "rewards/xmlcount_reward_func": 0.125,
       "step": 4
     },
     {
-      "completion_length": 86.33333333333333,
+      "completion_length": 58.666666666666664,
       "epoch": 2.0,
-      "grad_norm": 5.744239807128906,
-      "kl": 0.0014734831638634205,
+      "grad_norm": 7.197663307189941,
+      "kl": 0.010560240286091963,
       "learning_rate": 2.934120444167326e-07,
-      "loss": -0.0,
-      "reward": 1.585513969262441,
-      "reward_std": 0.005068938247859478,
+      "loss": 0.0,
+      "reward": 0.9169059892495474,
+      "reward_std": 0.0018884773987034957,
       "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": 0.12718065331379572,
+      "rewards/final_correctness_reward_func": 0.6666666666666666,
+      "rewards/question_recreation_reward_func": 0.16690599421660104,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.125,
+      "rewards/xmlcount_reward_func": 0.08333333333333333,
       "step": 6
     },
     {
-      "completion_length": 77.375,
+      "completion_length": 61.375,
       "epoch": 2.8,
-      "grad_norm": 5.462680339813232,
-      "kl": 0.0020970748155377805,
+      "grad_norm": 2.7515034675598145,
+      "kl": 0.004465199788683094,
       "learning_rate": 1.2500000000000005e-07,
       "loss": 0.0,
-      "reward": 1.8065865561366081,
-      "reward_std": 0.006950426250114106,
+      "reward": 1.361022423952818,
+      "reward_std": 0.00136809196555987,
       "rewards/concensus_correctness_reward_func": 0.0,
       "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 1.5,
-      "rewards/question_recreation_reward_func": 0.18158655054867268,
+      "rewards/final_correctness_reward_func": 1.0,
+      "rewards/question_recreation_reward_func": 0.26727242209017277,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.125,
+      "rewards/xmlcount_reward_func": 0.09375,
       "step": 8
     },
     {
-      "completion_length": 81.5,
+      "completion_length": 70.0,
       "epoch": 3.4,
-      "grad_norm": 4.858218669891357,
-      "kl": 0.009592992641652623,
+      "grad_norm": 6.803511619567871,
+      "kl": 0.01901205163449049,
       "learning_rate": 1.507684480352292e-08,
-      "loss": 0.0,
-      "reward": 0.9994062383969625,
-      "reward_std": 0.0025496524904156104,
+      "loss": -0.0,
+      "reward": 0.9308561086654663,
+      "reward_std": 0.005250229577844341,
       "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": 0.2077395518620809,
+      "rewards/question_recreation_reward_func": 0.1391894891858101,
       "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": 1.2114644050598144e-06,
-      "train_runtime": 644.7074,
-      "train_samples_per_second": 0.062,
-      "train_steps_per_second": 0.016
+      "train_loss": -3.1709017093817236e-06,
+      "train_runtime": 590.6267,
+      "train_samples_per_second": 0.068,
+      "train_steps_per_second": 0.017
     }
   ],
   "logging_steps": 2,