End of training

adapter_config.json

@@ -25,12 +25,12 @@
   "revision": null,
   "target_modules": [
     "v_proj",
-    "down_proj",
-    "q_proj",
-    "o_proj",
     "up_proj",
     "k_proj",
-    "gate_proj"
+    "gate_proj",
+    "q_proj",
+    "down_proj",
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM",
   "trainable_token_indices": null,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b32d9a8e154d85fc4936689e41302d5522c1e8980d53399bff92ed91cfdd10e7
+oid sha256:d5aa03f5f63734fc9cb08d204dc9801b70d80b6a766ee92884c4551f0a934f7e
 size 73911112

all_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 8.028497632039944e-07,
-    "train_runtime": 305.3022,
-    "train_samples": 14,
-    "train_samples_per_second": 1.048,
-    "train_steps_per_second": 0.066
+    "train_loss": 4.897040355444915e-07,
+    "train_runtime": 1090.847,
+    "train_samples": 11,
+    "train_samples_per_second": 0.293,
+    "train_steps_per_second": 0.018
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "total_flos": 0.0,
-    "train_loss": 8.028497632039944e-07,
-    "train_runtime": 305.3022,
-    "train_samples": 14,
-    "train_samples_per_second": 1.048,
-    "train_steps_per_second": 0.066
+    "train_loss": 4.897040355444915e-07,
+    "train_runtime": 1090.847,
+    "train_samples": 11,
+    "train_samples_per_second": 0.293,
+    "train_steps_per_second": 0.018
 }

trainer_state.json

@@ -2,7 +2,7 @@
   "best_global_step": null,
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 5.0,
+  "epoch": 6.7272727272727275,
   "eval_steps": 500,
   "global_step": 20,
   "is_hyper_param_search": false,
@@ -10,209 +10,209 @@
   "is_world_process_zero": true,
   "log_history": [
     {
-      "completion_length": 336.71875,
-      "epoch": 0.5714285714285714,
-      "grad_norm": 0.549845278263092,
-      "kl": 0.0009074829140445217,
+      "completion_length": 308.3125,
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.6000280380249023,
+      "kl": 0.0,
       "learning_rate": 5e-07,
-      "loss": 0.0,
-      "reward": 1.3631393127143383,
-      "reward_std": 1.0872603710740805,
-      "rewards/concensus_correctness_reward_func": 0.01093749818392098,
-      "rewards/consensus_reward_func": 0.3125,
+      "loss": -0.0,
+      "reward": 0.909564670175314,
+      "reward_std": 0.8405304190819152,
+      "rewards/concensus_correctness_reward_func": 0.0,
+      "rewards/consensus_reward_func": 0.0625,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.125,
-      "rewards/question_recreation_reward_func": 0.5159518122673035,
+      "rewards/final_correctness_reward_func": 0.0,
+      "rewards/question_recreation_reward_func": 0.5372834168374538,
       "rewards/soft_format_reward_func": 0.0,
-      "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.3987500111106783,
+      "rewards/strict_format_reward_func": 0.03125,
+      "rewards/xmlcount_reward_func": 0.2785312468186021,
       "step": 2
     },
     {
-      "completion_length": 215.875,
-      "epoch": 1.0,
-      "grad_norm": 0.4852951467037201,
-      "kl": 0.0008874579701417437,
+      "completion_length": 387.89285714285717,
+      "epoch": 1.3636363636363638,
+      "grad_norm": 0.47270771861076355,
+      "kl": 0.0008392650883511773,
       "learning_rate": 4.864543104251586e-07,
       "loss": 0.0,
-      "reward": 5.954347332318624,
-      "reward_std": 4.38392773270607,
-      "rewards/concensus_correctness_reward_func": 3.632083316644033,
-      "rewards/consensus_reward_func": 0.5,
+      "reward": 2.5110427141189575,
+      "reward_std": 2.5342068799904416,
+      "rewards/concensus_correctness_reward_func": 0.75,
+      "rewards/consensus_reward_func": 0.21428571428571427,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.75,
-      "rewards/question_recreation_reward_func": 0.48168083280324936,
+      "rewards/final_correctness_reward_func": 0.35714285714285715,
+      "rewards/question_recreation_reward_func": 0.5590784060103553,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.5905833343664805,
+      "rewards/xmlcount_reward_func": 0.6305357217788696,
       "step": 4
     },
     {
-      "completion_length": 329.53125,
-      "epoch": 1.5714285714285714,
-      "grad_norm": 0.6936220526695251,
-      "kl": 0.0008142793085426092,
+      "completion_length": 288.32142857142856,
+      "epoch": 2.0,
+      "grad_norm": 0.9542892575263977,
+      "kl": 0.0009663021774031222,
       "learning_rate": 4.472851273490984e-07,
       "loss": 0.0,
-      "reward": 2.0221757367253304,
-      "reward_std": 2.2516359835863113,
-      "rewards/concensus_correctness_reward_func": 0.6874999981373549,
-      "rewards/consensus_reward_func": 0.25,
+      "reward": 1.170509974871363,
+      "reward_std": 0.7048059260206563,
+      "rewards/concensus_correctness_reward_func": 0.0,
+      "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.3125,
-      "rewards/question_recreation_reward_func": 0.465550834313035,
+      "rewards/final_correctness_reward_func": 0.35714285714285715,
+      "rewards/question_recreation_reward_func": 0.4116528176569513,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.3066249960102141,
+      "rewards/xmlcount_reward_func": 0.4017142823764256,
       "step": 6
     },
     {
-      "completion_length": 239.75,
-      "epoch": 2.0,
-      "grad_norm": 0.3220447599887848,
-      "kl": 0.0009201703360304236,
+      "completion_length": 368.5625,
+      "epoch": 2.7272727272727275,
+      "grad_norm": 0.4186118245124817,
+      "kl": 0.0008336683677043766,
       "learning_rate": 3.867370395306068e-07,
       "loss": 0.0,
-      "reward": 1.879896601041158,
-      "reward_std": 1.3349867115418117,
-      "rewards/concensus_correctness_reward_func": 0.34416667371988297,
-      "rewards/consensus_reward_func": 0.16666666666666666,
+      "reward": 1.4782422371208668,
+      "reward_std": 1.4278910955326864,
+      "rewards/concensus_correctness_reward_func": 0.35118750110268593,
+      "rewards/consensus_reward_func": 0.125,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.5,
-      "rewards/question_recreation_reward_func": 0.4502299043039481,
+      "rewards/final_correctness_reward_func": 0.4375,
+      "rewards/question_recreation_reward_func": 0.3739297694992274,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.4188333222021659,
+      "rewards/xmlcount_reward_func": 0.19062500353902578,
       "step": 8
     },
     {
-      "completion_length": 319.75,
-      "epoch": 2.571428571428571,
-      "grad_norm": 0.462271124124527,
-      "kl": 0.0008407420027651824,
+      "completion_length": 363.82142857142856,
+      "epoch": 3.3636363636363638,
+      "grad_norm": 0.5186935067176819,
+      "kl": 0.0008661752134295446,
       "learning_rate": 3.1137137178519977e-07,
       "loss": 0.0,
-      "reward": 2.5945708453655243,
-      "reward_std": 2.1639809533953667,
-      "rewards/concensus_correctness_reward_func": 0.8006250308826566,
-      "rewards/consensus_reward_func": 0.4375,
+      "reward": 2.3520706466266086,
+      "reward_std": 2.343413374253682,
+      "rewards/concensus_correctness_reward_func": 0.8392857142857143,
+      "rewards/consensus_reward_func": 0.14285714285714285,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.25,
-      "rewards/question_recreation_reward_func": 0.5328521132469177,
+      "rewards/final_correctness_reward_func": 0.21428571428571427,
+      "rewards/question_recreation_reward_func": 0.6217134882296834,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.5735937505960464,
+      "rewards/xmlcount_reward_func": 0.5339285731315613,
       "step": 10
     },
     {
-      "completion_length": 276.4583333333333,
-      "epoch": 3.0,
-      "grad_norm": 0.4189568758010864,
-      "kl": 0.0011232300797322143,
+      "completion_length": 335.32142857142856,
+      "epoch": 4.0,
+      "grad_norm": 0.40881654620170593,
+      "kl": 0.0008830937939429921,
       "learning_rate": 2.2935516363191693e-07,
       "loss": 0.0,
-      "reward": 2.281927933295568,
-      "reward_std": 2.3937632540861764,
-      "rewards/concensus_correctness_reward_func": 0.9300000021855036,
-      "rewards/consensus_reward_func": 0.3333333333333333,
+      "reward": 0.9431978527988706,
+      "reward_std": 0.9636511317346178,
+      "rewards/concensus_correctness_reward_func": 0.0442142869744982,
+      "rewards/consensus_reward_func": 0.07142857142857142,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.4166666666666667,
-      "rewards/question_recreation_reward_func": 0.34013621260722476,
+      "rewards/final_correctness_reward_func": 0.2857142857142857,
+      "rewards/question_recreation_reward_func": 0.28391214460134506,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.2617916688323021,
+      "rewards/xmlcount_reward_func": 0.2579285715307508,
       "step": 12
     },
     {
-      "completion_length": 309.40625,
-      "epoch": 3.571428571428571,
-      "grad_norm": 0.49406951665878296,
-      "kl": 0.0008451644243905321,
+      "completion_length": 395.0,
+      "epoch": 4.7272727272727275,
+      "grad_norm": 0.4665830135345459,
+      "kl": 0.0007938470225781202,
       "learning_rate": 1.4957614383675767e-07,
       "loss": 0.0,
-      "reward": 3.1388991102576256,
-      "reward_std": 3.611638531088829,
-      "rewards/concensus_correctness_reward_func": 1.4150625318288803,
-      "rewards/consensus_reward_func": 0.3125,
+      "reward": 2.6766241267323494,
+      "reward_std": 2.357332043349743,
+      "rewards/concensus_correctness_reward_func": 1.25,
+      "rewards/consensus_reward_func": 0.1875,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.3125,
-      "rewards/question_recreation_reward_func": 0.635461613535881,
+      "rewards/final_correctness_reward_func": 0.1875,
+      "rewards/question_recreation_reward_func": 0.6196865616366267,
       "rewards/soft_format_reward_func": 0.0,
-      "rewards/strict_format_reward_func": 0.015625,
-      "rewards/xmlcount_reward_func": 0.4477499946951866,
+      "rewards/strict_format_reward_func": 0.0,
+      "rewards/xmlcount_reward_func": 0.4319375120103359,
       "step": 14
     },
     {
-      "completion_length": 262.3333333333333,
-      "epoch": 4.0,
-      "grad_norm": 0.3369537889957428,
-      "kl": 0.0012845705399134506,
+      "completion_length": 363.5357142857143,
+      "epoch": 5.363636363636363,
+      "grad_norm": 0.4923468232154846,
+      "kl": 0.0008862724145209151,
       "learning_rate": 8.067960709356478e-08,
-      "loss": 0.0,
-      "reward": 4.179730216662089,
-      "reward_std": 2.806457077463468,
-      "rewards/concensus_correctness_reward_func": 2.5208333333333335,
-      "rewards/consensus_reward_func": 0.6666666666666666,
+      "loss": -0.0,
+      "reward": 2.118965187243053,
+      "reward_std": 2.2687917195850917,
+      "rewards/concensus_correctness_reward_func": 0.7142857142857143,
+      "rewards/consensus_reward_func": 0.14285714285714285,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.08333333333333333,
-      "rewards/question_recreation_reward_func": 0.32602185383439064,
+      "rewards/final_correctness_reward_func": 0.42857142857142855,
+      "rewards/question_recreation_reward_func": 0.4556793837941119,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.5828749810655912,
+      "rewards/xmlcount_reward_func": 0.37757142526762827,
       "step": 16
     },
     {
-      "completion_length": 297.78125,
-      "epoch": 4.571428571428571,
-      "grad_norm": 0.504997730255127,
-      "kl": 0.0009279289661208168,
+      "completion_length": 322.64285714285717,
+      "epoch": 6.0,
+      "grad_norm": 0.450918048620224,
+      "kl": 0.0008349009772895702,
       "learning_rate": 3.013156219837776e-08,
-      "loss": 0.0,
-      "reward": 3.203057140111923,
-      "reward_std": 2.8058369271457195,
-      "rewards/concensus_correctness_reward_func": 1.4804375015664846,
-      "rewards/consensus_reward_func": 0.5,
+      "loss": -0.0,
+      "reward": 1.394048320395606,
+      "reward_std": 0.9764726414578035,
+      "rewards/concensus_correctness_reward_func": 0.0,
+      "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.3125,
-      "rewards/question_recreation_reward_func": 0.5565572455525398,
+      "rewards/final_correctness_reward_func": 0.2857142857142857,
+      "rewards/question_recreation_reward_func": 0.5164054871669838,
       "rewards/soft_format_reward_func": 0.0,
-      "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.35356250358745456,
+      "rewards/strict_format_reward_func": 0.017857142857142856,
+      "rewards/xmlcount_reward_func": 0.5740714243480137,
       "step": 18
     },
     {
-      "completion_length": 327.4166666666667,
-      "epoch": 5.0,
-      "grad_norm": 0.6248384714126587,
-      "kl": 0.0007756088064828267,
+      "completion_length": 311.28125,
+      "epoch": 6.7272727272727275,
+      "grad_norm": 0.8445176482200623,
+      "kl": 0.0009809081529965624,
       "learning_rate": 3.4096741493194193e-09,
       "loss": 0.0,
-      "reward": 1.846030871073405,
-      "reward_std": 2.4185179409881434,
-      "rewards/concensus_correctness_reward_func": 0.8295833331843218,
-      "rewards/consensus_reward_func": 0.16666666666666666,
+      "reward": 1.2761643175035715,
+      "reward_std": 0.9043446108698845,
+      "rewards/concensus_correctness_reward_func": 0.05431250110268593,
+      "rewards/consensus_reward_func": 0.0,
       "rewards/cumulative_reward_2": 0.0,
-      "rewards/final_correctness_reward_func": 0.16666666666666666,
-      "rewards/question_recreation_reward_func": 0.45132258037726086,
+      "rewards/final_correctness_reward_func": 0.3125,
+      "rewards/question_recreation_reward_func": 0.4313830805476755,
       "rewards/soft_format_reward_func": 0.0,
       "rewards/strict_format_reward_func": 0.0,
-      "rewards/xmlcount_reward_func": 0.23179167633255324,
+      "rewards/xmlcount_reward_func": 0.4779687486588955,
       "step": 20
     },
     {
-      "epoch": 5.0,
+      "epoch": 6.7272727272727275,
       "step": 20,
       "total_flos": 0.0,
-      "train_loss": 8.028497632039944e-07,
-      "train_runtime": 305.3022,
-      "train_samples_per_second": 1.048,
-      "train_steps_per_second": 0.066
+      "train_loss": 4.897040355444915e-07,
+      "train_runtime": 1090.847,
+      "train_samples_per_second": 0.293,
+      "train_steps_per_second": 0.018
     }
   ],
   "logging_steps": 2,
   "max_steps": 20,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 7,
+  "num_train_epochs": 10,
   "save_steps": 25,
   "stateful_callbacks": {
     "TrainerControl": {

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9a3221d4ec3964e3a416c996864dc87672dfa6a85e026a4b6cc77d174fda4632
+oid sha256:b808538c4f97e65834cbeaa89a34a4e6e283623f2778ee5610d4ad34e3163cde
 size 6008