Model save

README.md

@@ -1,13 +1,11 @@
 ---
+license: apache-2.0
 library_name: peft
 tags:
-- alignment-handbook
 - trl
 - dpo
 - generated_from_trainer
 base_model: norallm/normistral-7b-warm
-datasets:
-- hugodk-sch/aftonposten_title_prefs
 model-index:
 - name: ap-normistral-7b-align-scan
   results: []
@@ -18,17 +16,17 @@ should probably proofread and complete it, then remove this comment. -->
 
 # ap-normistral-7b-align-scan
 
-This model is a fine-tuned version of [data/ap-normistral-7b-sft-qlora](https://huggingface.co/data/ap-normistral-7b-sft-qlora) on the hugodk-sch/aftonposten_title_prefs dataset.
+This model is a fine-tuned version of [norallm/normistral-7b-warm](https://huggingface.co/norallm/normistral-7b-warm) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.6864
-- Rewards/chosen: -0.0790
-- Rewards/rejected: -0.1771
-- Rewards/accuracies: 0.5685
-- Rewards/margins: 0.0982
-- Logps/rejected: -36.4094
-- Logps/chosen: -32.6406
-- Logits/rejected: 98.4364
-- Logits/chosen: 98.4629
+- Loss: 0.6733
+- Rewards/chosen: -0.0561
+- Rewards/rejected: -0.1040
+- Rewards/accuracies: 0.6121
+- Rewards/margins: 0.0480
+- Logps/rejected: -37.0066
+- Logps/chosen: -33.0038
+- Logits/rejected: 97.9574
+- Logits/chosen: 97.9847
 
 ## Model description
 
@@ -63,9 +61,9 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
 |:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
-| 0.6649        | 0.26  | 100  | 0.7134          | -0.0147        | -0.0314          | 0.5220             | 0.0167          | -36.0449       | -32.4799     | 98.6822         | 98.6961       |
-| 0.6015        | 0.52  | 200  | 0.6984          | -0.1481        | -0.2178          | 0.5403             | 0.0696          | -36.5109       | -32.8134     | 98.4369         | 98.4609       |
-| 0.5603        | 0.78  | 300  | 0.7084          | -0.0908        | -0.1390          | 0.5399             | 0.0482          | -36.3141       | -32.6701     | 98.4570         | 98.4833       |
+| 0.6795        | 0.26  | 100  | 0.6967          | 0.0004         | 0.0027           | 0.4846             | -0.0023         | -35.9393       | -32.4388     | 98.7026         | 98.7111       |
+| 0.6355        | 0.52  | 200  | 0.6793          | -0.0461        | -0.0803          | 0.5835             | 0.0342          | -36.7700       | -32.9042     | 98.1082         | 98.1292       |
+| 0.6306        | 0.78  | 300  | 0.6733          | -0.0561        | -0.1040          | 0.6121             | 0.0480          | -37.0066       | -33.0038     | 97.9574         | 97.9847       |
 
 
 ### Framework versions

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8beb8439f31b24abb57cd224b5fd2d430a90408c58d00b3253d8fa51d2d2a640
+oid sha256:47431e82b8123afc83e0cf1e9391a2c9c8afe09db86fd986151fa85019d9af9f
 size 671150064

all_results.json

@@ -1,21 +1,8 @@
 {
     "epoch": 1.0,
-    "eval_logits/chosen": 98.4628677368164,
-    "eval_logits/rejected": 98.43635559082031,
-    "eval_logps/chosen": -32.64057159423828,
-    "eval_logps/rejected": -36.409393310546875,
-    "eval_loss": 0.686401903629303,
-    "eval_rewards/accuracies": 0.5685215592384338,
-    "eval_rewards/chosen": -0.07895812392234802,
-    "eval_rewards/margins": 0.09818949550390244,
-    "eval_rewards/rejected": -0.17714762687683105,
-    "eval_runtime": 103.7748,
-    "eval_samples": 343,
-    "eval_samples_per_second": 3.305,
-    "eval_steps_per_second": 0.414,
-    "train_loss": 0.5890550520512965,
-    "train_runtime": 2558.5725,
+    "train_loss": 0.6461187455561254,
+    "train_runtime": 2556.0398,
     "train_samples": 3079,
-    "train_samples_per_second": 1.203,
-    "train_steps_per_second": 0.15
+    "train_samples_per_second": 1.205,
+    "train_steps_per_second": 0.151
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 1.0,
-    "train_loss": 0.5890550520512965,
-    "train_runtime": 2558.5725,
+    "train_loss": 0.6461187455561254,
+    "train_runtime": 2556.0398,
     "train_samples": 3079,
-    "train_samples_per_second": 1.203,
-    "train_steps_per_second": 0.15
+    "train_samples_per_second": 1.205,
+    "train_steps_per_second": 0.151
 }

trainer_state.json

@@ -10,7 +10,7 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "grad_norm": 26.125,
+      "grad_norm": 6.53125,
       "learning_rate": 1.282051282051282e-07,
       "logits/chosen": 88.18099975585938,
       "logits/rejected": 88.25153350830078,
@@ -25,630 +25,630 @@
     },
     {
       "epoch": 0.03,
-      "grad_norm": 20.75,
+      "grad_norm": 5.375,
       "learning_rate": 1.282051282051282e-06,
-      "logits/chosen": 81.08618927001953,
-      "logits/rejected": 80.79019165039062,
-      "logps/chosen": -34.27778625488281,
-      "logps/rejected": -33.089111328125,
-      "loss": 0.6843,
-      "rewards/accuracies": 0.4861111044883728,
-      "rewards/chosen": -0.014423643238842487,
-      "rewards/margins": 0.037857502698898315,
-      "rewards/rejected": -0.05228114500641823,
+      "logits/chosen": 81.08163452148438,
+      "logits/rejected": 80.7850341796875,
+      "logps/chosen": -34.20207977294922,
+      "logps/rejected": -33.01838684082031,
+      "loss": 0.6889,
+      "rewards/accuracies": 0.4722222089767456,
+      "rewards/chosen": 0.003964493051171303,
+      "rewards/margins": 0.009962627664208412,
+      "rewards/rejected": -0.005998134613037109,
       "step": 10
     },
     {
       "epoch": 0.05,
-      "grad_norm": 23.5,
+      "grad_norm": 6.03125,
       "learning_rate": 2.564102564102564e-06,
-      "logits/chosen": 80.65704345703125,
-      "logits/rejected": 80.54637908935547,
-      "logps/chosen": -33.64678192138672,
-      "logps/rejected": -30.826171875,
-      "loss": 0.6907,
-      "rewards/accuracies": 0.5249999761581421,
-      "rewards/chosen": 0.017251869663596153,
-      "rewards/margins": 0.03672502189874649,
-      "rewards/rejected": -0.019473150372505188,
+      "logits/chosen": 80.66960144042969,
+      "logits/rejected": 80.55855560302734,
+      "logps/chosen": -33.57989501953125,
+      "logps/rejected": -30.827892303466797,
+      "loss": 0.6865,
+      "rewards/accuracies": 0.5625,
+      "rewards/chosen": 0.01100135874003172,
+      "rewards/margins": 0.016041692346334457,
+      "rewards/rejected": -0.005040335468947887,
       "step": 20
     },
     {
       "epoch": 0.08,
-      "grad_norm": 23.375,
+      "grad_norm": 5.875,
       "learning_rate": 3.846153846153847e-06,
-      "logits/chosen": 82.53121185302734,
-      "logits/rejected": 82.56344604492188,
-      "logps/chosen": -33.85637664794922,
-      "logps/rejected": -31.18195152282715,
-      "loss": 0.7305,
-      "rewards/accuracies": 0.4375,
-      "rewards/chosen": 0.06257696449756622,
-      "rewards/margins": -0.03417497128248215,
-      "rewards/rejected": 0.09675192832946777,
+      "logits/chosen": 82.521240234375,
+      "logits/rejected": 82.55521392822266,
+      "logps/chosen": -33.74619674682617,
+      "logps/rejected": -31.233224868774414,
+      "loss": 0.6911,
+      "rewards/accuracies": 0.4749999940395355,
+      "rewards/chosen": 0.02666253224015236,
+      "rewards/margins": 0.007601564284414053,
+      "rewards/rejected": 0.01906096562743187,
       "step": 30
     },
     {
       "epoch": 0.1,
-      "grad_norm": 21.625,
+      "grad_norm": 5.46875,
       "learning_rate": 4.999896948438434e-06,
-      "logits/chosen": 81.05611419677734,
-      "logits/rejected": 81.05009460449219,
-      "logps/chosen": -32.67264175415039,
-      "logps/rejected": -33.195556640625,
-      "loss": 0.6623,
-      "rewards/accuracies": 0.5874999761581421,
-      "rewards/chosen": 0.20056962966918945,
-      "rewards/margins": 0.13138213753700256,
-      "rewards/rejected": 0.0691874772310257,
+      "logits/chosen": 81.10136413574219,
+      "logits/rejected": 81.09754943847656,
+      "logps/chosen": -32.667869567871094,
+      "logps/rejected": -33.15234375,
+      "loss": 0.682,
+      "rewards/accuracies": 0.5625,
+      "rewards/chosen": 0.05061974376440048,
+      "rewards/margins": 0.02900140918791294,
+      "rewards/rejected": 0.02161833457648754,
       "step": 40
     },
     {
       "epoch": 0.13,
-      "grad_norm": 16.5,
+      "grad_norm": 4.96875,
       "learning_rate": 4.987541037542187e-06,
-      "logits/chosen": 78.71464538574219,
-      "logits/rejected": 78.72676086425781,
-      "logps/chosen": -30.6053466796875,
-      "logps/rejected": -30.86154556274414,
-      "loss": 0.6674,
-      "rewards/accuracies": 0.5874999761581421,
-      "rewards/chosen": 0.23996099829673767,
-      "rewards/margins": 0.15526394546031952,
-      "rewards/rejected": 0.08469705283641815,
+      "logits/chosen": 78.831787109375,
+      "logits/rejected": 78.83922576904297,
+      "logps/chosen": -30.56161117553711,
+      "logps/rejected": -30.64150619506836,
+      "loss": 0.6869,
+      "rewards/accuracies": 0.5375000238418579,
+      "rewards/chosen": 0.06436345726251602,
+      "rewards/margins": 0.021185602992773056,
+      "rewards/rejected": 0.043177854269742966,
       "step": 50
     },
     {
       "epoch": 0.16,
-      "grad_norm": 20.375,
+      "grad_norm": 5.09375,
       "learning_rate": 4.954691471941119e-06,
-      "logits/chosen": 83.24500274658203,
-      "logits/rejected": 83.30210876464844,
-      "logps/chosen": -30.880474090576172,
-      "logps/rejected": -29.403039932250977,
-      "loss": 0.7121,
-      "rewards/accuracies": 0.512499988079071,
-      "rewards/chosen": 0.11787523329257965,
-      "rewards/margins": 0.04288250952959061,
-      "rewards/rejected": 0.07499273121356964,
+      "logits/chosen": 83.43026733398438,
+      "logits/rejected": 83.48649597167969,
+      "logps/chosen": -30.803913116455078,
+      "logps/rejected": -29.231754302978516,
+      "loss": 0.6958,
+      "rewards/accuracies": 0.4749999940395355,
+      "rewards/chosen": 0.03712441027164459,
+      "rewards/margins": 0.0012475885450839996,
+      "rewards/rejected": 0.03587682172656059,
       "step": 60
     },
     {
       "epoch": 0.18,
-      "grad_norm": 30.75,
+      "grad_norm": 6.625,
       "learning_rate": 4.901618883413549e-06,
-      "logits/chosen": 83.83995056152344,
-      "logits/rejected": 83.8766098022461,
-      "logps/chosen": -30.424224853515625,
-      "logps/rejected": -32.918212890625,
-      "loss": 0.704,
-      "rewards/accuracies": 0.5625,
-      "rewards/chosen": 0.08192038536071777,
-      "rewards/margins": 0.03412293642759323,
-      "rewards/rejected": 0.04779745265841484,
+      "logits/chosen": 84.04580688476562,
+      "logits/rejected": 84.08148193359375,
+      "logps/chosen": -30.23891830444336,
+      "logps/rejected": -32.60674285888672,
+      "loss": 0.6974,
+      "rewards/accuracies": 0.4625000059604645,
+      "rewards/chosen": 0.03901054710149765,
+      "rewards/margins": -0.004085768014192581,
+      "rewards/rejected": 0.04309631139039993,
       "step": 70
     },
     {
       "epoch": 0.21,
-      "grad_norm": 20.5,
+      "grad_norm": 6.1875,
       "learning_rate": 4.828760511501322e-06,
-      "logits/chosen": 81.50079345703125,
-      "logits/rejected": 81.48374938964844,
-      "logps/chosen": -31.288936614990234,
-      "logps/rejected": -30.928760528564453,
-      "loss": 0.6548,
-      "rewards/accuracies": 0.550000011920929,
-      "rewards/chosen": 0.08640192449092865,
-      "rewards/margins": 0.15484753251075745,
-      "rewards/rejected": -0.0684456080198288,
+      "logits/chosen": 81.75585174560547,
+      "logits/rejected": 81.73826599121094,
+      "logps/chosen": -31.039302825927734,
+      "logps/rejected": -30.76405906677246,
+      "loss": 0.6731,
+      "rewards/accuracies": 0.6000000238418579,
+      "rewards/chosen": 0.046564072370529175,
+      "rewards/margins": 0.047205500304698944,
+      "rewards/rejected": -0.0006414322415366769,
       "step": 80
     },
     {
       "epoch": 0.23,
-      "grad_norm": 27.375,
+      "grad_norm": 7.1875,
       "learning_rate": 4.7367166013034295e-06,
-      "logits/chosen": 78.15573120117188,
-      "logits/rejected": 78.12377166748047,
-      "logps/chosen": -32.45014190673828,
-      "logps/rejected": -31.189159393310547,
-      "loss": 0.6587,
+      "logits/chosen": 78.49934387207031,
+      "logits/rejected": 78.46571350097656,
+      "logps/chosen": -32.25204849243164,
+      "logps/rejected": -30.996267318725586,
+      "loss": 0.6747,
       "rewards/accuracies": 0.5625,
-      "rewards/chosen": 0.07521401345729828,
-      "rewards/margins": 0.17028900980949402,
-      "rewards/rejected": -0.09507499635219574,
+      "rewards/chosen": 0.038613028824329376,
+      "rewards/margins": 0.04309249296784401,
+      "rewards/rejected": -0.004479461349546909,
       "step": 90
     },
     {
       "epoch": 0.26,
-      "grad_norm": 21.25,
+      "grad_norm": 6.125,
       "learning_rate": 4.626245458345211e-06,
-      "logits/chosen": 83.35581970214844,
-      "logits/rejected": 83.3793716430664,
-      "logps/chosen": -33.93380355834961,
-      "logps/rejected": -31.7375431060791,
-      "loss": 0.6649,
-      "rewards/accuracies": 0.625,
-      "rewards/chosen": 0.14847493171691895,
-      "rewards/margins": 0.14327403903007507,
-      "rewards/rejected": 0.005200871266424656,
+      "logits/chosen": 83.57372283935547,
+      "logits/rejected": 83.60108947753906,
+      "logps/chosen": -33.96754455566406,
+      "logps/rejected": -31.743358612060547,
+      "loss": 0.6795,
+      "rewards/accuracies": 0.550000011920929,
+      "rewards/chosen": 0.0337446853518486,
+      "rewards/margins": 0.033025771379470825,
+      "rewards/rejected": 0.0007189197349362075,
       "step": 100
     },
     {
       "epoch": 0.26,
-      "eval_logits/chosen": 98.69609832763672,
-      "eval_logits/rejected": 98.68218231201172,
-      "eval_logps/chosen": -32.47989273071289,
-      "eval_logps/rejected": -36.04493713378906,
-      "eval_loss": 0.7133548259735107,
-      "eval_rewards/accuracies": 0.5220099687576294,
-      "eval_rewards/chosen": -0.01468663476407528,
-      "eval_rewards/margins": 0.01667696051299572,
-      "eval_rewards/rejected": -0.0313635915517807,
-      "eval_runtime": 104.2778,
-      "eval_samples_per_second": 3.289,
-      "eval_steps_per_second": 0.412,
+      "eval_logits/chosen": 98.71112060546875,
+      "eval_logits/rejected": 98.70260620117188,
+      "eval_logps/chosen": -32.43876647949219,
+      "eval_logps/rejected": -35.93931198120117,
+      "eval_loss": 0.696650505065918,
+      "eval_rewards/accuracies": 0.4846345782279968,
+      "eval_rewards/chosen": 0.0004411655245348811,
+      "eval_rewards/margins": -0.0022806311026215553,
+      "eval_rewards/rejected": 0.00272179557941854,
+      "eval_runtime": 104.2161,
+      "eval_samples_per_second": 3.291,
+      "eval_steps_per_second": 0.413,
       "step": 100
     },
     {
       "epoch": 0.29,
-      "grad_norm": 27.375,
+      "grad_norm": 7.03125,
       "learning_rate": 4.498257201263691e-06,
-      "logits/chosen": 83.54657745361328,
-      "logits/rejected": 83.43326568603516,
-      "logps/chosen": -32.3030891418457,
-      "logps/rejected": -32.73809051513672,
-      "loss": 0.6007,
-      "rewards/accuracies": 0.675000011920929,
-      "rewards/chosen": 0.2878861129283905,
-      "rewards/margins": 0.3207677900791168,
-      "rewards/rejected": -0.03288168087601662,
+      "logits/chosen": 83.73343658447266,
+      "logits/rejected": 83.620361328125,
+      "logps/chosen": -32.29249954223633,
+      "logps/rejected": -32.76134490966797,
+      "loss": 0.6562,
+      "rewards/accuracies": 0.6625000238418579,
+      "rewards/chosen": 0.07303053885698318,
+      "rewards/margins": 0.08357587456703186,
+      "rewards/rejected": -0.010545337572693825,
       "step": 110
     },
     {
       "epoch": 0.31,
-      "grad_norm": 30.625,
+      "grad_norm": 6.84375,
       "learning_rate": 4.353806263777678e-06,
-      "logits/chosen": 83.636474609375,
-      "logits/rejected": 83.75120544433594,
-      "logps/chosen": -28.302623748779297,
-      "logps/rejected": -35.44631576538086,
-      "loss": 0.6186,
-      "rewards/accuracies": 0.6499999761581421,
-      "rewards/chosen": 0.2507820725440979,
-      "rewards/margins": 0.2456505298614502,
-      "rewards/rejected": 0.005131528712809086,
+      "logits/chosen": 83.75756072998047,
+      "logits/rejected": 83.86710357666016,
+      "logps/chosen": -28.15452003479004,
+      "logps/rejected": -35.40947723388672,
+      "loss": 0.6604,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": 0.07750620692968369,
+      "rewards/margins": 0.07253950834274292,
+      "rewards/rejected": 0.004966698121279478,
       "step": 120
     },
     {
       "epoch": 0.34,
-      "grad_norm": 19.375,
+      "grad_norm": 4.71875,
       "learning_rate": 4.1940827077152755e-06,
-      "logits/chosen": 80.81904602050781,
-      "logits/rejected": 80.84658813476562,
-      "logps/chosen": -30.323780059814453,
-      "logps/rejected": -32.11851119995117,
-      "loss": 0.6101,
-      "rewards/accuracies": 0.625,
-      "rewards/chosen": 0.2334306687116623,
-      "rewards/margins": 0.3081851899623871,
-      "rewards/rejected": -0.07475452125072479,
+      "logits/chosen": 80.84075164794922,
+      "logits/rejected": 80.86768341064453,
+      "logps/chosen": -30.288599014282227,
+      "logps/rejected": -32.0759391784668,
+      "loss": 0.6608,
+      "rewards/accuracies": 0.612500011920929,
+      "rewards/chosen": 0.06187562271952629,
+      "rewards/margins": 0.0763070285320282,
+      "rewards/rejected": -0.014431402087211609,
       "step": 130
     },
     {
       "epoch": 0.36,
-      "grad_norm": 16.375,
+      "grad_norm": 5.3125,
       "learning_rate": 4.0204024186666215e-06,
-      "logits/chosen": 81.91621398925781,
-      "logits/rejected": 81.9150161743164,
-      "logps/chosen": -26.948715209960938,
-      "logps/rejected": -33.053627014160156,
-      "loss": 0.5561,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": 0.19944298267364502,
-      "rewards/margins": 0.45154237747192383,
-      "rewards/rejected": -0.2520993649959564,
+      "logits/chosen": 81.7759780883789,
+      "logits/rejected": 81.78620910644531,
+      "logps/chosen": -26.95816421508789,
+      "logps/rejected": -33.1744270324707,
+      "loss": 0.6385,
+      "rewards/accuracies": 0.6625000238418579,
+      "rewards/chosen": 0.04891597107052803,
+      "rewards/margins": 0.12402068078517914,
+      "rewards/rejected": -0.07510470598936081,
       "step": 140
     },
     {
       "epoch": 0.39,
-      "grad_norm": 17.75,
+      "grad_norm": 5.78125,
       "learning_rate": 3.834196265035119e-06,
-      "logits/chosen": 80.3634262084961,
-      "logits/rejected": 80.32890319824219,
-      "logps/chosen": -28.899206161499023,
-      "logps/rejected": -33.3048095703125,
-      "loss": 0.5375,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": 0.1959628015756607,
-      "rewards/margins": 0.49128589034080505,
-      "rewards/rejected": -0.29532313346862793,
+      "logits/chosen": 80.12449645996094,
+      "logits/rejected": 80.0939712524414,
+      "logps/chosen": -29.01108741760254,
+      "logps/rejected": -33.390743255615234,
+      "loss": 0.6406,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": 0.03780210763216019,
+      "rewards/margins": 0.12022628635168076,
+      "rewards/rejected": -0.08242417871952057,
       "step": 150
     },
     {
       "epoch": 0.42,
-      "grad_norm": 29.625,
+      "grad_norm": 6.9375,
       "learning_rate": 3.636998309800573e-06,
-      "logits/chosen": 82.13862609863281,
-      "logits/rejected": 82.1554183959961,
-      "logps/chosen": -33.77667236328125,
-      "logps/rejected": -30.610042572021484,
-      "loss": 0.6209,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": 0.14403103291988373,
-      "rewards/margins": 0.4070332646369934,
-      "rewards/rejected": -0.2630022168159485,
+      "logits/chosen": 81.77169036865234,
+      "logits/rejected": 81.79722595214844,
+      "logps/chosen": -34.082550048828125,
+      "logps/rejected": -30.868389129638672,
+      "loss": 0.653,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": 0.0054204328916966915,
+      "rewards/margins": 0.09700557589530945,
+      "rewards/rejected": -0.09158514440059662,
       "step": 160
     },
     {
       "epoch": 0.44,
-      "grad_norm": 21.375,
+      "grad_norm": 6.8125,
       "learning_rate": 3.4304331721118078e-06,
-      "logits/chosen": 82.94571685791016,
-      "logits/rejected": 82.90376281738281,
-      "logps/chosen": -30.815759658813477,
-      "logps/rejected": -32.71514129638672,
-      "loss": 0.5622,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": 0.17741146683692932,
-      "rewards/margins": 0.45544299483299255,
-      "rewards/rejected": -0.2780315577983856,
+      "logits/chosen": 82.42708587646484,
+      "logits/rejected": 82.38101959228516,
+      "logps/chosen": -30.999557495117188,
+      "logps/rejected": -33.025474548339844,
+      "loss": 0.6393,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": 0.025972897186875343,
+      "rewards/margins": 0.1265145242214203,
+      "rewards/rejected": -0.10054163634777069,
       "step": 170
     },
     {
       "epoch": 0.47,
-      "grad_norm": 17.25,
+      "grad_norm": 5.375,
       "learning_rate": 3.2162026428305436e-06,
-      "logits/chosen": 80.42796325683594,
-      "logits/rejected": 80.40777587890625,
-      "logps/chosen": -30.608139038085938,
-      "logps/rejected": -31.78671646118164,
-      "loss": 0.5453,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": 0.23560428619384766,
-      "rewards/margins": 0.48740649223327637,
-      "rewards/rejected": -0.25180214643478394,
+      "logits/chosen": 79.67488861083984,
+      "logits/rejected": 79.65385437011719,
+      "logps/chosen": -30.78965187072754,
+      "logps/rejected": -32.08345031738281,
+      "loss": 0.6346,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": 0.04075000807642937,
+      "rewards/margins": 0.1333744376897812,
+      "rewards/rejected": -0.09262441098690033,
       "step": 180
     },
     {
       "epoch": 0.49,
-      "grad_norm": 10.5,
+      "grad_norm": 3.859375,
       "learning_rate": 2.996071664294641e-06,
-      "logits/chosen": 82.06446838378906,
-      "logits/rejected": 82.06089782714844,
-      "logps/chosen": -30.304092407226562,
-      "logps/rejected": -30.887014389038086,
-      "loss": 0.6087,
-      "rewards/accuracies": 0.675000011920929,
-      "rewards/chosen": 0.1849292516708374,
-      "rewards/margins": 0.3525257706642151,
-      "rewards/rejected": -0.16759653389453888,
+      "logits/chosen": 81.18582153320312,
+      "logits/rejected": 81.16909790039062,
+      "logps/chosen": -30.42611312866211,
+      "logps/rejected": -31.0355224609375,
+      "loss": 0.6585,
+      "rewards/accuracies": 0.625,
+      "rewards/chosen": 0.03403033688664436,
+      "rewards/margins": 0.09077958762645721,
+      "rewards/rejected": -0.056749243289232254,
       "step": 190
     },
     {
       "epoch": 0.52,
-      "grad_norm": 14.75,
+      "grad_norm": 5.75,
       "learning_rate": 2.7718537898066833e-06,
-      "logits/chosen": 77.5021743774414,
-      "logits/rejected": 77.44898223876953,
-      "logps/chosen": -33.84641647338867,
-      "logps/rejected": -32.781463623046875,
-      "loss": 0.6015,
-      "rewards/accuracies": 0.6625000238418579,
-      "rewards/chosen": 0.36252856254577637,
-      "rewards/margins": 0.477752149105072,
-      "rewards/rejected": -0.11522357165813446,
+      "logits/chosen": 76.43097686767578,
+      "logits/rejected": 76.38182067871094,
+      "logps/chosen": -34.055965423583984,
+      "logps/rejected": -33.20988082885742,
+      "loss": 0.6355,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/chosen": 0.06967721879482269,
+      "rewards/margins": 0.1413251906633377,
+      "rewards/rejected": -0.07164796441793442,
       "step": 200
     },
     {
       "epoch": 0.52,
-      "eval_logits/chosen": 98.46090698242188,
-      "eval_logits/rejected": 98.43693542480469,
-      "eval_logps/chosen": -32.81344985961914,
-      "eval_logps/rejected": -36.51091384887695,
-      "eval_loss": 0.6983966827392578,
-      "eval_rewards/accuracies": 0.5402824282646179,
-      "eval_rewards/chosen": -0.14810949563980103,
-      "eval_rewards/margins": 0.0696462020277977,
-      "eval_rewards/rejected": -0.21775569021701813,
-      "eval_runtime": 104.0141,
-      "eval_samples_per_second": 3.298,
-      "eval_steps_per_second": 0.413,
+      "eval_logits/chosen": 98.12921142578125,
+      "eval_logits/rejected": 98.10821533203125,
+      "eval_logps/chosen": -32.90422058105469,
+      "eval_logps/rejected": -36.77000427246094,
+      "eval_loss": 0.6793302893638611,
+      "eval_rewards/accuracies": 0.5834717750549316,
+      "eval_rewards/chosen": -0.04610438272356987,
+      "eval_rewards/margins": 0.03424324095249176,
+      "eval_rewards/rejected": -0.08034762740135193,
+      "eval_runtime": 103.9445,
+      "eval_samples_per_second": 3.3,
+      "eval_steps_per_second": 0.414,
       "step": 200
     },
     {
       "epoch": 0.55,
-      "grad_norm": 30.375,
+      "grad_norm": 8.4375,
       "learning_rate": 2.5453962426402006e-06,
-      "logits/chosen": 80.12822723388672,
-      "logits/rejected": 80.03518676757812,
-      "logps/chosen": -33.25244903564453,
-      "logps/rejected": -35.43198013305664,
-      "loss": 0.5658,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": 0.27730754017829895,
-      "rewards/margins": 0.4556616246700287,
-      "rewards/rejected": -0.17835409939289093,
+      "logits/chosen": 79.00591278076172,
+      "logits/rejected": 78.91060638427734,
+      "logps/chosen": -33.55596160888672,
+      "logps/rejected": -35.893375396728516,
+      "loss": 0.6418,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": 0.03897558152675629,
+      "rewards/margins": 0.12970347702503204,
+      "rewards/rejected": -0.09072789549827576,
       "step": 210
     },
     {
       "epoch": 0.57,
-      "grad_norm": 17.625,
+      "grad_norm": 6.46875,
       "learning_rate": 2.3185646976551794e-06,
-      "logits/chosen": 82.24729919433594,
-      "logits/rejected": 82.3295669555664,
-      "logps/chosen": -31.125244140625,
-      "logps/rejected": -31.354183197021484,
-      "loss": 0.5216,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": 0.3294834792613983,
-      "rewards/margins": 0.57957524061203,
-      "rewards/rejected": -0.2500917315483093,
+      "logits/chosen": 81.00145721435547,
+      "logits/rejected": 81.09220886230469,
+      "logps/chosen": -31.351266860961914,
+      "logps/rejected": -31.709590911865234,
+      "loss": 0.6274,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": 0.0597686693072319,
+      "rewards/margins": 0.15783223509788513,
+      "rewards/rejected": -0.09806357324123383,
       "step": 220
     },
     {
       "epoch": 0.6,
-      "grad_norm": 21.125,
+      "grad_norm": 6.25,
       "learning_rate": 2.0932279108998323e-06,
-      "logits/chosen": 79.34232330322266,
-      "logits/rejected": 79.39698791503906,
-      "logps/chosen": -32.36313247680664,
-      "logps/rejected": -34.425838470458984,
-      "loss": 0.618,
+      "logits/chosen": 78.02706146240234,
+      "logits/rejected": 78.08647155761719,
+      "logps/chosen": -32.611751556396484,
+      "logps/rejected": -34.99976348876953,
+      "loss": 0.6457,
       "rewards/accuracies": 0.6625000238418579,
-      "rewards/chosen": 0.17704486846923828,
-      "rewards/margins": 0.34251970052719116,
-      "rewards/rejected": -0.16547484695911407,
+      "rewards/chosen": 0.019399229437112808,
+      "rewards/margins": 0.118160679936409,
+      "rewards/rejected": -0.09876143932342529,
       "step": 230
     },
     {
       "epoch": 0.62,
-      "grad_norm": 22.5,
+      "grad_norm": 6.8125,
       "learning_rate": 1.8712423238279358e-06,
-      "logits/chosen": 81.70603942871094,
-      "logits/rejected": 81.9966049194336,
-      "logps/chosen": -30.78457260131836,
-      "logps/rejected": -32.03525161743164,
-      "loss": 0.5166,
+      "logits/chosen": 80.3001708984375,
+      "logits/rejected": 80.60163879394531,
+      "logps/chosen": -31.092309951782227,
+      "logps/rejected": -32.644691467285156,
+      "loss": 0.6182,
       "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": 0.3613971471786499,
-      "rewards/margins": 0.5626804232597351,
-      "rewards/rejected": -0.20128324627876282,
+      "rewards/chosen": 0.059575408697128296,
+      "rewards/margins": 0.17084039747714996,
+      "rewards/rejected": -0.11126496642827988,
       "step": 240
     },
     {
       "epoch": 0.65,
-      "grad_norm": 23.375,
+      "grad_norm": 7.0,
       "learning_rate": 1.6544367689701824e-06,
-      "logits/chosen": 80.38089752197266,
-      "logits/rejected": 80.44111633300781,
-      "logps/chosen": -26.974994659423828,
-      "logps/rejected": -30.13335609436035,
-      "loss": 0.6427,
+      "logits/chosen": 78.8541030883789,
+      "logits/rejected": 78.9134750366211,
+      "logps/chosen": -27.58013343811035,
+      "logps/rejected": -30.87213706970215,
+      "loss": 0.6545,
       "rewards/accuracies": 0.6000000238418579,
-      "rewards/chosen": 0.24836082756519318,
-      "rewards/margins": 0.34140732884407043,
-      "rewards/rejected": -0.09304650872945786,
+      "rewards/chosen": 0.001576287322677672,
+      "rewards/margins": 0.09871624410152435,
+      "rewards/rejected": -0.09713996946811676,
       "step": 250
     },
     {
       "epoch": 0.68,
-      "grad_norm": 20.125,
+      "grad_norm": 6.09375,
       "learning_rate": 1.4445974030621963e-06,
-      "logits/chosen": 77.59573364257812,
-      "logits/rejected": 77.74366760253906,
-      "logps/chosen": -30.3758487701416,
-      "logps/rejected": -36.788978576660156,
-      "loss": 0.4885,
-      "rewards/accuracies": 0.7749999761581421,
-      "rewards/chosen": 0.4740586280822754,
-      "rewards/margins": 0.749001145362854,
-      "rewards/rejected": -0.2749424874782562,
+      "logits/chosen": 75.93869018554688,
+      "logits/rejected": 76.0858154296875,
+      "logps/chosen": -30.762155532836914,
+      "logps/rejected": -37.70819091796875,
+      "loss": 0.5957,
+      "rewards/accuracies": 0.800000011920929,
+      "rewards/chosen": 0.07988408207893372,
+      "rewards/margins": 0.24054038524627686,
+      "rewards/rejected": -0.16065633296966553,
       "step": 260
     },
     {
       "epoch": 0.7,
-      "grad_norm": 14.875,
+      "grad_norm": 5.53125,
       "learning_rate": 1.243452991757889e-06,
-      "logits/chosen": 76.95594787597656,
-      "logits/rejected": 76.9817886352539,
-      "logps/chosen": -30.940093994140625,
-      "logps/rejected": -31.99422264099121,
-      "loss": 0.5356,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": 0.3355324864387512,
-      "rewards/margins": 0.5310525894165039,
-      "rewards/rejected": -0.19552013278007507,
+      "logits/chosen": 75.10862731933594,
+      "logits/rejected": 75.13311004638672,
+      "logps/chosen": -31.499658584594727,
+      "logps/rejected": -32.65572738647461,
+      "loss": 0.6335,
+      "rewards/accuracies": 0.7250000238418579,
+      "rewards/chosen": 0.027926713228225708,
+      "rewards/margins": 0.14295729994773865,
+      "rewards/rejected": -0.11503058671951294,
       "step": 270
     },
     {
       "epoch": 0.73,
-      "grad_norm": 22.375,
+      "grad_norm": 8.9375,
       "learning_rate": 1.0526606671603523e-06,
-      "logits/chosen": 79.73005676269531,
-      "logits/rejected": 79.521484375,
-      "logps/chosen": -31.06637954711914,
-      "logps/rejected": -30.05121421813965,
-      "loss": 0.5946,
-      "rewards/accuracies": 0.762499988079071,
-      "rewards/chosen": 0.2974059581756592,
-      "rewards/margins": 0.4510935842990875,
-      "rewards/rejected": -0.15368762612342834,
+      "logits/chosen": 77.99885559082031,
+      "logits/rejected": 77.7794418334961,
+      "logps/chosen": -31.871845245361328,
+      "logps/rejected": -30.71683692932129,
+      "loss": 0.657,
+      "rewards/accuracies": 0.637499988079071,
+      "rewards/chosen": -0.0061946348287165165,
+      "rewards/margins": 0.09878946840763092,
+      "rewards/rejected": -0.10498411953449249,
       "step": 280
     },
     {
       "epoch": 0.75,
-      "grad_norm": 18.125,
+      "grad_norm": 5.8125,
       "learning_rate": 8.737922755071455e-07,
-      "logits/chosen": 79.8319091796875,
-      "logits/rejected": 79.74632263183594,
-      "logps/chosen": -33.005733489990234,
-      "logps/rejected": -32.86864471435547,
-      "loss": 0.4772,
-      "rewards/accuracies": 0.7250000238418579,
-      "rewards/chosen": 0.43940457701683044,
-      "rewards/margins": 0.780119001865387,
-      "rewards/rejected": -0.3407144248485565,
+      "logits/chosen": 78.05221557617188,
+      "logits/rejected": 77.97148895263672,
+      "logps/chosen": -33.65934371948242,
+      "logps/rejected": -33.7969856262207,
+      "loss": 0.6008,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": 0.044489990919828415,
+      "rewards/margins": 0.22250187397003174,
+      "rewards/rejected": -0.17801189422607422,
       "step": 290
     },
     {
       "epoch": 0.78,
-      "grad_norm": 23.375,
+      "grad_norm": 6.6875,
       "learning_rate": 7.08321427484816e-07,
-      "logits/chosen": 75.52354431152344,
-      "logits/rejected": 75.6214599609375,
-      "logps/chosen": -32.215476989746094,
-      "logps/rejected": -29.42368507385254,
-      "loss": 0.5603,
-      "rewards/accuracies": 0.699999988079071,
-      "rewards/chosen": 0.47609180212020874,
-      "rewards/margins": 0.6047566533088684,
-      "rewards/rejected": -0.1286647617816925,
+      "logits/chosen": 73.7139892578125,
+      "logits/rejected": 73.8324203491211,
+      "logps/chosen": -32.73633575439453,
+      "logps/rejected": -30.082836151123047,
+      "loss": 0.6306,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": 0.06693719327449799,
+      "rewards/margins": 0.16501857340335846,
+      "rewards/rejected": -0.09808137267827988,
       "step": 300
     },
     {
       "epoch": 0.78,
-      "eval_logits/chosen": 98.4832534790039,
-      "eval_logits/rejected": 98.45703125,
-      "eval_logps/chosen": -32.67014694213867,
-      "eval_logps/rejected": -36.31406784057617,
-      "eval_loss": 0.7083631753921509,
-      "eval_rewards/accuracies": 0.5398671627044678,
-      "eval_rewards/chosen": -0.09078802913427353,
-      "eval_rewards/margins": 0.04822726547718048,
-      "eval_rewards/rejected": -0.1390153020620346,
-      "eval_runtime": 103.8446,
-      "eval_samples_per_second": 3.303,
+      "eval_logits/chosen": 97.9847412109375,
+      "eval_logits/rejected": 97.95744323730469,
+      "eval_logps/chosen": -33.00377655029297,
+      "eval_logps/rejected": -37.00663375854492,
+      "eval_loss": 0.6733371615409851,
+      "eval_rewards/accuracies": 0.6121262311935425,
+      "eval_rewards/chosen": -0.056060198694467545,
+      "eval_rewards/margins": 0.047950610518455505,
+      "eval_rewards/rejected": -0.10401081293821335,
+      "eval_runtime": 103.8262,
+      "eval_samples_per_second": 3.304,
       "eval_steps_per_second": 0.414,
       "step": 300
     },
     {
       "epoch": 0.81,
-      "grad_norm": 17.375,
+      "grad_norm": 5.65625,
       "learning_rate": 5.576113578589035e-07,
-      "logits/chosen": 82.69928741455078,
-      "logits/rejected": 82.7286376953125,
-      "logps/chosen": -30.082874298095703,
-      "logps/rejected": -32.789588928222656,
-      "loss": 0.5344,
-      "rewards/accuracies": 0.7875000238418579,
-      "rewards/chosen": 0.3122648298740387,
-      "rewards/margins": 0.5481952428817749,
-      "rewards/rejected": -0.2359304428100586,
+      "logits/chosen": 81.07884216308594,
+      "logits/rejected": 81.09075164794922,
+      "logps/chosen": -30.61749267578125,
+      "logps/rejected": -33.596153259277344,
+      "loss": 0.6246,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": 0.02460428513586521,
+      "rewards/margins": 0.1642439067363739,
+      "rewards/rejected": -0.13963960111141205,
       "step": 310
     },
     {
       "epoch": 0.83,
-      "grad_norm": 17.75,
+      "grad_norm": 6.25,
       "learning_rate": 4.229036944380913e-07,
-      "logits/chosen": 80.08171081542969,
-      "logits/rejected": 80.08326721191406,
-      "logps/chosen": -30.4410457611084,
-      "logps/rejected": -29.53485107421875,
-      "loss": 0.4679,
-      "rewards/accuracies": 0.800000011920929,
-      "rewards/chosen": 0.4630066454410553,
-      "rewards/margins": 0.7091516852378845,
-      "rewards/rejected": -0.24614505469799042,
+      "logits/chosen": 78.28865814208984,
+      "logits/rejected": 78.29974365234375,
+      "logps/chosen": -30.886798858642578,
+      "logps/rejected": -30.064910888671875,
+      "loss": 0.6141,
+      "rewards/accuracies": 0.7124999761581421,
+      "rewards/chosen": 0.07117630541324615,
+      "rewards/margins": 0.18571846187114716,
+      "rewards/rejected": -0.1145421713590622,
       "step": 320
     },
     {
       "epoch": 0.86,
-      "grad_norm": 14.6875,
+      "grad_norm": 6.625,
       "learning_rate": 3.053082288996112e-07,
-      "logits/chosen": 77.24263000488281,
-      "logits/rejected": 77.28250122070312,
-      "logps/chosen": -28.96456527709961,
-      "logps/rejected": -33.16306686401367,
-      "loss": 0.4741,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": 0.5552161931991577,
-      "rewards/margins": 0.7664985656738281,
-      "rewards/rejected": -0.21128229796886444,
+      "logits/chosen": 75.30354309082031,
+      "logits/rejected": 75.34764099121094,
+      "logps/chosen": -29.404388427734375,
+      "logps/rejected": -34.069969177246094,
+      "loss": 0.5902,
+      "rewards/accuracies": 0.762499988079071,
+      "rewards/chosen": 0.09482161700725555,
+      "rewards/margins": 0.23833179473876953,
+      "rewards/rejected": -0.14351019263267517,
       "step": 330
     },
     {
       "epoch": 0.88,
-      "grad_norm": 29.875,
+      "grad_norm": 8.1875,
       "learning_rate": 2.0579377374915805e-07,
-      "logits/chosen": 81.6305923461914,
-      "logits/rejected": 81.66109466552734,
-      "logps/chosen": -32.311283111572266,
-      "logps/rejected": -34.081214904785156,
-      "loss": 0.5334,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": 0.3603467345237732,
-      "rewards/margins": 0.6453540921211243,
-      "rewards/rejected": -0.2850072979927063,
+      "logits/chosen": 79.95478820800781,
+      "logits/rejected": 79.99383544921875,
+      "logps/chosen": -32.775882720947266,
+      "logps/rejected": -35.080204010009766,
+      "loss": 0.6075,
+      "rewards/accuracies": 0.7749999761581421,
+      "rewards/chosen": 0.043626897037029266,
+      "rewards/margins": 0.21477791666984558,
+      "rewards/rejected": -0.17115101218223572,
       "step": 340
     },
     {
       "epoch": 0.91,
-      "grad_norm": 11.625,
+      "grad_norm": 6.0625,
       "learning_rate": 1.2518018074041684e-07,
-      "logits/chosen": 80.61808776855469,
-      "logits/rejected": 80.62725067138672,
-      "logps/chosen": -32.49099349975586,
-      "logps/rejected": -33.57762908935547,
-      "loss": 0.5483,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": 0.4610978960990906,
-      "rewards/margins": 0.6554309129714966,
-      "rewards/rejected": -0.19433295726776123,
+      "logits/chosen": 78.82893371582031,
+      "logits/rejected": 78.83858489990234,
+      "logps/chosen": -32.75529479980469,
+      "logps/rejected": -34.289024353027344,
+      "loss": 0.6093,
+      "rewards/accuracies": 0.699999988079071,
+      "rewards/chosen": 0.08884432911872864,
+      "rewards/margins": 0.20856666564941406,
+      "rewards/rejected": -0.11972232908010483,
       "step": 350
     },
     {
       "epoch": 0.94,
-      "grad_norm": 15.8125,
+      "grad_norm": 5.625,
       "learning_rate": 6.41315865106129e-08,
-      "logits/chosen": 82.13656616210938,
-      "logits/rejected": 82.17839050292969,
-      "logps/chosen": -28.390300750732422,
-      "logps/rejected": -31.930139541625977,
-      "loss": 0.5259,
-      "rewards/accuracies": 0.7124999761581421,
-      "rewards/chosen": 0.46476393938064575,
-      "rewards/margins": 0.5768795013427734,
-      "rewards/rejected": -0.11211560666561127,
+      "logits/chosen": 80.45689392089844,
+      "logits/rejected": 80.4975357055664,
+      "logps/chosen": -28.772647857666016,
+      "logps/rejected": -32.65345764160156,
+      "loss": 0.619,
+      "rewards/accuracies": 0.675000011920929,
+      "rewards/chosen": 0.07795627415180206,
+      "rewards/margins": 0.17831732332706451,
+      "rewards/rejected": -0.10036104917526245,
       "step": 360
     },
     {
       "epoch": 0.96,
-      "grad_norm": 20.375,
+      "grad_norm": 7.0,
       "learning_rate": 2.3150941078050325e-08,
-      "logits/chosen": 81.61884307861328,
-      "logits/rejected": 81.6422348022461,
-      "logps/chosen": -31.991628646850586,
-      "logps/rejected": -35.56809616088867,
-      "loss": 0.5548,
-      "rewards/accuracies": 0.6499999761581421,
-      "rewards/chosen": 0.3538144528865814,
-      "rewards/margins": 0.5373049974441528,
-      "rewards/rejected": -0.1834905445575714,
+      "logits/chosen": 79.93882751464844,
+      "logits/rejected": 79.95525360107422,
+      "logps/chosen": -32.576927185058594,
+      "logps/rejected": -36.53556442260742,
+      "loss": 0.6267,
+      "rewards/accuracies": 0.6875,
+      "rewards/chosen": 0.02992354705929756,
+      "rewards/margins": 0.1725427806377411,
+      "rewards/rejected": -0.14261923730373383,
       "step": 370
     },
     {
       "epoch": 0.99,
-      "grad_norm": 21.375,
+      "grad_norm": 5.96875,
       "learning_rate": 2.575864278703266e-09,
-      "logits/chosen": 75.43504333496094,
-      "logits/rejected": 75.3118667602539,
-      "logps/chosen": -29.84956932067871,
-      "logps/rejected": -28.521799087524414,
-      "loss": 0.5873,
-      "rewards/accuracies": 0.6499999761581421,
-      "rewards/chosen": 0.29062420129776,
-      "rewards/margins": 0.4351147711277008,
-      "rewards/rejected": -0.1444905549287796,
+      "logits/chosen": 73.57047271728516,
+      "logits/rejected": 73.44520568847656,
+      "logps/chosen": -30.454349517822266,
+      "logps/rejected": -29.29287338256836,
+      "loss": 0.6434,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": 0.012177868746221066,
+      "rewards/margins": 0.12540772557258606,
+      "rewards/rejected": -0.11322984844446182,
       "step": 380
     },
     {
       "epoch": 1.0,
       "step": 385,
       "total_flos": 0.0,
-      "train_loss": 0.5890550520512965,
-      "train_runtime": 2558.5725,
-      "train_samples_per_second": 1.203,
-      "train_steps_per_second": 0.15
+      "train_loss": 0.6461187455561254,
+      "train_runtime": 2556.0398,
+      "train_samples_per_second": 1.205,
+      "train_steps_per_second": 0.151
     }
   ],
   "logging_steps": 10,