Model save

README.md

@@ -13,7 +13,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # zephyr-7b-dpo-full
 
-This model was trained from scratch on an unknown dataset.
+This model was trained from scratch on the None dataset.
 
 ## Model description
 
@@ -44,7 +44,7 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: cosine
 - lr_scheduler_warmup_ratio: 0.1
-- num_epochs: 2
+- num_epochs: 1
 
 ### Training results
 
@@ -54,5 +54,5 @@ The following hyperparameters were used during training:
 
 - Transformers 4.39.3
 - Pytorch 2.1.2+cu118
-- Datasets 2.19.1
+- Datasets 2.16.1
 - Tokenizers 0.15.2

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.99,
-    "train_loss": 0.5661277400923299,
-    "train_runtime": 10889.322,
-    "train_samples": 46672,
-    "train_samples_per_second": 8.572,
-    "train_steps_per_second": 0.033
+    "epoch": 1.0,
+    "train_loss": 0.7017018035803427,
+    "train_runtime": 5276.0496,
+    "train_samples": 46008,
+    "train_samples_per_second": 8.72,
+    "train_steps_per_second": 0.034
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:272835aec6a939eddb8f3c6761a09722f8293bf2e2879f737a53fca1bdafb26a
+oid sha256:78023790d283daa5ac216bca5118f7f16d3456d271b829d8d4c420bb806d2361
 size 4943162336

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:42495f2a7c8291991dc1126b3da9fe9851e33187e0aaaeef6cdd654ba5d4a54c
+oid sha256:46c6d2f9dc71af031bc8877acf06f617c43d4c9a595373f147771924b45f71a4
 size 4999819336

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:28778bd034ca6343f4ed1bdeedff7d601a33547dcaa0c4f056c9501d69a1ad0c
+oid sha256:d162b61250d806211784d7670553be059e9804860cd77fea33f82fe78db8eca4
 size 4540516344

runs/Jun19_01-19-06_n136-129-074/events.out.tfevents.1718731170.n136-129-074.2649948.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c70c87c1c2128491faecea7c3ca5937aa650e257bd092a1a43c2bce5981d83b7
-size 12372
+oid sha256:d3254c513bc67260bbf92cc3f5bcc62c9647dcc86c28afb97e596aff7d389083
+size 17518

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 1.99,
-    "train_loss": 0.5661277400923299,
-    "train_runtime": 10889.322,
-    "train_samples": 46672,
-    "train_samples_per_second": 8.572,
-    "train_steps_per_second": 0.033
+    "epoch": 1.0,
+    "train_loss": 0.7017018035803427,
+    "train_runtime": 5276.0496,
+    "train_samples": 46008,
+    "train_samples_per_second": 8.72,
+    "train_steps_per_second": 0.034
 }

trainer_state.json

@@ -1,21 +1,21 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 1.9945205479452055,
+  "epoch": 0.9958275382475661,
   "eval_steps": 500,
-  "global_step": 364,
+  "global_step": 179,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.01,
-      "grad_norm": 749.8044037682902,
-      "learning_rate": 2.702702702702703e-10,
-      "logits/chosen": -0.7030794620513916,
-      "logits/rejected": -0.3951629400253296,
-      "logps/chosen": -341.73382568359375,
-      "logps/rejected": -292.9862060546875,
+      "grad_norm": 265.8884735732537,
+      "learning_rate": 5.555555555555555e-10,
+      "logits/chosen": -1.145136833190918,
+      "logits/rejected": -0.7638764977455139,
+      "logps/chosen": -369.05181884765625,
+      "logps/rejected": -372.01507568359375,
       "loss": 0.6931,
       "rewards/accuracies": 0.0,
       "rewards/chosen": 0.0,
@@ -24,559 +24,274 @@
       "step": 1
     },
     {
-      "epoch": 0.05,
-      "grad_norm": 728.4025747943551,
-      "learning_rate": 2.702702702702703e-09,
-      "logits/chosen": -0.8550878167152405,
-      "logits/rejected": -0.5013476610183716,
-      "logps/chosen": -343.0583801269531,
-      "logps/rejected": -297.6263427734375,
-      "loss": 0.7129,
-      "rewards/accuracies": 0.40625,
-      "rewards/chosen": 0.02346261963248253,
-      "rewards/margins": -0.025641746819019318,
-      "rewards/rejected": 0.049104366451501846,
+      "epoch": 0.06,
+      "grad_norm": 336.072912015809,
+      "learning_rate": 5.555555555555556e-09,
+      "logits/chosen": -1.0928783416748047,
+      "logits/rejected": -0.9423567652702332,
+      "logps/chosen": -372.36029052734375,
+      "logps/rejected": -339.4759826660156,
+      "loss": 0.7082,
+      "rewards/accuracies": 0.4270833432674408,
+      "rewards/chosen": -0.016076795756816864,
+      "rewards/margins": -0.023205766454339027,
+      "rewards/rejected": 0.007128970697522163,
       "step": 10
     },
     {
       "epoch": 0.11,
-      "grad_norm": 769.1100121309379,
-      "learning_rate": 5.405405405405406e-09,
-      "logits/chosen": -0.89165860414505,
-      "logits/rejected": -0.5214177966117859,
-      "logps/chosen": -333.6175231933594,
-      "logps/rejected": -287.69482421875,
-      "loss": 0.7088,
-      "rewards/accuracies": 0.512499988079071,
-      "rewards/chosen": -0.0008101940038613975,
-      "rewards/margins": 0.021183893084526062,
-      "rewards/rejected": -0.02199408784508705,
+      "grad_norm": 322.5517219745342,
+      "learning_rate": 9.99619291237835e-09,
+      "logits/chosen": -0.8586678504943848,
+      "logits/rejected": -0.733450710773468,
+      "logps/chosen": -365.1810302734375,
+      "logps/rejected": -340.0671081542969,
+      "loss": 0.7124,
+      "rewards/accuracies": 0.48124998807907104,
+      "rewards/chosen": -0.01668541505932808,
+      "rewards/margins": -0.02395111322402954,
+      "rewards/rejected": 0.007265699096024036,
       "step": 20
     },
     {
-      "epoch": 0.16,
-      "grad_norm": 782.2455729371497,
-      "learning_rate": 8.108108108108109e-09,
-      "logits/chosen": -0.9365941882133484,
-      "logits/rejected": -0.5470438599586487,
-      "logps/chosen": -332.33966064453125,
-      "logps/rejected": -289.5589904785156,
-      "loss": 0.7083,
-      "rewards/accuracies": 0.5,
-      "rewards/chosen": -0.015217331238090992,
-      "rewards/margins": -0.010174614377319813,
-      "rewards/rejected": -0.005042716860771179,
+      "epoch": 0.17,
+      "grad_norm": 286.6512881766183,
+      "learning_rate": 9.863552602006434e-09,
+      "logits/chosen": -1.1635475158691406,
+      "logits/rejected": -0.8620313405990601,
+      "logps/chosen": -386.59136962890625,
+      "logps/rejected": -348.4241027832031,
+      "loss": 0.7091,
+      "rewards/accuracies": 0.5531250238418579,
+      "rewards/chosen": 0.008167916908860207,
+      "rewards/margins": 0.05863186717033386,
+      "rewards/rejected": -0.050463952124118805,
       "step": 30
     },
     {
       "epoch": 0.22,
-      "grad_norm": 736.3539975893915,
-      "learning_rate": 9.997923381619256e-09,
-      "logits/chosen": -0.9017000198364258,
-      "logits/rejected": -0.555226743221283,
-      "logps/chosen": -337.62115478515625,
-      "logps/rejected": -306.5302429199219,
-      "loss": 0.7072,
-      "rewards/accuracies": 0.5687500238418579,
-      "rewards/chosen": 0.05695560574531555,
-      "rewards/margins": 0.08540164679288864,
-      "rewards/rejected": -0.02844604291021824,
+      "grad_norm": 251.72489049021175,
+      "learning_rate": 9.546315917055362e-09,
+      "logits/chosen": -1.0636483430862427,
+      "logits/rejected": -0.8499566912651062,
+      "logps/chosen": -362.8810119628906,
+      "logps/rejected": -332.20843505859375,
+      "loss": 0.7101,
+      "rewards/accuracies": 0.48124998807907104,
+      "rewards/chosen": -0.03652422875165939,
+      "rewards/margins": -0.019151169806718826,
+      "rewards/rejected": -0.017373060807585716,
       "step": 40
     },
     {
-      "epoch": 0.27,
-      "grad_norm": 730.7089871756754,
-      "learning_rate": 9.961053687802851e-09,
-      "logits/chosen": -0.9294391870498657,
-      "logits/rejected": -0.5007705688476562,
-      "logps/chosen": -333.2597961425781,
-      "logps/rejected": -295.4553527832031,
-      "loss": 0.6941,
-      "rewards/accuracies": 0.53125,
-      "rewards/chosen": -0.01942090317606926,
-      "rewards/margins": 0.017053481191396713,
-      "rewards/rejected": -0.03647438436746597,
+      "epoch": 0.28,
+      "grad_norm": 296.71671546416985,
+      "learning_rate": 9.056523591268063e-09,
+      "logits/chosen": -1.0690479278564453,
+      "logits/rejected": -0.8786741495132446,
+      "logps/chosen": -377.68011474609375,
+      "logps/rejected": -340.79656982421875,
+      "loss": 0.7089,
+      "rewards/accuracies": 0.5249999761581421,
+      "rewards/chosen": -0.00048580169095657766,
+      "rewards/margins": 0.035611577332019806,
+      "rewards/rejected": -0.03609737753868103,
       "step": 50
     },
     {
       "epoch": 0.33,
-      "grad_norm": 730.7067078437899,
-      "learning_rate": 9.878428410862483e-09,
-      "logits/chosen": -0.8303865194320679,
-      "logits/rejected": -0.45489081740379333,
-      "logps/chosen": -331.8045349121094,
-      "logps/rejected": -295.13055419921875,
-      "loss": 0.6832,
-      "rewards/accuracies": 0.6031249761581421,
-      "rewards/chosen": -0.029834549874067307,
-      "rewards/margins": 0.05674406886100769,
-      "rewards/rejected": -0.0865786224603653,
+      "grad_norm": 244.469289636431,
+      "learning_rate": 8.41276571609327e-09,
+      "logits/chosen": -1.0523345470428467,
+      "logits/rejected": -0.8627673983573914,
+      "logps/chosen": -368.83758544921875,
+      "logps/rejected": -334.1421203613281,
+      "loss": 0.7039,
+      "rewards/accuracies": 0.574999988079071,
+      "rewards/chosen": 0.014057522639632225,
+      "rewards/margins": 0.05172234773635864,
+      "rewards/rejected": -0.03766482323408127,
       "step": 60
     },
     {
-      "epoch": 0.38,
-      "grad_norm": 674.6624722402527,
-      "learning_rate": 9.750809600145954e-09,
-      "logits/chosen": -1.0535982847213745,
-      "logits/rejected": -0.6284598112106323,
-      "logps/chosen": -351.1028747558594,
-      "logps/rejected": -304.6434631347656,
-      "loss": 0.6804,
-      "rewards/accuracies": 0.625,
-      "rewards/chosen": 0.0078976359218359,
-      "rewards/margins": 0.11121608316898346,
-      "rewards/rejected": -0.10331843793392181,
+      "epoch": 0.39,
+      "grad_norm": 218.81492341857816,
+      "learning_rate": 7.639476152864163e-09,
+      "logits/chosen": -1.0656211376190186,
+      "logits/rejected": -0.9337635040283203,
+      "logps/chosen": -361.3856506347656,
+      "logps/rejected": -338.86590576171875,
+      "loss": 0.7048,
+      "rewards/accuracies": 0.559374988079071,
+      "rewards/chosen": 0.010572463274002075,
+      "rewards/margins": 0.05579069256782532,
+      "rewards/rejected": -0.04521822929382324,
       "step": 70
     },
     {
-      "epoch": 0.44,
-      "grad_norm": 640.6205441818687,
-      "learning_rate": 9.579374278412817e-09,
-      "logits/chosen": -0.9343639612197876,
-      "logits/rejected": -0.3806554079055786,
-      "logps/chosen": -344.04443359375,
-      "logps/rejected": -294.193359375,
-      "loss": 0.6487,
-      "rewards/accuracies": 0.659375011920929,
-      "rewards/chosen": -0.01617306098341942,
-      "rewards/margins": 0.14446857571601868,
-      "rewards/rejected": -0.1606416255235672,
+      "epoch": 0.45,
+      "grad_norm": 233.90515893411515,
+      "learning_rate": 6.766005144407989e-09,
+      "logits/chosen": -0.9778022766113281,
+      "logits/rejected": -0.7526016235351562,
+      "logps/chosen": -369.4658508300781,
+      "logps/rejected": -332.49652099609375,
+      "loss": 0.6986,
+      "rewards/accuracies": 0.550000011920929,
+      "rewards/chosen": 0.050749681890010834,
+      "rewards/margins": 0.07469334453344345,
+      "rewards/rejected": -0.023943662643432617,
       "step": 80
     },
     {
-      "epoch": 0.49,
-      "grad_norm": 629.6802089882239,
-      "learning_rate": 9.365703586204494e-09,
-      "logits/chosen": -1.0297319889068604,
-      "logits/rejected": -0.654572606086731,
-      "logps/chosen": -340.8974609375,
-      "logps/rejected": -307.7338562011719,
-      "loss": 0.639,
-      "rewards/accuracies": 0.621874988079071,
-      "rewards/chosen": 0.009362801909446716,
-      "rewards/margins": 0.15911515057086945,
-      "rewards/rejected": -0.14975234866142273,
+      "epoch": 0.5,
+      "grad_norm": 236.56099599203856,
+      "learning_rate": 5.825505325157961e-09,
+      "logits/chosen": -0.9711936712265015,
+      "logits/rejected": -0.7899622917175293,
+      "logps/chosen": -360.23504638671875,
+      "logps/rejected": -336.3778076171875,
+      "loss": 0.7004,
+      "rewards/accuracies": 0.5874999761581421,
+      "rewards/chosen": 0.045762062072753906,
+      "rewards/margins": 0.08318910002708435,
+      "rewards/rejected": -0.037427037954330444,
       "step": 90
     },
     {
-      "epoch": 0.55,
-      "grad_norm": 654.9042346572915,
-      "learning_rate": 9.111768199053588e-09,
-      "logits/chosen": -0.8516262173652649,
-      "logits/rejected": -0.5497715473175049,
-      "logps/chosen": -318.9173889160156,
-      "logps/rejected": -284.2920837402344,
-      "loss": 0.6303,
-      "rewards/accuracies": 0.609375,
-      "rewards/chosen": -0.07476776093244553,
-      "rewards/margins": 0.15329131484031677,
-      "rewards/rejected": -0.2280590981245041,
+      "epoch": 0.56,
+      "grad_norm": 243.70225224613213,
+      "learning_rate": 4.853673411307564e-09,
+      "logits/chosen": -1.0938254594802856,
+      "logits/rejected": -0.9187259674072266,
+      "logps/chosen": -381.3570861816406,
+      "logps/rejected": -355.38067626953125,
+      "loss": 0.6995,
+      "rewards/accuracies": 0.590624988079071,
+      "rewards/chosen": 0.049023449420928955,
+      "rewards/margins": 0.06866957992315292,
+      "rewards/rejected": -0.019646134227514267,
       "step": 100
     },
     {
-      "epoch": 0.6,
-      "grad_norm": 568.2894000484466,
-      "learning_rate": 8.819910152028871e-09,
-      "logits/chosen": -0.7658065557479858,
-      "logits/rejected": -0.4776793122291565,
-      "logps/chosen": -335.7439270019531,
-      "logps/rejected": -302.370849609375,
-      "loss": 0.6175,
-      "rewards/accuracies": 0.640625,
-      "rewards/chosen": 0.008360857143998146,
-      "rewards/margins": 0.20505543053150177,
-      "rewards/rejected": -0.19669455289840698,
+      "epoch": 0.61,
+      "grad_norm": 260.62196642098917,
+      "learning_rate": 3.887395330218429e-09,
+      "logits/chosen": -1.0548704862594604,
+      "logits/rejected": -0.8432992696762085,
+      "logps/chosen": -379.90887451171875,
+      "logps/rejected": -349.34161376953125,
+      "loss": 0.6973,
+      "rewards/accuracies": 0.6343749761581421,
+      "rewards/chosen": 0.11532249301671982,
+      "rewards/margins": 0.1085214838385582,
+      "rewards/rejected": 0.006801015231758356,
       "step": 110
     },
     {
-      "epoch": 0.66,
-      "grad_norm": 569.4638082407581,
-      "learning_rate": 8.492821239247364e-09,
-      "logits/chosen": -0.851113498210907,
-      "logits/rejected": -0.49246683716773987,
-      "logps/chosen": -349.08477783203125,
-      "logps/rejected": -308.5341491699219,
-      "loss": 0.5935,
-      "rewards/accuracies": 0.7437499761581421,
-      "rewards/chosen": 0.11443378776311874,
-      "rewards/margins": 0.3531562387943268,
-      "rewards/rejected": -0.23872247338294983,
+      "epoch": 0.67,
+      "grad_norm": 253.16395063942525,
+      "learning_rate": 2.963346213260737e-09,
+      "logits/chosen": -1.1116909980773926,
+      "logits/rejected": -0.8191507458686829,
+      "logps/chosen": -378.15948486328125,
+      "logps/rejected": -347.91815185546875,
+      "loss": 0.7002,
+      "rewards/accuracies": 0.625,
+      "rewards/chosen": 0.11069490760564804,
+      "rewards/margins": 0.1103142723441124,
+      "rewards/rejected": 0.0003806292952504009,
       "step": 120
     },
     {
-      "epoch": 0.71,
-      "grad_norm": 581.8336234971125,
-      "learning_rate": 8.133518187573863e-09,
-      "logits/chosen": -0.8286861181259155,
-      "logits/rejected": -0.5028008818626404,
-      "logps/chosen": -328.44915771484375,
-      "logps/rejected": -298.2557373046875,
-      "loss": 0.5959,
-      "rewards/accuracies": 0.643750011920929,
-      "rewards/chosen": 0.06084731966257095,
-      "rewards/margins": 0.24498350918293,
-      "rewards/rejected": -0.18413618206977844,
+      "epoch": 0.72,
+      "grad_norm": 231.95660361604925,
+      "learning_rate": 2.1165983894256645e-09,
+      "logits/chosen": -1.1268573999404907,
+      "logits/rejected": -0.9338471293449402,
+      "logps/chosen": -374.75958251953125,
+      "logps/rejected": -351.5363464355469,
+      "loss": 0.6982,
+      "rewards/accuracies": 0.578125,
+      "rewards/chosen": 0.12768486142158508,
+      "rewards/margins": 0.12136401981115341,
+      "rewards/rejected": 0.006320851854979992,
       "step": 130
     },
     {
-      "epoch": 0.77,
-      "grad_norm": 566.985090918969,
-      "learning_rate": 7.745314833479833e-09,
-      "logits/chosen": -1.0121524333953857,
-      "logits/rejected": -0.5360809564590454,
-      "logps/chosen": -338.01531982421875,
-      "logps/rejected": -293.3200378417969,
-      "loss": 0.5864,
-      "rewards/accuracies": 0.7281249761581421,
-      "rewards/chosen": 0.14776429533958435,
-      "rewards/margins": 0.35846027731895447,
-      "rewards/rejected": -0.21069595217704773,
+      "epoch": 0.78,
+      "grad_norm": 253.82374639748912,
+      "learning_rate": 1.3792902133797692e-09,
+      "logits/chosen": -1.1068191528320312,
+      "logits/rejected": -0.99742591381073,
+      "logps/chosen": -356.70904541015625,
+      "logps/rejected": -332.5885009765625,
+      "loss": 0.6989,
+      "rewards/accuracies": 0.590624988079071,
+      "rewards/chosen": 0.13411644101142883,
+      "rewards/margins": 0.12641935050487518,
+      "rewards/rejected": 0.007697069551795721,
       "step": 140
     },
     {
-      "epoch": 0.82,
-      "grad_norm": 585.5307136387664,
-      "learning_rate": 7.33179155967327e-09,
-      "logits/chosen": -0.9024865031242371,
-      "logits/rejected": -0.5908719301223755,
-      "logps/chosen": -331.0682373046875,
-      "logps/rejected": -298.38677978515625,
-      "loss": 0.5718,
-      "rewards/accuracies": 0.675000011920929,
-      "rewards/chosen": 0.1406635046005249,
-      "rewards/margins": 0.32861852645874023,
-      "rewards/rejected": -0.18795502185821533,
+      "epoch": 0.83,
+      "grad_norm": 240.7477679446011,
+      "learning_rate": 7.794062526569734e-10,
+      "logits/chosen": -1.04066801071167,
+      "logits/rejected": -0.7631799578666687,
+      "logps/chosen": -373.73486328125,
+      "logps/rejected": -343.91748046875,
+      "loss": 0.6939,
+      "rewards/accuracies": 0.609375,
+      "rewards/chosen": 0.14227721095085144,
+      "rewards/margins": 0.10592744499444962,
+      "rewards/rejected": 0.03634977340698242,
       "step": 150
     },
     {
-      "epoch": 0.88,
-      "grad_norm": 608.4424256652742,
-      "learning_rate": 6.896762273384178e-09,
-      "logits/chosen": -0.9404014348983765,
-      "logits/rejected": -0.5414465069770813,
-      "logps/chosen": -340.7105407714844,
-      "logps/rejected": -301.7948303222656,
-      "loss": 0.5625,
-      "rewards/accuracies": 0.7406250238418579,
-      "rewards/chosen": 0.29647737741470337,
-      "rewards/margins": 0.45771312713623047,
-      "rewards/rejected": -0.1612357348203659,
+      "epoch": 0.89,
+      "grad_norm": 231.3816310913791,
+      "learning_rate": 3.397151320423647e-10,
+      "logits/chosen": -1.139892578125,
+      "logits/rejected": -0.8710765838623047,
+      "logps/chosen": -368.18963623046875,
+      "logps/rejected": -335.36260986328125,
+      "loss": 0.6985,
+      "rewards/accuracies": 0.6156250238418579,
+      "rewards/chosen": 0.1571863293647766,
+      "rewards/margins": 0.1425299346446991,
+      "rewards/rejected": 0.014656400308012962,
       "step": 160
     },
     {
-      "epoch": 0.93,
-      "grad_norm": 571.5105228832468,
-      "learning_rate": 6.444239230863505e-09,
-      "logits/chosen": -0.6664124131202698,
-      "logits/rejected": -0.3767244815826416,
-      "logps/chosen": -326.7174377441406,
-      "logps/rejected": -292.18304443359375,
-      "loss": 0.5516,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": 0.3695653975009918,
-      "rewards/margins": 0.49222081899642944,
-      "rewards/rejected": -0.12265541404485703,
+      "epoch": 0.95,
+      "grad_norm": 265.6241152762033,
+      "learning_rate": 7.690534931565518e-11,
+      "logits/chosen": -0.9450071454048157,
+      "logits/rejected": -0.8331435918807983,
+      "logps/chosen": -363.50225830078125,
+      "logps/rejected": -337.5380554199219,
+      "loss": 0.6982,
+      "rewards/accuracies": 0.612500011920929,
+      "rewards/chosen": 0.144387885928154,
+      "rewards/margins": 0.17387160658836365,
+      "rewards/rejected": -0.029483716934919357,
       "step": 170
     },
     {
-      "epoch": 0.99,
-      "grad_norm": 486.04829364539904,
-      "learning_rate": 5.97839603251764e-09,
-      "logits/chosen": -0.8600166440010071,
-      "logits/rejected": -0.5389689803123474,
-      "logps/chosen": -323.1247863769531,
-      "logps/rejected": -290.40802001953125,
-      "loss": 0.5408,
-      "rewards/accuracies": 0.784375011920929,
-      "rewards/chosen": 0.3320290446281433,
-      "rewards/margins": 0.5290471315383911,
-      "rewards/rejected": -0.19701813161373138,
-      "step": 180
-    },
-    {
-      "epoch": 1.04,
-      "grad_norm": 524.5665762946036,
-      "learning_rate": 5.5035291299727915e-09,
-      "logits/chosen": -1.0205037593841553,
-      "logits/rejected": -0.6305373907089233,
-      "logps/chosen": -351.1410827636719,
-      "logps/rejected": -313.13812255859375,
-      "loss": 0.5258,
-      "rewards/accuracies": 0.731249988079071,
-      "rewards/chosen": 0.40980952978134155,
-      "rewards/margins": 0.5211332440376282,
-      "rewards/rejected": -0.11132367700338364,
-      "step": 190
-    },
-    {
-      "epoch": 1.1,
-      "grad_norm": 474.67341086960977,
-      "learning_rate": 5.024018200087854e-09,
-      "logits/chosen": -0.836185097694397,
-      "logits/rejected": -0.33586248755455017,
-      "logps/chosen": -321.29168701171875,
-      "logps/rejected": -280.28753662109375,
-      "loss": 0.512,
-      "rewards/accuracies": 0.7562500238418579,
-      "rewards/chosen": 0.3657322824001312,
-      "rewards/margins": 0.5857880711555481,
-      "rewards/rejected": -0.22005578875541687,
-      "step": 200
-    },
-    {
-      "epoch": 1.15,
-      "grad_norm": 480.69197567309544,
-      "learning_rate": 4.544285751384584e-09,
-      "logits/chosen": -0.9048161506652832,
-      "logits/rejected": -0.6336894631385803,
-      "logps/chosen": -329.5954895019531,
-      "logps/rejected": -297.79901123046875,
-      "loss": 0.5067,
-      "rewards/accuracies": 0.721875011920929,
-      "rewards/chosen": 0.41357460618019104,
-      "rewards/margins": 0.5688098669052124,
-      "rewards/rejected": -0.1552352011203766,
-      "step": 210
-    },
-    {
-      "epoch": 1.21,
-      "grad_norm": 498.26717868032813,
-      "learning_rate": 4.068756335443199e-09,
-      "logits/chosen": -0.8500760793685913,
-      "logits/rejected": -0.501125693321228,
-      "logps/chosen": -339.07012939453125,
-      "logps/rejected": -299.48095703125,
-      "loss": 0.514,
-      "rewards/accuracies": 0.7906249761581421,
-      "rewards/chosen": 0.5280696153640747,
-      "rewards/margins": 0.6353160738945007,
-      "rewards/rejected": -0.10724644362926483,
-      "step": 220
-    },
-    {
-      "epoch": 1.26,
-      "grad_norm": 522.8370181213115,
-      "learning_rate": 3.6018157394549284e-09,
-      "logits/chosen": -0.8433824777603149,
-      "logits/rejected": -0.4627400040626526,
-      "logps/chosen": -332.1639709472656,
-      "logps/rejected": -297.5298156738281,
-      "loss": 0.5207,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": 0.4683798849582672,
-      "rewards/margins": 0.5844911336898804,
-      "rewards/rejected": -0.11611130088567734,
-      "step": 230
-    },
-    {
-      "epoch": 1.32,
-      "grad_norm": 521.1145144981034,
-      "learning_rate": 3.14777053629687e-09,
-      "logits/chosen": -1.0663576126098633,
-      "logits/rejected": -0.6788166761398315,
-      "logps/chosen": -336.2250671386719,
-      "logps/rejected": -296.2794494628906,
-      "loss": 0.5124,
-      "rewards/accuracies": 0.7593749761581421,
-      "rewards/chosen": 0.5691989660263062,
-      "rewards/margins": 0.702237069606781,
-      "rewards/rejected": -0.13303814828395844,
-      "step": 240
-    },
-    {
-      "epoch": 1.37,
-      "grad_norm": 487.7677359334702,
-      "learning_rate": 2.7108083651970002e-09,
-      "logits/chosen": -0.7713569402694702,
-      "logits/rejected": -0.4026997983455658,
-      "logps/chosen": -327.7779235839844,
-      "logps/rejected": -290.0624084472656,
-      "loss": 0.5045,
-      "rewards/accuracies": 0.7406250238418579,
-      "rewards/chosen": 0.5247091054916382,
-      "rewards/margins": 0.6842475533485413,
-      "rewards/rejected": -0.1595384031534195,
-      "step": 250
-    },
-    {
-      "epoch": 1.42,
-      "grad_norm": 524.4673235373901,
-      "learning_rate": 2.294959309319086e-09,
-      "logits/chosen": -0.8702203035354614,
-      "logits/rejected": -0.6658123135566711,
-      "logps/chosen": -328.04638671875,
-      "logps/rejected": -302.77777099609375,
-      "loss": 0.5103,
-      "rewards/accuracies": 0.737500011920929,
-      "rewards/chosen": 0.6019162535667419,
-      "rewards/margins": 0.5868763327598572,
-      "rewards/rejected": 0.015039998106658459,
-      "step": 260
-    },
-    {
-      "epoch": 1.48,
-      "grad_norm": 553.8708647904535,
-      "learning_rate": 1.9040587264803672e-09,
-      "logits/chosen": -0.9641194343566895,
-      "logits/rejected": -0.4711450934410095,
-      "logps/chosen": -340.99169921875,
-      "logps/rejected": -293.9861145019531,
-      "loss": 0.4939,
-      "rewards/accuracies": 0.831250011920929,
-      "rewards/chosen": 0.56943279504776,
-      "rewards/margins": 0.7665254473686218,
-      "rewards/rejected": -0.1970926821231842,
-      "step": 270
-    },
-    {
-      "epoch": 1.53,
-      "grad_norm": 453.07875520502444,
-      "learning_rate": 1.5417118758126409e-09,
-      "logits/chosen": -0.8815242648124695,
-      "logits/rejected": -0.4348044991493225,
-      "logps/chosen": -342.6517028808594,
-      "logps/rejected": -296.92169189453125,
-      "loss": 0.4934,
-      "rewards/accuracies": 0.7906249761581421,
-      "rewards/chosen": 0.5705780982971191,
-      "rewards/margins": 0.8008774518966675,
-      "rewards/rejected": -0.23029938340187073,
-      "step": 280
-    },
-    {
-      "epoch": 1.59,
-      "grad_norm": 499.34891989016415,
-      "learning_rate": 1.2112606666135601e-09,
-      "logits/chosen": -0.9316185712814331,
-      "logits/rejected": -0.5389292240142822,
-      "logps/chosen": -333.5206604003906,
-      "logps/rejected": -297.6076354980469,
-      "loss": 0.4978,
-      "rewards/accuracies": 0.765625,
-      "rewards/chosen": 0.5433750748634338,
-      "rewards/margins": 0.6790112257003784,
-      "rewards/rejected": -0.135636106133461,
-      "step": 290
-    },
-    {
-      "epoch": 1.64,
-      "grad_norm": 475.94785473220173,
-      "learning_rate": 9.157528360620415e-10,
-      "logits/chosen": -0.8223875164985657,
-      "logits/rejected": -0.5513488054275513,
-      "logps/chosen": -312.7162780761719,
-      "logps/rejected": -280.6097106933594,
-      "loss": 0.4933,
-      "rewards/accuracies": 0.7593749761581421,
-      "rewards/chosen": 0.5738901495933533,
-      "rewards/margins": 0.6515259146690369,
-      "rewards/rejected": -0.07763569802045822,
-      "step": 300
-    },
-    {
-      "epoch": 1.7,
-      "grad_norm": 530.0260445258433,
-      "learning_rate": 6.579138400703716e-10,
-      "logits/chosen": -0.8587888479232788,
-      "logits/rejected": -0.4476223587989807,
-      "logps/chosen": -334.20648193359375,
-      "logps/rejected": -295.112548828125,
-      "loss": 0.4913,
-      "rewards/accuracies": 0.7906249761581421,
-      "rewards/chosen": 0.6548627018928528,
-      "rewards/margins": 0.7542473077774048,
-      "rewards/rejected": -0.09938466548919678,
-      "step": 310
-    },
-    {
-      "epoch": 1.75,
-      "grad_norm": 487.39892256915766,
-      "learning_rate": 4.401217165224564e-10,
-      "logits/chosen": -0.7274507284164429,
-      "logits/rejected": -0.45758286118507385,
-      "logps/chosen": -322.195068359375,
-      "logps/rejected": -289.47064208984375,
-      "loss": 0.4876,
-      "rewards/accuracies": 0.78125,
-      "rewards/chosen": 0.6047518849372864,
-      "rewards/margins": 0.6844498515129089,
-      "rewards/rejected": -0.07969798892736435,
-      "step": 320
-    },
-    {
-      "epoch": 1.81,
-      "grad_norm": 474.27503066220174,
-      "learning_rate": 2.643851527335006e-10,
-      "logits/chosen": -0.9080432653427124,
-      "logits/rejected": -0.5377383232116699,
-      "logps/chosen": -331.7919921875,
-      "logps/rejected": -301.7752685546875,
-      "loss": 0.4856,
-      "rewards/accuracies": 0.7718750238418579,
-      "rewards/chosen": 0.6742402911186218,
-      "rewards/margins": 0.7511230111122131,
-      "rewards/rejected": -0.07688269764184952,
-      "step": 330
-    },
-    {
-      "epoch": 1.86,
-      "grad_norm": 515.521744630777,
-      "learning_rate": 1.3232495941396638e-10,
-      "logits/chosen": -0.9773859977722168,
-      "logits/rejected": -0.6425100564956665,
-      "logps/chosen": -329.1192321777344,
-      "logps/rejected": -293.9158020019531,
-      "loss": 0.4902,
-      "rewards/accuracies": 0.78125,
-      "rewards/chosen": 0.5788453817367554,
-      "rewards/margins": 0.6871310472488403,
-      "rewards/rejected": -0.10828564316034317,
-      "step": 340
-    },
-    {
-      "epoch": 1.92,
-      "grad_norm": 474.472293237774,
-      "learning_rate": 4.515912200264427e-11,
-      "logits/chosen": -0.9029630422592163,
-      "logits/rejected": -0.4482271075248718,
-      "logps/chosen": -331.9292907714844,
-      "logps/rejected": -287.7762451171875,
-      "loss": 0.4845,
-      "rewards/accuracies": 0.7906249761581421,
-      "rewards/chosen": 0.5662914514541626,
-      "rewards/margins": 0.7054949998855591,
-      "rewards/rejected": -0.1392035186290741,
-      "step": 350
-    },
-    {
-      "epoch": 1.97,
-      "grad_norm": 466.2251927040291,
-      "learning_rate": 3.691567239743621e-12,
-      "logits/chosen": -0.866126537322998,
-      "logits/rejected": -0.38964423537254333,
-      "logps/chosen": -346.421875,
-      "logps/rejected": -304.1325988769531,
-      "loss": 0.4772,
-      "rewards/accuracies": 0.8031250238418579,
-      "rewards/chosen": 0.6508310437202454,
-      "rewards/margins": 0.8615090250968933,
-      "rewards/rejected": -0.21067801117897034,
-      "step": 360
-    },
-    {
-      "epoch": 1.99,
-      "step": 364,
+      "epoch": 1.0,
+      "step": 179,
       "total_flos": 0.0,
-      "train_loss": 0.5661277400923299,
-      "train_runtime": 10889.322,
-      "train_samples_per_second": 8.572,
-      "train_steps_per_second": 0.033
+      "train_loss": 0.7017018035803427,
+      "train_runtime": 5276.0496,
+      "train_samples_per_second": 8.72,
+      "train_steps_per_second": 0.034
     }
   ],
   "logging_steps": 10,
-  "max_steps": 364,
+  "max_steps": 179,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 2,
+  "num_train_epochs": 1,
   "save_steps": 100,
   "total_flos": 0.0,
   "train_batch_size": 8,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8fd356bcfa780902cdf04fd2796c04a6d4a53cd985943fd18f9bc9687ace99b3
+oid sha256:040031e5c3d3e58a32f9908aff846ea8ca608590342dc4c08d071499174fce10
 size 6328