zephyr-7b-ipo-lora / trainer_state.json
sambar's picture
Model save
4fc4e61
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 2.9992254066615027,
"eval_steps": 100,
"global_step": 726,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 6.84931506849315e-09,
"logits/chosen": -2.2289741039276123,
"logits/rejected": -2.0226380825042725,
"logps/chosen": -263.3438720703125,
"logps/rejected": -201.271240234375,
"loss": 25.0,
"rewards/accuracies": 0.0,
"rewards/chosen": 0.0,
"rewards/margins": 0.0,
"rewards/rejected": 0.0,
"step": 1
},
{
"epoch": 0.04,
"learning_rate": 6.84931506849315e-08,
"logits/chosen": -2.266495943069458,
"logits/rejected": -1.9388139247894287,
"logps/chosen": -284.0570068359375,
"logps/rejected": -217.82611083984375,
"loss": 25.2097,
"rewards/accuracies": 0.4253472089767456,
"rewards/chosen": -0.0013144080294296145,
"rewards/margins": -0.0012916413834318519,
"rewards/rejected": -2.276669692946598e-05,
"step": 10
},
{
"epoch": 0.08,
"learning_rate": 1.36986301369863e-07,
"logits/chosen": -2.273313283920288,
"logits/rejected": -1.9863513708114624,
"logps/chosen": -269.41644287109375,
"logps/rejected": -208.65896606445312,
"loss": 25.3631,
"rewards/accuracies": 0.504687488079071,
"rewards/chosen": -0.00041677377885207534,
"rewards/margins": -0.0017736803274601698,
"rewards/rejected": 0.0013569066068157554,
"step": 20
},
{
"epoch": 0.12,
"learning_rate": 2.054794520547945e-07,
"logits/chosen": -2.218705177307129,
"logits/rejected": -1.993547797203064,
"logps/chosen": -243.9556427001953,
"logps/rejected": -199.10511779785156,
"loss": 25.202,
"rewards/accuracies": 0.5171874761581421,
"rewards/chosen": -0.0021037233527749777,
"rewards/margins": 0.0023984042927622795,
"rewards/rejected": -0.004502127878367901,
"step": 30
},
{
"epoch": 0.17,
"learning_rate": 2.73972602739726e-07,
"logits/chosen": -2.3098392486572266,
"logits/rejected": -1.996872901916504,
"logps/chosen": -274.44091796875,
"logps/rejected": -208.118896484375,
"loss": 25.1001,
"rewards/accuracies": 0.5062500238418579,
"rewards/chosen": 5.082110874354839e-05,
"rewards/margins": 0.0012452874798327684,
"rewards/rejected": -0.0011944664875045419,
"step": 40
},
{
"epoch": 0.21,
"learning_rate": 3.424657534246575e-07,
"logits/chosen": -2.2532970905303955,
"logits/rejected": -2.0162949562072754,
"logps/chosen": -258.4920959472656,
"logps/rejected": -215.529052734375,
"loss": 24.7793,
"rewards/accuracies": 0.53125,
"rewards/chosen": 0.0014612700324505568,
"rewards/margins": 0.004128460772335529,
"rewards/rejected": -0.0026671909727156162,
"step": 50
},
{
"epoch": 0.25,
"learning_rate": 4.10958904109589e-07,
"logits/chosen": -2.2521817684173584,
"logits/rejected": -1.9908069372177124,
"logps/chosen": -257.59222412109375,
"logps/rejected": -206.552490234375,
"loss": 24.2711,
"rewards/accuracies": 0.567187488079071,
"rewards/chosen": 0.008318379521369934,
"rewards/margins": 0.010114217177033424,
"rewards/rejected": -0.0017958376556634903,
"step": 60
},
{
"epoch": 0.29,
"learning_rate": 4.794520547945205e-07,
"logits/chosen": -2.2163357734680176,
"logits/rejected": -1.997807264328003,
"logps/chosen": -245.3988800048828,
"logps/rejected": -206.3892364501953,
"loss": 23.8848,
"rewards/accuracies": 0.604687511920929,
"rewards/chosen": 0.008774536661803722,
"rewards/margins": 0.015018345788121223,
"rewards/rejected": -0.006243808660656214,
"step": 70
},
{
"epoch": 0.33,
"learning_rate": 4.946401225114854e-07,
"logits/chosen": -2.272542715072632,
"logits/rejected": -1.9779363870620728,
"logps/chosen": -256.14208984375,
"logps/rejected": -210.4285125732422,
"loss": 23.3744,
"rewards/accuracies": 0.6187499761581421,
"rewards/chosen": 0.011616826988756657,
"rewards/margins": 0.023227987810969353,
"rewards/rejected": -0.01161116175353527,
"step": 80
},
{
"epoch": 0.37,
"learning_rate": 4.869831546707504e-07,
"logits/chosen": -2.310504913330078,
"logits/rejected": -2.018115520477295,
"logps/chosen": -262.4266052246094,
"logps/rejected": -216.05191040039062,
"loss": 22.5882,
"rewards/accuracies": 0.6484375,
"rewards/chosen": 0.02013319917023182,
"rewards/margins": 0.03405465558171272,
"rewards/rejected": -0.013921457342803478,
"step": 90
},
{
"epoch": 0.41,
"learning_rate": 4.793261868300153e-07,
"logits/chosen": -2.2590413093566895,
"logits/rejected": -1.9913495779037476,
"logps/chosen": -264.6045837402344,
"logps/rejected": -219.0672149658203,
"loss": 22.0702,
"rewards/accuracies": 0.6640625,
"rewards/chosen": 0.02459399588406086,
"rewards/margins": 0.04087246581912041,
"rewards/rejected": -0.016278475522994995,
"step": 100
},
{
"epoch": 0.45,
"learning_rate": 4.7166921898928023e-07,
"logits/chosen": -2.2836508750915527,
"logits/rejected": -2.039545774459839,
"logps/chosen": -243.48867797851562,
"logps/rejected": -214.2646942138672,
"loss": 21.7548,
"rewards/accuracies": 0.6656249761581421,
"rewards/chosen": 0.020637672394514084,
"rewards/margins": 0.043859176337718964,
"rewards/rejected": -0.02322150394320488,
"step": 110
},
{
"epoch": 0.5,
"learning_rate": 4.640122511485451e-07,
"logits/chosen": -2.2533435821533203,
"logits/rejected": -2.0158801078796387,
"logps/chosen": -247.85543823242188,
"logps/rejected": -211.3623504638672,
"loss": 21.608,
"rewards/accuracies": 0.635937511920929,
"rewards/chosen": 0.026100531220436096,
"rewards/margins": 0.05109390616416931,
"rewards/rejected": -0.024993373081088066,
"step": 120
},
{
"epoch": 0.54,
"learning_rate": 4.563552833078101e-07,
"logits/chosen": -2.2582602500915527,
"logits/rejected": -2.0190303325653076,
"logps/chosen": -256.9596252441406,
"logps/rejected": -217.3245849609375,
"loss": 20.8686,
"rewards/accuracies": 0.667187511920929,
"rewards/chosen": 0.028709357604384422,
"rewards/margins": 0.06462417542934418,
"rewards/rejected": -0.035914815962314606,
"step": 130
},
{
"epoch": 0.58,
"learning_rate": 4.4869831546707505e-07,
"logits/chosen": -2.2754273414611816,
"logits/rejected": -2.038130521774292,
"logps/chosen": -261.1288757324219,
"logps/rejected": -221.7465057373047,
"loss": 20.9236,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.03735818341374397,
"rewards/margins": 0.07491641491651535,
"rewards/rejected": -0.037558235228061676,
"step": 140
},
{
"epoch": 0.62,
"learning_rate": 4.4104134762633994e-07,
"logits/chosen": -2.2817165851593018,
"logits/rejected": -2.055285930633545,
"logps/chosen": -257.0137939453125,
"logps/rejected": -217.66183471679688,
"loss": 20.6598,
"rewards/accuracies": 0.6703125238418579,
"rewards/chosen": 0.03143654763698578,
"rewards/margins": 0.06876268237829208,
"rewards/rejected": -0.0373261384665966,
"step": 150
},
{
"epoch": 0.66,
"learning_rate": 4.333843797856049e-07,
"logits/chosen": -2.2860567569732666,
"logits/rejected": -1.999734878540039,
"logps/chosen": -270.0646057128906,
"logps/rejected": -229.95010375976562,
"loss": 19.9317,
"rewards/accuracies": 0.7015625238418579,
"rewards/chosen": 0.04298175126314163,
"rewards/margins": 0.08404376357793808,
"rewards/rejected": -0.04106200858950615,
"step": 160
},
{
"epoch": 0.7,
"learning_rate": 4.257274119448698e-07,
"logits/chosen": -2.3108818531036377,
"logits/rejected": -2.0907673835754395,
"logps/chosen": -261.53253173828125,
"logps/rejected": -230.0071563720703,
"loss": 20.2237,
"rewards/accuracies": 0.6546875238418579,
"rewards/chosen": 0.03258711099624634,
"rewards/margins": 0.07823493331670761,
"rewards/rejected": -0.04564782604575157,
"step": 170
},
{
"epoch": 0.74,
"learning_rate": 4.180704441041347e-07,
"logits/chosen": -2.2532687187194824,
"logits/rejected": -2.0309438705444336,
"logps/chosen": -247.6223907470703,
"logps/rejected": -216.98587036132812,
"loss": 20.6143,
"rewards/accuracies": 0.671875,
"rewards/chosen": 0.027079517021775246,
"rewards/margins": 0.08031971752643585,
"rewards/rejected": -0.05324019119143486,
"step": 180
},
{
"epoch": 0.78,
"learning_rate": 4.1041347626339966e-07,
"logits/chosen": -2.290727138519287,
"logits/rejected": -2.008113384246826,
"logps/chosen": -252.6389617919922,
"logps/rejected": -204.10769653320312,
"loss": 19.923,
"rewards/accuracies": 0.668749988079071,
"rewards/chosen": 0.02958676591515541,
"rewards/margins": 0.09163785725831985,
"rewards/rejected": -0.06205107644200325,
"step": 190
},
{
"epoch": 0.83,
"learning_rate": 4.027565084226646e-07,
"logits/chosen": -2.278743267059326,
"logits/rejected": -2.0434935092926025,
"logps/chosen": -251.91677856445312,
"logps/rejected": -208.04470825195312,
"loss": 19.5106,
"rewards/accuracies": 0.71875,
"rewards/chosen": 0.03315151482820511,
"rewards/margins": 0.09510111808776855,
"rewards/rejected": -0.061949603259563446,
"step": 200
},
{
"epoch": 0.87,
"learning_rate": 3.9509954058192954e-07,
"logits/chosen": -2.288477897644043,
"logits/rejected": -2.027742385864258,
"logps/chosen": -274.6870422363281,
"logps/rejected": -221.636962890625,
"loss": 19.5079,
"rewards/accuracies": 0.684374988079071,
"rewards/chosen": 0.03773919492959976,
"rewards/margins": 0.10934200137853622,
"rewards/rejected": -0.07160280644893646,
"step": 210
},
{
"epoch": 0.91,
"learning_rate": 3.874425727411945e-07,
"logits/chosen": -2.2594573497772217,
"logits/rejected": -2.008882999420166,
"logps/chosen": -267.6262512207031,
"logps/rejected": -219.49728393554688,
"loss": 19.6404,
"rewards/accuracies": 0.698437511920929,
"rewards/chosen": 0.04026947170495987,
"rewards/margins": 0.10674212872982025,
"rewards/rejected": -0.06647266447544098,
"step": 220
},
{
"epoch": 0.95,
"learning_rate": 3.797856049004594e-07,
"logits/chosen": -2.26505708694458,
"logits/rejected": -2.060857057571411,
"logps/chosen": -254.90005493164062,
"logps/rejected": -218.54141235351562,
"loss": 19.6698,
"rewards/accuracies": 0.6796875,
"rewards/chosen": 0.035312727093696594,
"rewards/margins": 0.0973130315542221,
"rewards/rejected": -0.062000297009944916,
"step": 230
},
{
"epoch": 0.99,
"learning_rate": 3.7212863705972436e-07,
"logits/chosen": -2.2956652641296387,
"logits/rejected": -2.047691822052002,
"logps/chosen": -261.9791564941406,
"logps/rejected": -212.89151000976562,
"loss": 19.3937,
"rewards/accuracies": 0.706250011920929,
"rewards/chosen": 0.035399921238422394,
"rewards/margins": 0.10358913987874985,
"rewards/rejected": -0.06818921864032745,
"step": 240
},
{
"epoch": 1.0,
"eval_logits/chosen": -2.0236928462982178,
"eval_logits/rejected": -1.7961955070495605,
"eval_logps/chosen": -255.23326110839844,
"eval_logps/rejected": -211.76084899902344,
"eval_loss": 19.3449764251709,
"eval_rewards/accuracies": 0.7039999961853027,
"eval_rewards/chosen": 0.029078969731926918,
"eval_rewards/margins": 0.1019509881734848,
"eval_rewards/rejected": -0.07287202030420303,
"eval_runtime": 239.3505,
"eval_samples_per_second": 8.356,
"eval_steps_per_second": 0.522,
"step": 242
},
{
"epoch": 1.03,
"learning_rate": 3.6447166921898925e-07,
"logits/chosen": -2.2042205333709717,
"logits/rejected": -2.014646053314209,
"logps/chosen": -248.44058227539062,
"logps/rejected": -200.34120178222656,
"loss": 19.9018,
"rewards/accuracies": 0.6812499761581421,
"rewards/chosen": 0.03346983343362808,
"rewards/margins": 0.10174000263214111,
"rewards/rejected": -0.06827016919851303,
"step": 250
},
{
"epoch": 1.07,
"learning_rate": 3.568147013782542e-07,
"logits/chosen": -2.2492403984069824,
"logits/rejected": -2.0223591327667236,
"logps/chosen": -252.7912139892578,
"logps/rejected": -216.12240600585938,
"loss": 19.0687,
"rewards/accuracies": 0.7265625,
"rewards/chosen": 0.03231377154588699,
"rewards/margins": 0.10541899502277374,
"rewards/rejected": -0.07310522347688675,
"step": 260
},
{
"epoch": 1.12,
"learning_rate": 3.4915773353751913e-07,
"logits/chosen": -2.26572847366333,
"logits/rejected": -1.9836113452911377,
"logps/chosen": -268.9698791503906,
"logps/rejected": -225.75338745117188,
"loss": 18.628,
"rewards/accuracies": 0.7484375238418579,
"rewards/chosen": 0.04413260146975517,
"rewards/margins": 0.1356559544801712,
"rewards/rejected": -0.09152336418628693,
"step": 270
},
{
"epoch": 1.16,
"learning_rate": 3.41500765696784e-07,
"logits/chosen": -2.270271062850952,
"logits/rejected": -2.0418267250061035,
"logps/chosen": -241.98959350585938,
"logps/rejected": -215.8519287109375,
"loss": 19.2251,
"rewards/accuracies": 0.734375,
"rewards/chosen": 0.03690015524625778,
"rewards/margins": 0.11016629636287689,
"rewards/rejected": -0.07326614856719971,
"step": 280
},
{
"epoch": 1.2,
"learning_rate": 3.33843797856049e-07,
"logits/chosen": -2.2635691165924072,
"logits/rejected": -2.0188465118408203,
"logps/chosen": -273.9185791015625,
"logps/rejected": -217.93167114257812,
"loss": 18.6598,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.03699145466089249,
"rewards/margins": 0.12567397952079773,
"rewards/rejected": -0.08868252485990524,
"step": 290
},
{
"epoch": 1.24,
"learning_rate": 3.2618683001531396e-07,
"logits/chosen": -2.2175521850585938,
"logits/rejected": -1.9899898767471313,
"logps/chosen": -251.8654327392578,
"logps/rejected": -217.54421997070312,
"loss": 18.5271,
"rewards/accuracies": 0.714062511920929,
"rewards/chosen": 0.03108426369726658,
"rewards/margins": 0.1172715276479721,
"rewards/rejected": -0.08618726581335068,
"step": 300
},
{
"epoch": 1.28,
"learning_rate": 3.1852986217457885e-07,
"logits/chosen": -2.287581205368042,
"logits/rejected": -2.035792827606201,
"logps/chosen": -250.79638671875,
"logps/rejected": -218.07278442382812,
"loss": 19.1524,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.03225615620613098,
"rewards/margins": 0.11478684097528458,
"rewards/rejected": -0.082530677318573,
"step": 310
},
{
"epoch": 1.32,
"learning_rate": 3.108728943338438e-07,
"logits/chosen": -2.2732996940612793,
"logits/rejected": -2.085688829421997,
"logps/chosen": -243.0460662841797,
"logps/rejected": -227.0532989501953,
"loss": 19.1273,
"rewards/accuracies": 0.692187488079071,
"rewards/chosen": 0.02955133654177189,
"rewards/margins": 0.10330448299646378,
"rewards/rejected": -0.07375315576791763,
"step": 320
},
{
"epoch": 1.36,
"learning_rate": 3.0321592649310873e-07,
"logits/chosen": -2.2556710243225098,
"logits/rejected": -1.9700673818588257,
"logps/chosen": -243.98757934570312,
"logps/rejected": -214.067138671875,
"loss": 18.57,
"rewards/accuracies": 0.7281249761581421,
"rewards/chosen": 0.03395627439022064,
"rewards/margins": 0.1173306480050087,
"rewards/rejected": -0.08337438106536865,
"step": 330
},
{
"epoch": 1.4,
"learning_rate": 2.955589586523736e-07,
"logits/chosen": -2.285491466522217,
"logits/rejected": -2.017014503479004,
"logps/chosen": -256.3018493652344,
"logps/rejected": -215.5684051513672,
"loss": 19.3792,
"rewards/accuracies": 0.7265625,
"rewards/chosen": 0.02712482213973999,
"rewards/margins": 0.11975017935037613,
"rewards/rejected": -0.09262534976005554,
"step": 340
},
{
"epoch": 1.45,
"learning_rate": 2.8790199081163856e-07,
"logits/chosen": -2.2827515602111816,
"logits/rejected": -1.9982925653457642,
"logps/chosen": -256.5948181152344,
"logps/rejected": -214.377197265625,
"loss": 19.2092,
"rewards/accuracies": 0.6890624761581421,
"rewards/chosen": 0.02536213956773281,
"rewards/margins": 0.10417892783880234,
"rewards/rejected": -0.07881677150726318,
"step": 350
},
{
"epoch": 1.49,
"learning_rate": 2.802450229709035e-07,
"logits/chosen": -2.226412296295166,
"logits/rejected": -1.9456377029418945,
"logps/chosen": -245.56753540039062,
"logps/rejected": -202.68087768554688,
"loss": 18.5533,
"rewards/accuracies": 0.7015625238418579,
"rewards/chosen": 0.023562483489513397,
"rewards/margins": 0.11015045642852783,
"rewards/rejected": -0.08658796548843384,
"step": 360
},
{
"epoch": 1.53,
"learning_rate": 2.725880551301684e-07,
"logits/chosen": -2.2886481285095215,
"logits/rejected": -2.075536012649536,
"logps/chosen": -265.2494812011719,
"logps/rejected": -222.0449981689453,
"loss": 19.4234,
"rewards/accuracies": 0.690625011920929,
"rewards/chosen": 0.0262996107339859,
"rewards/margins": 0.11352996528148651,
"rewards/rejected": -0.08723036199808121,
"step": 370
},
{
"epoch": 1.57,
"learning_rate": 2.649310872894334e-07,
"logits/chosen": -2.2954063415527344,
"logits/rejected": -2.019110918045044,
"logps/chosen": -265.41534423828125,
"logps/rejected": -211.8571319580078,
"loss": 18.7522,
"rewards/accuracies": 0.734375,
"rewards/chosen": 0.030247170478105545,
"rewards/margins": 0.12689557671546936,
"rewards/rejected": -0.09664840996265411,
"step": 380
},
{
"epoch": 1.61,
"learning_rate": 2.572741194486983e-07,
"logits/chosen": -2.307234525680542,
"logits/rejected": -1.9544427394866943,
"logps/chosen": -263.83050537109375,
"logps/rejected": -196.88644409179688,
"loss": 18.7036,
"rewards/accuracies": 0.721875011920929,
"rewards/chosen": 0.02584310993552208,
"rewards/margins": 0.11804970353841782,
"rewards/rejected": -0.09220659732818604,
"step": 390
},
{
"epoch": 1.65,
"learning_rate": 2.496171516079632e-07,
"logits/chosen": -2.3390731811523438,
"logits/rejected": -2.020245313644409,
"logps/chosen": -283.559326171875,
"logps/rejected": -228.0408477783203,
"loss": 18.9635,
"rewards/accuracies": 0.734375,
"rewards/chosen": 0.03759358078241348,
"rewards/margins": 0.13340887427330017,
"rewards/rejected": -0.09581530094146729,
"step": 400
},
{
"epoch": 1.69,
"learning_rate": 2.4196018376722816e-07,
"logits/chosen": -2.25607967376709,
"logits/rejected": -2.005556344985962,
"logps/chosen": -268.84881591796875,
"logps/rejected": -216.67996215820312,
"loss": 19.2263,
"rewards/accuracies": 0.7265625,
"rewards/chosen": 0.027539696544408798,
"rewards/margins": 0.12173386663198471,
"rewards/rejected": -0.09419417381286621,
"step": 410
},
{
"epoch": 1.74,
"learning_rate": 2.343032159264931e-07,
"logits/chosen": -2.234701633453369,
"logits/rejected": -1.9779908657073975,
"logps/chosen": -259.31536865234375,
"logps/rejected": -212.6099853515625,
"loss": 18.7807,
"rewards/accuracies": 0.729687511920929,
"rewards/chosen": 0.03047511912882328,
"rewards/margins": 0.13214388489723206,
"rewards/rejected": -0.10166877508163452,
"step": 420
},
{
"epoch": 1.78,
"learning_rate": 2.26646248085758e-07,
"logits/chosen": -2.231544017791748,
"logits/rejected": -1.981184720993042,
"logps/chosen": -249.824951171875,
"logps/rejected": -205.8656005859375,
"loss": 18.6868,
"rewards/accuracies": 0.729687511920929,
"rewards/chosen": 0.026928503066301346,
"rewards/margins": 0.128057599067688,
"rewards/rejected": -0.10112909972667694,
"step": 430
},
{
"epoch": 1.82,
"learning_rate": 2.1898928024502298e-07,
"logits/chosen": -2.2951433658599854,
"logits/rejected": -2.026249408721924,
"logps/chosen": -253.8539581298828,
"logps/rejected": -210.1437225341797,
"loss": 18.3582,
"rewards/accuracies": 0.739062488079071,
"rewards/chosen": 0.03310644254088402,
"rewards/margins": 0.12848404049873352,
"rewards/rejected": -0.09537758678197861,
"step": 440
},
{
"epoch": 1.86,
"learning_rate": 2.113323124042879e-07,
"logits/chosen": -2.2722887992858887,
"logits/rejected": -2.0053603649139404,
"logps/chosen": -261.27166748046875,
"logps/rejected": -219.5053253173828,
"loss": 18.8048,
"rewards/accuracies": 0.7265625,
"rewards/chosen": 0.034187205135822296,
"rewards/margins": 0.1302899569272995,
"rewards/rejected": -0.0961027592420578,
"step": 450
},
{
"epoch": 1.9,
"learning_rate": 2.036753445635528e-07,
"logits/chosen": -2.279463052749634,
"logits/rejected": -1.9979301691055298,
"logps/chosen": -273.9200744628906,
"logps/rejected": -225.6119842529297,
"loss": 18.3929,
"rewards/accuracies": 0.7093750238418579,
"rewards/chosen": 0.0315621979534626,
"rewards/margins": 0.1322147399187088,
"rewards/rejected": -0.1006525382399559,
"step": 460
},
{
"epoch": 1.94,
"learning_rate": 1.9601837672281775e-07,
"logits/chosen": -2.261075735092163,
"logits/rejected": -2.026048183441162,
"logps/chosen": -260.4441223144531,
"logps/rejected": -224.8106231689453,
"loss": 18.4695,
"rewards/accuracies": 0.7359374761581421,
"rewards/chosen": 0.025445517152547836,
"rewards/margins": 0.12411808967590332,
"rewards/rejected": -0.09867255389690399,
"step": 470
},
{
"epoch": 1.98,
"learning_rate": 1.883614088820827e-07,
"logits/chosen": -2.296330213546753,
"logits/rejected": -1.9999678134918213,
"logps/chosen": -258.93145751953125,
"logps/rejected": -212.75003051757812,
"loss": 19.376,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.027268463745713234,
"rewards/margins": 0.12198734283447266,
"rewards/rejected": -0.09471887350082397,
"step": 480
},
{
"epoch": 2.0,
"eval_logits/chosen": -2.023204803466797,
"eval_logits/rejected": -1.7954164743423462,
"eval_logps/chosen": -255.2545623779297,
"eval_logps/rejected": -211.98085021972656,
"eval_loss": 18.819795608520508,
"eval_rewards/accuracies": 0.7020000219345093,
"eval_rewards/chosen": 0.026950573548674583,
"eval_rewards/margins": 0.12182173132896423,
"eval_rewards/rejected": -0.0948711559176445,
"eval_runtime": 239.4946,
"eval_samples_per_second": 8.351,
"eval_steps_per_second": 0.522,
"step": 484
},
{
"epoch": 2.02,
"learning_rate": 1.807044410413476e-07,
"logits/chosen": -2.2521510124206543,
"logits/rejected": -2.021901845932007,
"logps/chosen": -248.35025024414062,
"logps/rejected": -224.5894775390625,
"loss": 18.8769,
"rewards/accuracies": 0.7359374761581421,
"rewards/chosen": 0.026256313547492027,
"rewards/margins": 0.1273646056652069,
"rewards/rejected": -0.10110831260681152,
"step": 490
},
{
"epoch": 2.07,
"learning_rate": 1.7304747320061255e-07,
"logits/chosen": -2.2362194061279297,
"logits/rejected": -2.0276010036468506,
"logps/chosen": -253.75991821289062,
"logps/rejected": -218.86349487304688,
"loss": 18.6885,
"rewards/accuracies": 0.7265625,
"rewards/chosen": 0.03784112259745598,
"rewards/margins": 0.12961548566818237,
"rewards/rejected": -0.0917743593454361,
"step": 500
},
{
"epoch": 2.11,
"learning_rate": 1.6539050535987747e-07,
"logits/chosen": -2.3277292251586914,
"logits/rejected": -2.005606174468994,
"logps/chosen": -260.1662902832031,
"logps/rejected": -212.98428344726562,
"loss": 18.2008,
"rewards/accuracies": 0.731249988079071,
"rewards/chosen": 0.030466770753264427,
"rewards/margins": 0.13587407767772675,
"rewards/rejected": -0.10540731251239777,
"step": 510
},
{
"epoch": 2.15,
"learning_rate": 1.5773353751914243e-07,
"logits/chosen": -2.354032516479492,
"logits/rejected": -2.0204906463623047,
"logps/chosen": -275.2497863769531,
"logps/rejected": -218.10595703125,
"loss": 18.827,
"rewards/accuracies": 0.7515624761581421,
"rewards/chosen": 0.0328175388276577,
"rewards/margins": 0.13892371952533722,
"rewards/rejected": -0.10610618442296982,
"step": 520
},
{
"epoch": 2.19,
"learning_rate": 1.5007656967840735e-07,
"logits/chosen": -2.227428436279297,
"logits/rejected": -1.9895546436309814,
"logps/chosen": -244.3919677734375,
"logps/rejected": -218.86886596679688,
"loss": 18.3559,
"rewards/accuracies": 0.753125011920929,
"rewards/chosen": 0.027428537607192993,
"rewards/margins": 0.13661186397075653,
"rewards/rejected": -0.10918332636356354,
"step": 530
},
{
"epoch": 2.23,
"learning_rate": 1.4241960183767226e-07,
"logits/chosen": -2.2682414054870605,
"logits/rejected": -1.9858787059783936,
"logps/chosen": -272.84637451171875,
"logps/rejected": -223.4131622314453,
"loss": 18.6971,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.024813225492835045,
"rewards/margins": 0.12149915844202042,
"rewards/rejected": -0.09668593108654022,
"step": 540
},
{
"epoch": 2.27,
"learning_rate": 1.347626339969372e-07,
"logits/chosen": -2.2588775157928467,
"logits/rejected": -2.093949794769287,
"logps/chosen": -261.2510986328125,
"logps/rejected": -230.35330200195312,
"loss": 18.6719,
"rewards/accuracies": 0.723437488079071,
"rewards/chosen": 0.034928444772958755,
"rewards/margins": 0.12344489991664886,
"rewards/rejected": -0.08851645886898041,
"step": 550
},
{
"epoch": 2.31,
"learning_rate": 1.2710566615620215e-07,
"logits/chosen": -2.2328484058380127,
"logits/rejected": -1.9578449726104736,
"logps/chosen": -264.450439453125,
"logps/rejected": -212.9776611328125,
"loss": 17.9566,
"rewards/accuracies": 0.734375,
"rewards/chosen": 0.026903945952653885,
"rewards/margins": 0.13341760635375977,
"rewards/rejected": -0.10651366412639618,
"step": 560
},
{
"epoch": 2.35,
"learning_rate": 1.1944869831546706e-07,
"logits/chosen": -2.237755060195923,
"logits/rejected": -2.0300750732421875,
"logps/chosen": -255.99853515625,
"logps/rejected": -226.13626098632812,
"loss": 18.5265,
"rewards/accuracies": 0.698437511920929,
"rewards/chosen": 0.023317929357290268,
"rewards/margins": 0.12482543289661407,
"rewards/rejected": -0.1015075072646141,
"step": 570
},
{
"epoch": 2.4,
"learning_rate": 1.11791730474732e-07,
"logits/chosen": -2.257673740386963,
"logits/rejected": -1.9982837438583374,
"logps/chosen": -252.7151336669922,
"logps/rejected": -213.78366088867188,
"loss": 18.0724,
"rewards/accuracies": 0.7593749761581421,
"rewards/chosen": 0.023391084745526314,
"rewards/margins": 0.14269840717315674,
"rewards/rejected": -0.11930731683969498,
"step": 580
},
{
"epoch": 2.44,
"learning_rate": 1.0413476263399694e-07,
"logits/chosen": -2.297306537628174,
"logits/rejected": -2.012129783630371,
"logps/chosen": -259.78497314453125,
"logps/rejected": -206.7128448486328,
"loss": 18.4955,
"rewards/accuracies": 0.7093750238418579,
"rewards/chosen": 0.017748359590768814,
"rewards/margins": 0.12208724021911621,
"rewards/rejected": -0.1043388843536377,
"step": 590
},
{
"epoch": 2.48,
"learning_rate": 9.647779479326186e-08,
"logits/chosen": -2.2344954013824463,
"logits/rejected": -2.0501608848571777,
"logps/chosen": -240.86679077148438,
"logps/rejected": -212.7677764892578,
"loss": 18.9662,
"rewards/accuracies": 0.7203124761581421,
"rewards/chosen": 0.02210834063589573,
"rewards/margins": 0.11001463979482651,
"rewards/rejected": -0.08790630102157593,
"step": 600
},
{
"epoch": 2.52,
"learning_rate": 8.88208269525268e-08,
"logits/chosen": -2.239448308944702,
"logits/rejected": -1.992950201034546,
"logps/chosen": -250.8835906982422,
"logps/rejected": -211.41421508789062,
"loss": 18.8806,
"rewards/accuracies": 0.7250000238418579,
"rewards/chosen": 0.02492038905620575,
"rewards/margins": 0.1194721907377243,
"rewards/rejected": -0.09455180168151855,
"step": 610
},
{
"epoch": 2.56,
"learning_rate": 8.116385911179173e-08,
"logits/chosen": -2.2440013885498047,
"logits/rejected": -1.960422158241272,
"logps/chosen": -270.11187744140625,
"logps/rejected": -207.26058959960938,
"loss": 18.2745,
"rewards/accuracies": 0.737500011920929,
"rewards/chosen": 0.03377198800444603,
"rewards/margins": 0.13419213891029358,
"rewards/rejected": -0.10042013972997665,
"step": 620
},
{
"epoch": 2.6,
"learning_rate": 7.350689127105667e-08,
"logits/chosen": -2.2937111854553223,
"logits/rejected": -1.987168550491333,
"logps/chosen": -262.86199951171875,
"logps/rejected": -219.86508178710938,
"loss": 18.1999,
"rewards/accuracies": 0.7093750238418579,
"rewards/chosen": 0.03628316521644592,
"rewards/margins": 0.14258472621440887,
"rewards/rejected": -0.10630156099796295,
"step": 630
},
{
"epoch": 2.64,
"learning_rate": 6.584992343032159e-08,
"logits/chosen": -2.2060506343841553,
"logits/rejected": -1.9665876626968384,
"logps/chosen": -253.2500457763672,
"logps/rejected": -216.85476684570312,
"loss": 18.8148,
"rewards/accuracies": 0.7203124761581421,
"rewards/chosen": 0.03216860815882683,
"rewards/margins": 0.12755930423736572,
"rewards/rejected": -0.0953906923532486,
"step": 640
},
{
"epoch": 2.69,
"learning_rate": 5.819295558958652e-08,
"logits/chosen": -2.2680506706237793,
"logits/rejected": -1.9561970233917236,
"logps/chosen": -277.43157958984375,
"logps/rejected": -209.50244140625,
"loss": 18.345,
"rewards/accuracies": 0.7437499761581421,
"rewards/chosen": 0.03760701045393944,
"rewards/margins": 0.14586040377616882,
"rewards/rejected": -0.10825341939926147,
"step": 650
},
{
"epoch": 2.73,
"learning_rate": 5.0535987748851455e-08,
"logits/chosen": -2.3038992881774902,
"logits/rejected": -2.04447078704834,
"logps/chosen": -265.35980224609375,
"logps/rejected": -219.22085571289062,
"loss": 18.1755,
"rewards/accuracies": 0.7421875,
"rewards/chosen": 0.027750462293624878,
"rewards/margins": 0.12442169338464737,
"rewards/rejected": -0.09667123109102249,
"step": 660
},
{
"epoch": 2.77,
"learning_rate": 4.287901990811638e-08,
"logits/chosen": -2.2642054557800293,
"logits/rejected": -2.0056357383728027,
"logps/chosen": -258.467041015625,
"logps/rejected": -232.66946411132812,
"loss": 18.3268,
"rewards/accuracies": 0.7203124761581421,
"rewards/chosen": 0.033985260874032974,
"rewards/margins": 0.14256197214126587,
"rewards/rejected": -0.1085767149925232,
"step": 670
},
{
"epoch": 2.81,
"learning_rate": 3.522205206738132e-08,
"logits/chosen": -2.2641890048980713,
"logits/rejected": -2.0118279457092285,
"logps/chosen": -265.30517578125,
"logps/rejected": -217.5371856689453,
"loss": 18.4502,
"rewards/accuracies": 0.7359374761581421,
"rewards/chosen": 0.027841120958328247,
"rewards/margins": 0.12672238051891327,
"rewards/rejected": -0.09888125956058502,
"step": 680
},
{
"epoch": 2.85,
"learning_rate": 2.7565084226646246e-08,
"logits/chosen": -2.2387919425964355,
"logits/rejected": -2.0168097019195557,
"logps/chosen": -254.23617553710938,
"logps/rejected": -224.2781219482422,
"loss": 18.6327,
"rewards/accuracies": 0.7281249761581421,
"rewards/chosen": 0.03157157823443413,
"rewards/margins": 0.13236665725708008,
"rewards/rejected": -0.10079507529735565,
"step": 690
},
{
"epoch": 2.89,
"learning_rate": 1.9908116385911178e-08,
"logits/chosen": -2.2574362754821777,
"logits/rejected": -2.012924909591675,
"logps/chosen": -250.27371215820312,
"logps/rejected": -215.03756713867188,
"loss": 18.7916,
"rewards/accuracies": 0.690625011920929,
"rewards/chosen": 0.023055683821439743,
"rewards/margins": 0.11122564226388931,
"rewards/rejected": -0.08816995471715927,
"step": 700
},
{
"epoch": 2.93,
"learning_rate": 1.225114854517611e-08,
"logits/chosen": -2.309433937072754,
"logits/rejected": -2.0074551105499268,
"logps/chosen": -270.73040771484375,
"logps/rejected": -209.7691192626953,
"loss": 18.5426,
"rewards/accuracies": 0.723437488079071,
"rewards/chosen": 0.025719935074448586,
"rewards/margins": 0.13181468844413757,
"rewards/rejected": -0.10609474033117294,
"step": 710
},
{
"epoch": 2.97,
"learning_rate": 4.594180704441042e-09,
"logits/chosen": -2.2386531829833984,
"logits/rejected": -2.058807849884033,
"logps/chosen": -248.2914581298828,
"logps/rejected": -213.177490234375,
"loss": 18.4503,
"rewards/accuracies": 0.7124999761581421,
"rewards/chosen": 0.0301600880920887,
"rewards/margins": 0.12142640352249146,
"rewards/rejected": -0.09126633405685425,
"step": 720
},
{
"epoch": 3.0,
"eval_logits/chosen": -2.0242695808410645,
"eval_logits/rejected": -1.7967454195022583,
"eval_logps/chosen": -255.23193359375,
"eval_logps/rejected": -212.03785705566406,
"eval_loss": 18.33973503112793,
"eval_rewards/accuracies": 0.7200000286102295,
"eval_rewards/chosen": 0.029216337949037552,
"eval_rewards/margins": 0.12978971004486084,
"eval_rewards/rejected": -0.10057336091995239,
"eval_runtime": 239.5813,
"eval_samples_per_second": 8.348,
"eval_steps_per_second": 0.522,
"step": 726
},
{
"epoch": 3.0,
"step": 726,
"total_flos": 0.0,
"train_loss": 19.76867720969124,
"train_runtime": 32942.8373,
"train_samples_per_second": 5.643,
"train_steps_per_second": 0.022
}
],
"logging_steps": 10,
"max_steps": 726,
"num_train_epochs": 3,
"save_steps": 500,
"total_flos": 0.0,
"trial_name": null,
"trial_params": null
}