| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.7904487443392343, | |
| "eval_steps": 500, | |
| "global_step": 60, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.013174145738987238, | |
| "grad_norm": 0.5370480418205261, | |
| "learning_rate": 1.25e-07, | |
| "logits/chosen": 10.071717262268066, | |
| "logits/rejected": 10.610974311828613, | |
| "logps/chosen": -0.5208129286766052, | |
| "logps/rejected": -0.5466158390045166, | |
| "loss": 1.3666, | |
| "rewards/accuracies": 0.5703125, | |
| "rewards/chosen": -1.0416258573532104, | |
| "rewards/margins": 0.05160584673285484, | |
| "rewards/rejected": -1.0932316780090332, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.026348291477974475, | |
| "grad_norm": 0.6556285619735718, | |
| "learning_rate": 2.5e-07, | |
| "logits/chosen": 10.222262382507324, | |
| "logits/rejected": 10.754176139831543, | |
| "logps/chosen": -0.5501401424407959, | |
| "logps/rejected": -0.580172598361969, | |
| "loss": 1.363, | |
| "rewards/accuracies": 0.546875, | |
| "rewards/chosen": -1.1002802848815918, | |
| "rewards/margins": 0.06006482243537903, | |
| "rewards/rejected": -1.160345196723938, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.03952243721696171, | |
| "grad_norm": 0.5852232575416565, | |
| "learning_rate": 3.75e-07, | |
| "logits/chosen": 10.029905319213867, | |
| "logits/rejected": 10.635485649108887, | |
| "logps/chosen": -0.6488087773323059, | |
| "logps/rejected": -0.6685145497322083, | |
| "loss": 1.3891, | |
| "rewards/accuracies": 0.484375, | |
| "rewards/chosen": -1.2976175546646118, | |
| "rewards/margins": 0.03941144049167633, | |
| "rewards/rejected": -1.3370290994644165, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05269658295594895, | |
| "grad_norm": 0.880297839641571, | |
| "learning_rate": 5e-07, | |
| "logits/chosen": 9.950821876525879, | |
| "logits/rejected": 10.651039123535156, | |
| "logps/chosen": -0.5544578433036804, | |
| "logps/rejected": -0.5768871307373047, | |
| "loss": 1.3895, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -1.1089156866073608, | |
| "rewards/margins": 0.04485849663615227, | |
| "rewards/rejected": -1.1537742614746094, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06587072869493618, | |
| "grad_norm": 1.0345523357391357, | |
| "learning_rate": 6.249999999999999e-07, | |
| "logits/chosen": 10.145177841186523, | |
| "logits/rejected": 10.952865600585938, | |
| "logps/chosen": -0.5600388646125793, | |
| "logps/rejected": -0.5631997585296631, | |
| "loss": 1.3982, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -1.1200777292251587, | |
| "rewards/margins": 0.006321802735328674, | |
| "rewards/rejected": -1.1263995170593262, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.07904487443392343, | |
| "grad_norm": 0.9534112811088562, | |
| "learning_rate": 7.5e-07, | |
| "logits/chosen": 10.677449226379395, | |
| "logits/rejected": 11.309142112731934, | |
| "logps/chosen": -0.5716575384140015, | |
| "logps/rejected": -0.6011527180671692, | |
| "loss": 1.3753, | |
| "rewards/accuracies": 0.5859375, | |
| "rewards/chosen": -1.143315076828003, | |
| "rewards/margins": 0.05899051949381828, | |
| "rewards/rejected": -1.2023054361343384, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.09221902017291066, | |
| "grad_norm": 0.7576708793640137, | |
| "learning_rate": 8.75e-07, | |
| "logits/chosen": 9.934803009033203, | |
| "logits/rejected": 10.656447410583496, | |
| "logps/chosen": -0.5172385573387146, | |
| "logps/rejected": -0.5407333374023438, | |
| "loss": 1.3735, | |
| "rewards/accuracies": 0.5234375, | |
| "rewards/chosen": -1.0344771146774292, | |
| "rewards/margins": 0.04698962718248367, | |
| "rewards/rejected": -1.0814666748046875, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.1053931659118979, | |
| "grad_norm": 0.6589964628219604, | |
| "learning_rate": 1e-06, | |
| "logits/chosen": 10.398344039916992, | |
| "logits/rejected": 11.005778312683105, | |
| "logps/chosen": -0.5838003754615784, | |
| "logps/rejected": -0.5947949290275574, | |
| "loss": 1.396, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -1.1676007509231567, | |
| "rewards/margins": 0.02198910340666771, | |
| "rewards/rejected": -1.1895898580551147, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.11856731165088513, | |
| "grad_norm": 0.7832041382789612, | |
| "learning_rate": 9.994504457428556e-07, | |
| "logits/chosen": 10.207743644714355, | |
| "logits/rejected": 11.032607078552246, | |
| "logps/chosen": -0.6102317571640015, | |
| "logps/rejected": -0.5968730449676514, | |
| "loss": 1.4486, | |
| "rewards/accuracies": 0.5234375, | |
| "rewards/chosen": -1.220463514328003, | |
| "rewards/margins": -0.02671748585999012, | |
| "rewards/rejected": -1.1937460899353027, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.13174145738987236, | |
| "grad_norm": 1.3487299680709839, | |
| "learning_rate": 9.97802991010949e-07, | |
| "logits/chosen": 9.977300643920898, | |
| "logits/rejected": 10.761427879333496, | |
| "logps/chosen": -0.49488401412963867, | |
| "logps/rejected": -0.47264525294303894, | |
| "loss": 1.4363, | |
| "rewards/accuracies": 0.5078125, | |
| "rewards/chosen": -0.9897680282592773, | |
| "rewards/margins": -0.04447745531797409, | |
| "rewards/rejected": -0.9452905058860779, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.14491560312885962, | |
| "grad_norm": 0.9786595702171326, | |
| "learning_rate": 9.950612572673255e-07, | |
| "logits/chosen": 10.096546173095703, | |
| "logits/rejected": 10.999103546142578, | |
| "logps/chosen": -0.6210685968399048, | |
| "logps/rejected": -0.6260945796966553, | |
| "loss": 1.4145, | |
| "rewards/accuracies": 0.4453125, | |
| "rewards/chosen": -1.2421371936798096, | |
| "rewards/margins": 0.010051850229501724, | |
| "rewards/rejected": -1.2521891593933105, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.15808974886784685, | |
| "grad_norm": 0.9959055781364441, | |
| "learning_rate": 9.912312714377879e-07, | |
| "logits/chosen": 10.130002975463867, | |
| "logits/rejected": 10.666738510131836, | |
| "logps/chosen": -0.5434242486953735, | |
| "logps/rejected": -0.5788334608078003, | |
| "loss": 1.3496, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -1.086848497390747, | |
| "rewards/margins": 0.07081828266382217, | |
| "rewards/rejected": -1.1576669216156006, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.17126389460683408, | |
| "grad_norm": 0.7984468340873718, | |
| "learning_rate": 9.863214526624063e-07, | |
| "logits/chosen": 9.885704040527344, | |
| "logits/rejected": 10.887924194335938, | |
| "logps/chosen": -0.5868184566497803, | |
| "logps/rejected": -0.5810582041740417, | |
| "loss": 1.4268, | |
| "rewards/accuracies": 0.4453125, | |
| "rewards/chosen": -1.1736369132995605, | |
| "rewards/margins": -0.011520491912961006, | |
| "rewards/rejected": -1.1621164083480835, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.1844380403458213, | |
| "grad_norm": 1.042403221130371, | |
| "learning_rate": 9.8034259378842e-07, | |
| "logits/chosen": 10.426769256591797, | |
| "logits/rejected": 11.37084674835205, | |
| "logps/chosen": -0.5749006271362305, | |
| "logps/rejected": -0.6211482882499695, | |
| "loss": 1.3485, | |
| "rewards/accuracies": 0.5703125, | |
| "rewards/chosen": -1.149801254272461, | |
| "rewards/margins": 0.09249527752399445, | |
| "rewards/rejected": -1.242296576499939, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.19761218608480857, | |
| "grad_norm": 1.288346290588379, | |
| "learning_rate": 9.73307837645217e-07, | |
| "logits/chosen": 9.93845272064209, | |
| "logits/rejected": 10.669981002807617, | |
| "logps/chosen": -0.5796323418617249, | |
| "logps/rejected": -0.5988457202911377, | |
| "loss": 1.3761, | |
| "rewards/accuracies": 0.4765625, | |
| "rewards/chosen": -1.1592646837234497, | |
| "rewards/margins": 0.03842674568295479, | |
| "rewards/rejected": -1.1976914405822754, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.2107863318237958, | |
| "grad_norm": 1.540350079536438, | |
| "learning_rate": 9.652326481535433e-07, | |
| "logits/chosen": 10.660343170166016, | |
| "logits/rejected": 11.23757553100586, | |
| "logps/chosen": -0.5483916997909546, | |
| "logps/rejected": -0.5503613948822021, | |
| "loss": 1.3984, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -1.0967833995819092, | |
| "rewards/margins": 0.003939321264624596, | |
| "rewards/rejected": -1.1007227897644043, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.22396047756278303, | |
| "grad_norm": 1.0571225881576538, | |
| "learning_rate": 9.561347763324483e-07, | |
| "logits/chosen": 10.136300086975098, | |
| "logits/rejected": 10.82028579711914, | |
| "logps/chosen": -0.5688516497612, | |
| "logps/rejected": -0.553267240524292, | |
| "loss": 1.4383, | |
| "rewards/accuracies": 0.546875, | |
| "rewards/chosen": -1.1377032995224, | |
| "rewards/margins": -0.031168799847364426, | |
| "rewards/rejected": -1.106534481048584, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.23713462330177026, | |
| "grad_norm": 1.095320463180542, | |
| "learning_rate": 9.460342212786932e-07, | |
| "logits/chosen": 10.693653106689453, | |
| "logits/rejected": 11.115797996520996, | |
| "logps/chosen": -0.6066192388534546, | |
| "logps/rejected": -0.5900355577468872, | |
| "loss": 1.4382, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -1.2132384777069092, | |
| "rewards/margins": -0.03316720202565193, | |
| "rewards/rejected": -1.1800711154937744, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2503087690407575, | |
| "grad_norm": 0.6557356119155884, | |
| "learning_rate": 9.349531862043951e-07, | |
| "logits/chosen": 10.337396621704102, | |
| "logits/rejected": 10.86961841583252, | |
| "logps/chosen": -0.525456964969635, | |
| "logps/rejected": -0.5541884899139404, | |
| "loss": 1.3703, | |
| "rewards/accuracies": 0.609375, | |
| "rewards/chosen": -1.05091392993927, | |
| "rewards/margins": 0.057463016360998154, | |
| "rewards/rejected": -1.1083769798278809, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.2634829147797447, | |
| "grad_norm": 0.7737817168235779, | |
| "learning_rate": 9.229160296295487e-07, | |
| "logits/chosen": 10.50332260131836, | |
| "logits/rejected": 11.313106536865234, | |
| "logps/chosen": -0.6586668491363525, | |
| "logps/rejected": -0.6300007104873657, | |
| "loss": 1.4657, | |
| "rewards/accuracies": 0.5390625, | |
| "rewards/chosen": -1.317333698272705, | |
| "rewards/margins": -0.05733209103345871, | |
| "rewards/rejected": -1.2600014209747314, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.276657060518732, | |
| "grad_norm": 0.5888201594352722, | |
| "learning_rate": 9.099492118367122e-07, | |
| "logits/chosen": 9.65850830078125, | |
| "logits/rejected": 10.547868728637695, | |
| "logps/chosen": -0.6071429252624512, | |
| "logps/rejected": -0.6548156142234802, | |
| "loss": 1.3526, | |
| "rewards/accuracies": 0.5390625, | |
| "rewards/chosen": -1.2142858505249023, | |
| "rewards/margins": 0.09534531086683273, | |
| "rewards/rejected": -1.3096312284469604, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.28983120625771924, | |
| "grad_norm": 1.3495151996612549, | |
| "learning_rate": 8.960812367055646e-07, | |
| "logits/chosen": 9.991419792175293, | |
| "logits/rejected": 10.768990516662598, | |
| "logps/chosen": -0.6086853742599487, | |
| "logps/rejected": -0.6002129912376404, | |
| "loss": 1.4365, | |
| "rewards/accuracies": 0.4375, | |
| "rewards/chosen": -1.2173707485198975, | |
| "rewards/margins": -0.01694471389055252, | |
| "rewards/rejected": -1.2004259824752808, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.3030053519967065, | |
| "grad_norm": 0.40277352929115295, | |
| "learning_rate": 8.813425890551909e-07, | |
| "logits/chosen": 9.981124877929688, | |
| "logits/rejected": 10.845569610595703, | |
| "logps/chosen": -0.6795557141304016, | |
| "logps/rejected": -0.630460262298584, | |
| "loss": 1.5098, | |
| "rewards/accuracies": 0.5078125, | |
| "rewards/chosen": -1.3591114282608032, | |
| "rewards/margins": -0.0981910228729248, | |
| "rewards/rejected": -1.260920524597168, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.3161794977356937, | |
| "grad_norm": 0.738304853439331, | |
| "learning_rate": 8.657656676318345e-07, | |
| "logits/chosen": 10.05403995513916, | |
| "logits/rejected": 10.649165153503418, | |
| "logps/chosen": -0.5632920265197754, | |
| "logps/rejected": -0.5841237902641296, | |
| "loss": 1.3813, | |
| "rewards/accuracies": 0.515625, | |
| "rewards/chosen": -1.1265840530395508, | |
| "rewards/margins": 0.0416635125875473, | |
| "rewards/rejected": -1.1682475805282593, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.32935364347468093, | |
| "grad_norm": 1.1053329706192017, | |
| "learning_rate": 8.493847138894208e-07, | |
| "logits/chosen": 10.050838470458984, | |
| "logits/rejected": 10.84270191192627, | |
| "logps/chosen": -0.597644567489624, | |
| "logps/rejected": -0.595725417137146, | |
| "loss": 1.4282, | |
| "rewards/accuracies": 0.515625, | |
| "rewards/chosen": -1.195289134979248, | |
| "rewards/margins": -0.0038383007049560547, | |
| "rewards/rejected": -1.191450834274292, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.34252778921366817, | |
| "grad_norm": 1.4238759279251099, | |
| "learning_rate": 8.322357367194108e-07, | |
| "logits/chosen": 10.223604202270508, | |
| "logits/rejected": 11.166160583496094, | |
| "logps/chosen": -0.6837385892868042, | |
| "logps/rejected": -0.7063555717468262, | |
| "loss": 1.4015, | |
| "rewards/accuracies": 0.5078125, | |
| "rewards/chosen": -1.3674771785736084, | |
| "rewards/margins": 0.04523405432701111, | |
| "rewards/rejected": -1.4127111434936523, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.3557019349526554, | |
| "grad_norm": 2.064854860305786, | |
| "learning_rate": 8.143564332954425e-07, | |
| "logits/chosen": 10.116180419921875, | |
| "logits/rejected": 11.071605682373047, | |
| "logps/chosen": -0.581619381904602, | |
| "logps/rejected": -0.5917808413505554, | |
| "loss": 1.4029, | |
| "rewards/accuracies": 0.4921875, | |
| "rewards/chosen": -1.163238763809204, | |
| "rewards/margins": 0.020322859287261963, | |
| "rewards/rejected": -1.1835616827011108, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.3688760806916426, | |
| "grad_norm": 0.7787203788757324, | |
| "learning_rate": 7.957861062067612e-07, | |
| "logits/chosen": 9.396055221557617, | |
| "logits/rejected": 10.454724311828613, | |
| "logps/chosen": -0.5871963500976562, | |
| "logps/rejected": -0.575207531452179, | |
| "loss": 1.4384, | |
| "rewards/accuracies": 0.515625, | |
| "rewards/chosen": -1.1743927001953125, | |
| "rewards/margins": -0.023977583274245262, | |
| "rewards/rejected": -1.150415062904358, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.3820502264306299, | |
| "grad_norm": 0.6901921629905701, | |
| "learning_rate": 7.765655770625996e-07, | |
| "logits/chosen": 10.049378395080566, | |
| "logits/rejected": 10.916838645935059, | |
| "logps/chosen": -0.559207558631897, | |
| "logps/rejected": -0.5522322654724121, | |
| "loss": 1.4303, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -1.118415117263794, | |
| "rewards/margins": -0.013950513675808907, | |
| "rewards/rejected": -1.1044645309448242, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.39522437216961714, | |
| "grad_norm": 0.40703555941581726, | |
| "learning_rate": 7.567370967574209e-07, | |
| "logits/chosen": 10.099987030029297, | |
| "logits/rejected": 11.389348983764648, | |
| "logps/chosen": -0.6557003855705261, | |
| "logps/rejected": -0.637586236000061, | |
| "loss": 1.4561, | |
| "rewards/accuracies": 0.4921875, | |
| "rewards/chosen": -1.3114007711410522, | |
| "rewards/margins": -0.03622820973396301, | |
| "rewards/rejected": -1.275172472000122, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4083985179086044, | |
| "grad_norm": 0.42555299401283264, | |
| "learning_rate": 7.363442525942826e-07, | |
| "logits/chosen": 9.72238826751709, | |
| "logits/rejected": 10.818886756896973, | |
| "logps/chosen": -0.5414553284645081, | |
| "logps/rejected": -0.5017333626747131, | |
| "loss": 1.4659, | |
| "rewards/accuracies": 0.4921875, | |
| "rewards/chosen": -1.0829106569290161, | |
| "rewards/margins": -0.07944400608539581, | |
| "rewards/rejected": -1.0034667253494263, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.4215726636475916, | |
| "grad_norm": 0.5622223615646362, | |
| "learning_rate": 7.154318724704851e-07, | |
| "logits/chosen": 9.967964172363281, | |
| "logits/rejected": 11.171170234680176, | |
| "logps/chosen": -0.5367560386657715, | |
| "logps/rejected": -0.5056421756744385, | |
| "loss": 1.4642, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -1.073512077331543, | |
| "rewards/margins": -0.06222765892744064, | |
| "rewards/rejected": -1.011284351348877, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.43474680938657884, | |
| "grad_norm": 1.4852371215820312, | |
| "learning_rate": 6.940459263361248e-07, | |
| "logits/chosen": 9.997882843017578, | |
| "logits/rejected": 10.80778980255127, | |
| "logps/chosen": -0.5545799732208252, | |
| "logps/rejected": -0.5040555000305176, | |
| "loss": 1.4922, | |
| "rewards/accuracies": 0.4765625, | |
| "rewards/chosen": -1.1091599464416504, | |
| "rewards/margins": -0.10104897618293762, | |
| "rewards/rejected": -1.0081110000610352, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.44792095512556607, | |
| "grad_norm": 0.7598586082458496, | |
| "learning_rate": 6.722334251421664e-07, | |
| "logits/chosen": 9.955677032470703, | |
| "logits/rejected": 10.815716743469238, | |
| "logps/chosen": -0.6858883500099182, | |
| "logps/rejected": -0.6911861896514893, | |
| "loss": 1.4234, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -1.3717767000198364, | |
| "rewards/margins": 0.010595660656690598, | |
| "rewards/rejected": -1.3823723793029785, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.4610951008645533, | |
| "grad_norm": 0.6476173996925354, | |
| "learning_rate": 6.500423175001703e-07, | |
| "logits/chosen": 10.530078887939453, | |
| "logits/rejected": 11.22971248626709, | |
| "logps/chosen": -0.532319962978363, | |
| "logps/rejected": -0.5957823395729065, | |
| "loss": 1.3528, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -1.064639925956726, | |
| "rewards/margins": 0.12692469358444214, | |
| "rewards/rejected": -1.191564679145813, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.47426924660354053, | |
| "grad_norm": 0.7018894553184509, | |
| "learning_rate": 6.275213842808382e-07, | |
| "logits/chosen": 10.111711502075195, | |
| "logits/rejected": 10.461627006530762, | |
| "logps/chosen": -0.5287503004074097, | |
| "logps/rejected": -0.5425370335578918, | |
| "loss": 1.3894, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -1.0575006008148193, | |
| "rewards/margins": 0.027573533356189728, | |
| "rewards/rejected": -1.0850740671157837, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.4874433923425278, | |
| "grad_norm": 0.5113664269447327, | |
| "learning_rate": 6.047201313830723e-07, | |
| "logits/chosen": 10.279487609863281, | |
| "logits/rejected": 10.891258239746094, | |
| "logps/chosen": -0.5863273739814758, | |
| "logps/rejected": -0.6054317951202393, | |
| "loss": 1.3829, | |
| "rewards/accuracies": 0.484375, | |
| "rewards/chosen": -1.1726547479629517, | |
| "rewards/margins": 0.03820895403623581, | |
| "rewards/rejected": -1.2108635902404785, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.500617538081515, | |
| "grad_norm": 0.8292818069458008, | |
| "learning_rate": 5.816886809092651e-07, | |
| "logits/chosen": 9.993082046508789, | |
| "logits/rejected": 10.848751068115234, | |
| "logps/chosen": -0.5642987489700317, | |
| "logps/rejected": -0.6022392511367798, | |
| "loss": 1.3685, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -1.1285974979400635, | |
| "rewards/margins": 0.07588109374046326, | |
| "rewards/rejected": -1.2044785022735596, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.5137916838205022, | |
| "grad_norm": 2.4941494464874268, | |
| "learning_rate": 5.584776609860413e-07, | |
| "logits/chosen": 10.164732933044434, | |
| "logits/rejected": 10.930498123168945, | |
| "logps/chosen": -0.5763741135597229, | |
| "logps/rejected": -0.5431419610977173, | |
| "loss": 1.4765, | |
| "rewards/accuracies": 0.46875, | |
| "rewards/chosen": -1.1527482271194458, | |
| "rewards/margins": -0.06646420061588287, | |
| "rewards/rejected": -1.0862839221954346, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.5269658295594895, | |
| "grad_norm": 3.3130643367767334, | |
| "learning_rate": 5.351380944726465e-07, | |
| "logits/chosen": 10.054292678833008, | |
| "logits/rejected": 11.316207885742188, | |
| "logps/chosen": -0.5626562833786011, | |
| "logps/rejected": -0.5585920810699463, | |
| "loss": 1.4175, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -1.1253125667572021, | |
| "rewards/margins": -0.008128294721245766, | |
| "rewards/rejected": -1.1171841621398926, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.5401399752984768, | |
| "grad_norm": 0.6443184018135071, | |
| "learning_rate": 5.117212868016303e-07, | |
| "logits/chosen": 10.326967239379883, | |
| "logits/rejected": 10.83257007598877, | |
| "logps/chosen": -0.5476542115211487, | |
| "logps/rejected": -0.5671020746231079, | |
| "loss": 1.3819, | |
| "rewards/accuracies": 0.5234375, | |
| "rewards/chosen": -1.0953084230422974, | |
| "rewards/margins": 0.038895681500434875, | |
| "rewards/rejected": -1.1342041492462158, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.553314121037464, | |
| "grad_norm": 1.2466775178909302, | |
| "learning_rate": 4.882787131983697e-07, | |
| "logits/chosen": 9.467147827148438, | |
| "logits/rejected": 10.408858299255371, | |
| "logps/chosen": -0.5122556686401367, | |
| "logps/rejected": -0.5029544234275818, | |
| "loss": 1.4341, | |
| "rewards/accuracies": 0.4296875, | |
| "rewards/chosen": -1.0245113372802734, | |
| "rewards/margins": -0.01860247366130352, | |
| "rewards/rejected": -1.0059088468551636, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.5664882667764513, | |
| "grad_norm": 1.0065938234329224, | |
| "learning_rate": 4.648619055273537e-07, | |
| "logits/chosen": 9.575172424316406, | |
| "logits/rejected": 10.501733779907227, | |
| "logps/chosen": -0.4882752299308777, | |
| "logps/rejected": -0.5106842517852783, | |
| "loss": 1.3748, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -0.9765504598617554, | |
| "rewards/margins": 0.044818080961704254, | |
| "rewards/rejected": -1.0213685035705566, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.5796624125154385, | |
| "grad_norm": 0.7030373215675354, | |
| "learning_rate": 4.4152233901395875e-07, | |
| "logits/chosen": 9.824498176574707, | |
| "logits/rejected": 10.55920696258545, | |
| "logps/chosen": -0.545955240726471, | |
| "logps/rejected": -0.5752911567687988, | |
| "loss": 1.3742, | |
| "rewards/accuracies": 0.6328125, | |
| "rewards/chosen": -1.091910481452942, | |
| "rewards/margins": 0.05867195501923561, | |
| "rewards/rejected": -1.1505823135375977, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.5928365582544257, | |
| "grad_norm": 0.6114833354949951, | |
| "learning_rate": 4.183113190907348e-07, | |
| "logits/chosen": 9.935275077819824, | |
| "logits/rejected": 10.47735595703125, | |
| "logps/chosen": -0.5813839435577393, | |
| "logps/rejected": -0.5967380404472351, | |
| "loss": 1.4037, | |
| "rewards/accuracies": 0.53125, | |
| "rewards/chosen": -1.1627678871154785, | |
| "rewards/margins": 0.030708128586411476, | |
| "rewards/rejected": -1.1934760808944702, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.606010703993413, | |
| "grad_norm": 4.388708114624023, | |
| "learning_rate": 3.9527986861692785e-07, | |
| "logits/chosen": 9.954986572265625, | |
| "logits/rejected": 10.432456016540527, | |
| "logps/chosen": -0.5504661798477173, | |
| "logps/rejected": -0.5747936964035034, | |
| "loss": 1.3774, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -1.1009323596954346, | |
| "rewards/margins": 0.04865497350692749, | |
| "rewards/rejected": -1.1495873928070068, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.6191848497324002, | |
| "grad_norm": 0.6564725041389465, | |
| "learning_rate": 3.724786157191618e-07, | |
| "logits/chosen": 9.773380279541016, | |
| "logits/rejected": 10.888383865356445, | |
| "logps/chosen": -0.6393429040908813, | |
| "logps/rejected": -0.6083090305328369, | |
| "loss": 1.4628, | |
| "rewards/accuracies": 0.484375, | |
| "rewards/chosen": -1.2786858081817627, | |
| "rewards/margins": -0.06206771731376648, | |
| "rewards/rejected": -1.2166180610656738, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.6323589954713874, | |
| "grad_norm": 1.6401126384735107, | |
| "learning_rate": 3.499576824998297e-07, | |
| "logits/chosen": 10.472443580627441, | |
| "logits/rejected": 11.252219200134277, | |
| "logps/chosen": -0.5614030361175537, | |
| "logps/rejected": -0.582251787185669, | |
| "loss": 1.3921, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": -1.1228060722351074, | |
| "rewards/margins": 0.04169747233390808, | |
| "rewards/rejected": -1.164503574371338, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.6455331412103746, | |
| "grad_norm": 0.5104884505271912, | |
| "learning_rate": 3.2776657485783356e-07, | |
| "logits/chosen": 10.491229057312012, | |
| "logits/rejected": 11.15767765045166, | |
| "logps/chosen": -0.5550377368927002, | |
| "logps/rejected": -0.5468860864639282, | |
| "loss": 1.4177, | |
| "rewards/accuracies": 0.5, | |
| "rewards/chosen": -1.1100754737854004, | |
| "rewards/margins": -0.016303202137351036, | |
| "rewards/rejected": -1.0937721729278564, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.6587072869493619, | |
| "grad_norm": 0.5247014164924622, | |
| "learning_rate": 3.0595407366387506e-07, | |
| "logits/chosen": 10.169745445251465, | |
| "logits/rejected": 10.953508377075195, | |
| "logps/chosen": -0.4428662955760956, | |
| "logps/rejected": -0.4541059136390686, | |
| "loss": 1.3806, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -0.8857325911521912, | |
| "rewards/margins": 0.02247920073568821, | |
| "rewards/rejected": -0.9082118272781372, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.6718814326883491, | |
| "grad_norm": 0.28417685627937317, | |
| "learning_rate": 2.845681275295148e-07, | |
| "logits/chosen": 10.14460563659668, | |
| "logits/rejected": 10.639837265014648, | |
| "logps/chosen": -0.5080307126045227, | |
| "logps/rejected": -0.5019209384918213, | |
| "loss": 1.4123, | |
| "rewards/accuracies": 0.5546875, | |
| "rewards/chosen": -1.0160614252090454, | |
| "rewards/margins": -0.012219585478305817, | |
| "rewards/rejected": -1.0038418769836426, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.6850555784273363, | |
| "grad_norm": 0.40843820571899414, | |
| "learning_rate": 2.636557474057173e-07, | |
| "logits/chosen": 9.437065124511719, | |
| "logits/rejected": 10.381149291992188, | |
| "logps/chosen": -0.5462982654571533, | |
| "logps/rejected": -0.5879113674163818, | |
| "loss": 1.3609, | |
| "rewards/accuracies": 0.578125, | |
| "rewards/chosen": -1.0925965309143066, | |
| "rewards/margins": 0.0832262635231018, | |
| "rewards/rejected": -1.1758227348327637, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.6982297241663236, | |
| "grad_norm": 0.5657537579536438, | |
| "learning_rate": 2.432629032425789e-07, | |
| "logits/chosen": 10.003761291503906, | |
| "logits/rejected": 10.856680870056152, | |
| "logps/chosen": -0.4738156497478485, | |
| "logps/rejected": -0.5128645896911621, | |
| "loss": 1.3441, | |
| "rewards/accuracies": 0.5390625, | |
| "rewards/chosen": -0.947631299495697, | |
| "rewards/margins": 0.07809784263372421, | |
| "rewards/rejected": -1.0257291793823242, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.7114038699053108, | |
| "grad_norm": 0.7596526741981506, | |
| "learning_rate": 2.2343442293740028e-07, | |
| "logits/chosen": 10.338200569152832, | |
| "logits/rejected": 11.057929992675781, | |
| "logps/chosen": -0.6268498301506042, | |
| "logps/rejected": -0.6259283423423767, | |
| "loss": 1.4169, | |
| "rewards/accuracies": 0.4765625, | |
| "rewards/chosen": -1.2536996603012085, | |
| "rewards/margins": -0.001843004021793604, | |
| "rewards/rejected": -1.2518566846847534, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.724578015644298, | |
| "grad_norm": 0.5075994729995728, | |
| "learning_rate": 2.0421389379323877e-07, | |
| "logits/chosen": 10.308176040649414, | |
| "logits/rejected": 10.985513687133789, | |
| "logps/chosen": -0.578580379486084, | |
| "logps/rejected": -0.6193315386772156, | |
| "loss": 1.3714, | |
| "rewards/accuracies": 0.578125, | |
| "rewards/chosen": -1.157160758972168, | |
| "rewards/margins": 0.08150239288806915, | |
| "rewards/rejected": -1.2386630773544312, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.7377521613832853, | |
| "grad_norm": 0.7418858408927917, | |
| "learning_rate": 1.8564356670455767e-07, | |
| "logits/chosen": 9.866720199584961, | |
| "logits/rejected": 10.861518859863281, | |
| "logps/chosen": -0.5241114497184753, | |
| "logps/rejected": -0.5731189250946045, | |
| "loss": 1.35, | |
| "rewards/accuracies": 0.5703125, | |
| "rewards/chosen": -1.0482228994369507, | |
| "rewards/margins": 0.09801498055458069, | |
| "rewards/rejected": -1.146237850189209, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.7509263071222725, | |
| "grad_norm": 1.9169734716415405, | |
| "learning_rate": 1.6776426328058919e-07, | |
| "logits/chosen": 10.54930305480957, | |
| "logits/rejected": 11.324594497680664, | |
| "logps/chosen": -0.5696872472763062, | |
| "logps/rejected": -0.5661669969558716, | |
| "loss": 1.4095, | |
| "rewards/accuracies": 0.546875, | |
| "rewards/chosen": -1.1393744945526123, | |
| "rewards/margins": -0.007040377240628004, | |
| "rewards/rejected": -1.1323339939117432, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.7641004528612598, | |
| "grad_norm": 1.0950486660003662, | |
| "learning_rate": 1.5061528611057915e-07, | |
| "logits/chosen": 9.858892440795898, | |
| "logits/rejected": 10.457954406738281, | |
| "logps/chosen": -0.524871289730072, | |
| "logps/rejected": -0.5936727523803711, | |
| "loss": 1.3407, | |
| "rewards/accuracies": 0.6328125, | |
| "rewards/chosen": -1.049742579460144, | |
| "rewards/margins": 0.13760297000408173, | |
| "rewards/rejected": -1.1873455047607422, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.7772745986002471, | |
| "grad_norm": 0.5938051342964172, | |
| "learning_rate": 1.3423433236816562e-07, | |
| "logits/chosen": 10.020545959472656, | |
| "logits/rejected": 10.919637680053711, | |
| "logps/chosen": -0.5659824013710022, | |
| "logps/rejected": -0.5747073292732239, | |
| "loss": 1.406, | |
| "rewards/accuracies": 0.578125, | |
| "rewards/chosen": -1.1319648027420044, | |
| "rewards/margins": 0.017449870705604553, | |
| "rewards/rejected": -1.1494146585464478, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.7904487443392343, | |
| "grad_norm": 0.42677873373031616, | |
| "learning_rate": 1.1865741094480908e-07, | |
| "logits/chosen": 10.143865585327148, | |
| "logits/rejected": 11.086860656738281, | |
| "logps/chosen": -0.514181911945343, | |
| "logps/rejected": -0.5444334745407104, | |
| "loss": 1.3637, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": -1.028363823890686, | |
| "rewards/margins": 0.060503169894218445, | |
| "rewards/rejected": -1.088866949081421, | |
| "step": 60 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 75, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 12, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |