Invalid JSON:Unexpected token 'N', ..."ad_norm": NaN,
"... is not valid JSON
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 1407, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.010666666666666666, | |
| "grad_norm": 5.690470218658447, | |
| "learning_rate": 4.651162790697675e-07, | |
| "logits/chosen": -1.3886802196502686, | |
| "logits/rejected": -1.288570761680603, | |
| "logps/chosen": -198.13186645507812, | |
| "logps/rejected": -261.1497802734375, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.3125, | |
| "rewards/chosen": -0.00018911839288193733, | |
| "rewards/margins": 0.00016672378114890307, | |
| "rewards/rejected": -0.0003558421158231795, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.021333333333333333, | |
| "grad_norm": 2.7565765380859375, | |
| "learning_rate": 1.0465116279069768e-06, | |
| "logits/chosen": -1.3378444910049438, | |
| "logits/rejected": -1.2759315967559814, | |
| "logps/chosen": -206.43960571289062, | |
| "logps/rejected": -279.51348876953125, | |
| "loss": 0.691, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": 0.001392951002344489, | |
| "rewards/margins": 0.0042605213820934296, | |
| "rewards/rejected": -0.002867569914087653, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.032, | |
| "grad_norm": 3.2212350368499756, | |
| "learning_rate": 1.6279069767441862e-06, | |
| "logits/chosen": -1.3505547046661377, | |
| "logits/rejected": -1.2644460201263428, | |
| "logps/chosen": -206.5700225830078, | |
| "logps/rejected": -259.0060729980469, | |
| "loss": 0.6853, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": 0.0011921453988179564, | |
| "rewards/margins": 0.015914104878902435, | |
| "rewards/rejected": -0.01472195703536272, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.042666666666666665, | |
| "grad_norm": NaN, | |
| "learning_rate": 2.2093023255813954e-06, | |
| "logits/chosen": -1.4066317081451416, | |
| "logits/rejected": -1.2923839092254639, | |
| "logps/chosen": -223.789306640625, | |
| "logps/rejected": -258.15020751953125, | |
| "loss": 0.6749, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": 0.0023522863630205393, | |
| "rewards/margins": 0.03771216794848442, | |
| "rewards/rejected": -0.035359881818294525, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.05333333333333334, | |
| "grad_norm": 2.9260594844818115, | |
| "learning_rate": 2.790697674418605e-06, | |
| "logits/chosen": -1.350759506225586, | |
| "logits/rejected": -1.16400146484375, | |
| "logps/chosen": -203.46731567382812, | |
| "logps/rejected": -283.2779235839844, | |
| "loss": 0.658, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.0039240168407559395, | |
| "rewards/margins": 0.07307498157024384, | |
| "rewards/rejected": -0.0769990086555481, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.064, | |
| "grad_norm": 3.8544366359710693, | |
| "learning_rate": 3.372093023255814e-06, | |
| "logits/chosen": -1.4426774978637695, | |
| "logits/rejected": -1.3100595474243164, | |
| "logps/chosen": -264.96258544921875, | |
| "logps/rejected": -255.35922241210938, | |
| "loss": 0.6423, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.01604917272925377, | |
| "rewards/margins": 0.10922543704509735, | |
| "rewards/rejected": -0.12527461349964142, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.07466666666666667, | |
| "grad_norm": 2.8372976779937744, | |
| "learning_rate": 3.953488372093024e-06, | |
| "logits/chosen": -1.488498568534851, | |
| "logits/rejected": -1.4097141027450562, | |
| "logps/chosen": -210.7440643310547, | |
| "logps/rejected": -289.41046142578125, | |
| "loss": 0.6216, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.030715733766555786, | |
| "rewards/margins": 0.15629956126213074, | |
| "rewards/rejected": -0.1870153248310089, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08533333333333333, | |
| "grad_norm": 2.8906612396240234, | |
| "learning_rate": 4.5348837209302326e-06, | |
| "logits/chosen": -1.4510371685028076, | |
| "logits/rejected": -1.288226842880249, | |
| "logps/chosen": -207.5216064453125, | |
| "logps/rejected": -260.95416259765625, | |
| "loss": 0.6082, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.0654064267873764, | |
| "rewards/margins": 0.19735947251319885, | |
| "rewards/rejected": -0.26276591420173645, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.096, | |
| "grad_norm": 2.679903984069824, | |
| "learning_rate": 4.99999336897255e-06, | |
| "logits/chosen": -1.4529269933700562, | |
| "logits/rejected": -1.2380075454711914, | |
| "logps/chosen": -211.7548370361328, | |
| "logps/rejected": -255.37753295898438, | |
| "loss": 0.5597, | |
| "rewards/accuracies": 0.8187500238418579, | |
| "rewards/chosen": -0.07748017460107803, | |
| "rewards/margins": 0.32700955867767334, | |
| "rewards/rejected": -0.4044896960258484, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10666666666666667, | |
| "grad_norm": 5.634088039398193, | |
| "learning_rate": 4.999761286705285e-06, | |
| "logits/chosen": -1.4369691610336304, | |
| "logits/rejected": -1.2739875316619873, | |
| "logps/chosen": -221.0635528564453, | |
| "logps/rejected": -271.1972351074219, | |
| "loss": 0.5247, | |
| "rewards/accuracies": 0.84375, | |
| "rewards/chosen": -0.11849220842123032, | |
| "rewards/margins": 0.4242885112762451, | |
| "rewards/rejected": -0.5427806973457336, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11733333333333333, | |
| "grad_norm": 2.9434328079223633, | |
| "learning_rate": 4.999197688241076e-06, | |
| "logits/chosen": -1.5113855600357056, | |
| "logits/rejected": -1.3324053287506104, | |
| "logps/chosen": -202.31285095214844, | |
| "logps/rejected": -265.8841552734375, | |
| "loss": 0.491, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.10869394242763519, | |
| "rewards/margins": 0.5555111765861511, | |
| "rewards/rejected": -0.6642050743103027, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.128, | |
| "grad_norm": 3.161414623260498, | |
| "learning_rate": 4.9983026483238685e-06, | |
| "logits/chosen": -1.438110113143921, | |
| "logits/rejected": -1.2655045986175537, | |
| "logps/chosen": -211.4470977783203, | |
| "logps/rejected": -260.45208740234375, | |
| "loss": 0.4723, | |
| "rewards/accuracies": 0.8062499761581421, | |
| "rewards/chosen": -0.2564672529697418, | |
| "rewards/margins": 0.6585448980331421, | |
| "rewards/rejected": -0.9150121808052063, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13866666666666666, | |
| "grad_norm": 2.4061851501464844, | |
| "learning_rate": 4.997076285653087e-06, | |
| "logits/chosen": -1.5721609592437744, | |
| "logits/rejected": -1.3645689487457275, | |
| "logps/chosen": -247.45956420898438, | |
| "logps/rejected": -268.1156921386719, | |
| "loss": 0.5025, | |
| "rewards/accuracies": 0.768750011920929, | |
| "rewards/chosen": -0.37047362327575684, | |
| "rewards/margins": 0.6632369756698608, | |
| "rewards/rejected": -1.0337107181549072, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.14933333333333335, | |
| "grad_norm": 3.2865662574768066, | |
| "learning_rate": 4.995518762867897e-06, | |
| "logits/chosen": -1.4936473369598389, | |
| "logits/rejected": -1.2969480752944946, | |
| "logps/chosen": -206.06124877929688, | |
| "logps/rejected": -283.84466552734375, | |
| "loss": 0.4005, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.36351141333580017, | |
| "rewards/margins": 0.9965918660163879, | |
| "rewards/rejected": -1.3601032495498657, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 2.552584409713745, | |
| "learning_rate": 4.993630286525634e-06, | |
| "logits/chosen": -1.548957109451294, | |
| "logits/rejected": -1.4676214456558228, | |
| "logps/chosen": -203.4427947998047, | |
| "logps/rejected": -263.97698974609375, | |
| "loss": 0.414, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.3925500214099884, | |
| "rewards/margins": 0.9759253263473511, | |
| "rewards/rejected": -1.368475317955017, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.17066666666666666, | |
| "grad_norm": 2.2154462337493896, | |
| "learning_rate": 4.99141110707441e-06, | |
| "logits/chosen": -1.4633119106292725, | |
| "logits/rejected": -1.3160654306411743, | |
| "logps/chosen": -214.0872039794922, | |
| "logps/rejected": -276.56048583984375, | |
| "loss": 0.3877, | |
| "rewards/accuracies": 0.8374999761581421, | |
| "rewards/chosen": -0.44031819701194763, | |
| "rewards/margins": 1.116113305091858, | |
| "rewards/rejected": -1.556431531906128, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18133333333333335, | |
| "grad_norm": 2.944103479385376, | |
| "learning_rate": 4.9888615188199e-06, | |
| "logits/chosen": -1.5001494884490967, | |
| "logits/rejected": -1.3682790994644165, | |
| "logps/chosen": -207.25057983398438, | |
| "logps/rejected": -285.5174560546875, | |
| "loss": 0.3776, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.530932605266571, | |
| "rewards/margins": 1.2580114603042603, | |
| "rewards/rejected": -1.7889440059661865, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.192, | |
| "grad_norm": 2.180025339126587, | |
| "learning_rate": 4.9859818598863104e-06, | |
| "logits/chosen": -1.5337731838226318, | |
| "logits/rejected": -1.3553698062896729, | |
| "logps/chosen": -201.07247924804688, | |
| "logps/rejected": -264.13848876953125, | |
| "loss": 0.3508, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.4138107895851135, | |
| "rewards/margins": 1.3532263040542603, | |
| "rewards/rejected": -1.767037034034729, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20266666666666666, | |
| "grad_norm": 2.4048197269439697, | |
| "learning_rate": 4.982772512171538e-06, | |
| "logits/chosen": -1.6030489206314087, | |
| "logits/rejected": -1.501514196395874, | |
| "logps/chosen": -198.14962768554688, | |
| "logps/rejected": -262.2642517089844, | |
| "loss": 0.3704, | |
| "rewards/accuracies": 0.8500000238418579, | |
| "rewards/chosen": -0.35309070348739624, | |
| "rewards/margins": 1.3474071025848389, | |
| "rewards/rejected": -1.7004978656768799, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.21333333333333335, | |
| "grad_norm": 3.966644287109375, | |
| "learning_rate": 4.979233901296523e-06, | |
| "logits/chosen": -1.5420665740966797, | |
| "logits/rejected": -1.4332940578460693, | |
| "logps/chosen": -196.133056640625, | |
| "logps/rejected": -273.1980285644531, | |
| "loss": 0.3125, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": -0.418867826461792, | |
| "rewards/margins": 1.6799125671386719, | |
| "rewards/rejected": -2.0987801551818848, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.224, | |
| "grad_norm": 3.5509183406829834, | |
| "learning_rate": 4.975366496548804e-06, | |
| "logits/chosen": -1.6272293329238892, | |
| "logits/rejected": -1.396296739578247, | |
| "logps/chosen": -218.2744140625, | |
| "logps/rejected": -290.47222900390625, | |
| "loss": 0.3072, | |
| "rewards/accuracies": 0.8687499761581421, | |
| "rewards/chosen": -0.4910121560096741, | |
| "rewards/margins": 1.7056410312652588, | |
| "rewards/rejected": -2.196653127670288, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.23466666666666666, | |
| "grad_norm": 3.0346481800079346, | |
| "learning_rate": 4.971170810820279e-06, | |
| "logits/chosen": -1.5354888439178467, | |
| "logits/rejected": -1.4109222888946533, | |
| "logps/chosen": -227.25634765625, | |
| "logps/rejected": -285.1552734375, | |
| "loss": 0.3081, | |
| "rewards/accuracies": 0.875, | |
| "rewards/chosen": -0.5940927267074585, | |
| "rewards/margins": 1.8030599355697632, | |
| "rewards/rejected": -2.397152900695801, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.24533333333333332, | |
| "grad_norm": 2.6833956241607666, | |
| "learning_rate": 4.96664740053919e-06, | |
| "logits/chosen": -1.5940526723861694, | |
| "logits/rejected": -1.4896529912948608, | |
| "logps/chosen": -219.3295440673828, | |
| "logps/rejected": -300.86102294921875, | |
| "loss": 0.2949, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.6788660287857056, | |
| "rewards/margins": 1.8880093097686768, | |
| "rewards/rejected": -2.5668752193450928, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.256, | |
| "grad_norm": 2.7038156986236572, | |
| "learning_rate": 4.961796865596328e-06, | |
| "logits/chosen": -1.5613062381744385, | |
| "logits/rejected": -1.3715859651565552, | |
| "logps/chosen": -203.54425048828125, | |
| "logps/rejected": -283.4847412109375, | |
| "loss": 0.2571, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.3901914358139038, | |
| "rewards/margins": 2.150984287261963, | |
| "rewards/rejected": -2.5411758422851562, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.26666666666666666, | |
| "grad_norm": 3.027423858642578, | |
| "learning_rate": 4.956619849265473e-06, | |
| "logits/chosen": -1.6930328607559204, | |
| "logits/rejected": -1.507812261581421, | |
| "logps/chosen": -207.44992065429688, | |
| "logps/rejected": -286.91607666015625, | |
| "loss": 0.2844, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.6250053644180298, | |
| "rewards/margins": 1.9750416278839111, | |
| "rewards/rejected": -2.6000471115112305, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.2773333333333333, | |
| "grad_norm": 1.8523262739181519, | |
| "learning_rate": 4.951117038118091e-06, | |
| "logits/chosen": -1.655664086341858, | |
| "logits/rejected": -1.4919339418411255, | |
| "logps/chosen": -208.61209106445312, | |
| "logps/rejected": -279.31695556640625, | |
| "loss": 0.2413, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.5459321737289429, | |
| "rewards/margins": 2.22091007232666, | |
| "rewards/rejected": -2.7668423652648926, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.288, | |
| "grad_norm": 3.026885986328125, | |
| "learning_rate": 4.945289161932273e-06, | |
| "logits/chosen": -1.6513007879257202, | |
| "logits/rejected": -1.5145455598831177, | |
| "logps/chosen": -219.18704223632812, | |
| "logps/rejected": -293.86993408203125, | |
| "loss": 0.2555, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.6282526254653931, | |
| "rewards/margins": 2.321913957595825, | |
| "rewards/rejected": -2.950166702270508, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.2986666666666667, | |
| "grad_norm": 1.7478291988372803, | |
| "learning_rate": 4.93913699359596e-06, | |
| "logits/chosen": -1.662444829940796, | |
| "logits/rejected": -1.4403254985809326, | |
| "logps/chosen": -199.58511352539062, | |
| "logps/rejected": -296.8211364746094, | |
| "loss": 0.26, | |
| "rewards/accuracies": 0.893750011920929, | |
| "rewards/chosen": -0.5024666786193848, | |
| "rewards/margins": 2.4235363006591797, | |
| "rewards/rejected": -2.9260029792785645, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.30933333333333335, | |
| "grad_norm": 2.777461528778076, | |
| "learning_rate": 4.9326613490044364e-06, | |
| "logits/chosen": -1.6423578262329102, | |
| "logits/rejected": -1.5219056606292725, | |
| "logps/chosen": -211.6616668701172, | |
| "logps/rejected": -273.14337158203125, | |
| "loss": 0.2553, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.5515410304069519, | |
| "rewards/margins": 2.2121779918670654, | |
| "rewards/rejected": -2.763719081878662, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 2.59855055809021, | |
| "learning_rate": 4.925863086952132e-06, | |
| "logits/chosen": -1.6805490255355835, | |
| "logits/rejected": -1.4335730075836182, | |
| "logps/chosen": -219.8978729248047, | |
| "logps/rejected": -290.7606506347656, | |
| "loss": 0.284, | |
| "rewards/accuracies": 0.862500011920929, | |
| "rewards/chosen": -0.5696079134941101, | |
| "rewards/margins": 2.2634024620056152, | |
| "rewards/rejected": -2.833010196685791, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33066666666666666, | |
| "grad_norm": 2.194394111633301, | |
| "learning_rate": 4.918743109018727e-06, | |
| "logits/chosen": -1.7670875787734985, | |
| "logits/rejected": -1.5925384759902954, | |
| "logps/chosen": -198.6534881591797, | |
| "logps/rejected": -291.90155029296875, | |
| "loss": 0.1863, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.3875270187854767, | |
| "rewards/margins": 2.7576119899749756, | |
| "rewards/rejected": -3.145139217376709, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.3413333333333333, | |
| "grad_norm": 1.8163162469863892, | |
| "learning_rate": 4.911302359449585e-06, | |
| "logits/chosen": -1.767751693725586, | |
| "logits/rejected": -1.504002571105957, | |
| "logps/chosen": -224.132568359375, | |
| "logps/rejected": -301.08026123046875, | |
| "loss": 0.2234, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.610785722732544, | |
| "rewards/margins": 2.765779733657837, | |
| "rewards/rejected": -3.376565456390381, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.352, | |
| "grad_norm": 2.5315065383911133, | |
| "learning_rate": 4.9035418250305314e-06, | |
| "logits/chosen": -1.5424492359161377, | |
| "logits/rejected": -1.4309223890304565, | |
| "logps/chosen": -223.1277618408203, | |
| "logps/rejected": -289.2735290527344, | |
| "loss": 0.2373, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.6303970217704773, | |
| "rewards/margins": 2.7104063034057617, | |
| "rewards/rejected": -3.340803623199463, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3626666666666667, | |
| "grad_norm": 2.4730045795440674, | |
| "learning_rate": 4.895462534956981e-06, | |
| "logits/chosen": -1.701702356338501, | |
| "logits/rejected": -1.4543676376342773, | |
| "logps/chosen": -225.279052734375, | |
| "logps/rejected": -295.2032470703125, | |
| "loss": 0.2161, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.5884218215942383, | |
| "rewards/margins": 2.8115923404693604, | |
| "rewards/rejected": -3.4000144004821777, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.37333333333333335, | |
| "grad_norm": 2.8877482414245605, | |
| "learning_rate": 4.887065560697449e-06, | |
| "logits/chosen": -1.659183144569397, | |
| "logits/rejected": -1.4441927671432495, | |
| "logps/chosen": -220.6221466064453, | |
| "logps/rejected": -289.29254150390625, | |
| "loss": 0.2201, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.5501753687858582, | |
| "rewards/margins": 2.771678924560547, | |
| "rewards/rejected": -3.3218541145324707, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.384, | |
| "grad_norm": 2.3038156032562256, | |
| "learning_rate": 4.878352015851459e-06, | |
| "logits/chosen": -1.657855749130249, | |
| "logits/rejected": -1.468774676322937, | |
| "logps/chosen": -234.57339477539062, | |
| "logps/rejected": -275.18017578125, | |
| "loss": 0.2154, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": -0.5347112417221069, | |
| "rewards/margins": 2.8941256999969482, | |
| "rewards/rejected": -3.4288368225097656, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.39466666666666667, | |
| "grad_norm": 1.4912587404251099, | |
| "learning_rate": 4.869323056001852e-06, | |
| "logits/chosen": -1.6985185146331787, | |
| "logits/rejected": -1.4987022876739502, | |
| "logps/chosen": -215.89541625976562, | |
| "logps/rejected": -299.5584411621094, | |
| "loss": 0.2257, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.597873330116272, | |
| "rewards/margins": 2.936903476715088, | |
| "rewards/rejected": -3.5347766876220703, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.4053333333333333, | |
| "grad_norm": 1.758817434310913, | |
| "learning_rate": 4.859979878561536e-06, | |
| "logits/chosen": -1.6548064947128296, | |
| "logits/rejected": -1.480167031288147, | |
| "logps/chosen": -215.01345825195312, | |
| "logps/rejected": -302.09722900390625, | |
| "loss": 0.1969, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.7040780186653137, | |
| "rewards/margins": 2.9713985919952393, | |
| "rewards/rejected": -3.675476551055908, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.416, | |
| "grad_norm": 3.2202625274658203, | |
| "learning_rate": 4.850323722614685e-06, | |
| "logits/chosen": -1.6637979745864868, | |
| "logits/rejected": -1.534681797027588, | |
| "logps/chosen": -223.47787475585938, | |
| "logps/rejected": -298.9809875488281, | |
| "loss": 0.2318, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.7906392812728882, | |
| "rewards/margins": 2.9284045696258545, | |
| "rewards/rejected": -3.7190442085266113, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.4266666666666667, | |
| "grad_norm": 3.042365789413452, | |
| "learning_rate": 4.840355868752416e-06, | |
| "logits/chosen": -1.6638110876083374, | |
| "logits/rejected": -1.4790096282958984, | |
| "logps/chosen": -259.5401306152344, | |
| "logps/rejected": -279.5474853515625, | |
| "loss": 0.2538, | |
| "rewards/accuracies": 0.8812500238418579, | |
| "rewards/chosen": -0.3952628970146179, | |
| "rewards/margins": 2.784644603729248, | |
| "rewards/rejected": -3.179908037185669, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.43733333333333335, | |
| "grad_norm": 3.7375149726867676, | |
| "learning_rate": 4.830077638902955e-06, | |
| "logits/chosen": -1.772063970565796, | |
| "logits/rejected": -1.5635404586791992, | |
| "logps/chosen": -192.68406677246094, | |
| "logps/rejected": -279.232421875, | |
| "loss": 0.2313, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.6237959861755371, | |
| "rewards/margins": 3.2594666481018066, | |
| "rewards/rejected": -3.8832626342773438, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.448, | |
| "grad_norm": 2.7821362018585205, | |
| "learning_rate": 4.8194903961563255e-06, | |
| "logits/chosen": -1.8043403625488281, | |
| "logits/rejected": -1.5912580490112305, | |
| "logps/chosen": -212.60244750976562, | |
| "logps/rejected": -303.67816162109375, | |
| "loss": 0.176, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.5918294191360474, | |
| "rewards/margins": 3.186065196990967, | |
| "rewards/rejected": -3.7778944969177246, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.45866666666666667, | |
| "grad_norm": 3.463007926940918, | |
| "learning_rate": 4.808595544583578e-06, | |
| "logits/chosen": -1.763313889503479, | |
| "logits/rejected": -1.6367692947387695, | |
| "logps/chosen": -202.7022705078125, | |
| "logps/rejected": -306.5892028808594, | |
| "loss": 0.2357, | |
| "rewards/accuracies": 0.887499988079071, | |
| "rewards/chosen": -0.530913233757019, | |
| "rewards/margins": 3.3871676921844482, | |
| "rewards/rejected": -3.9180808067321777, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.4693333333333333, | |
| "grad_norm": 2.2896623611450195, | |
| "learning_rate": 4.797394529050577e-06, | |
| "logits/chosen": -1.6672195196151733, | |
| "logits/rejected": -1.4932286739349365, | |
| "logps/chosen": -249.15420532226562, | |
| "logps/rejected": -311.45587158203125, | |
| "loss": 0.1972, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.7217354774475098, | |
| "rewards/margins": 3.4966182708740234, | |
| "rewards/rejected": -4.218353271484375, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 1.703731656074524, | |
| "learning_rate": 4.785888835026393e-06, | |
| "logits/chosen": -1.7073551416397095, | |
| "logits/rejected": -1.517589807510376, | |
| "logps/chosen": -207.189208984375, | |
| "logps/rejected": -306.99774169921875, | |
| "loss": 0.1687, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -0.24462179839611053, | |
| "rewards/margins": 3.6275787353515625, | |
| "rewards/rejected": -3.8722004890441895, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.49066666666666664, | |
| "grad_norm": 2.8105475902557373, | |
| "learning_rate": 4.7740799883862966e-06, | |
| "logits/chosen": -1.708707571029663, | |
| "logits/rejected": -1.5929806232452393, | |
| "logps/chosen": -206.3466339111328, | |
| "logps/rejected": -301.8596496582031, | |
| "loss": 0.1369, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.34434959292411804, | |
| "rewards/margins": 3.6625962257385254, | |
| "rewards/rejected": -4.006945610046387, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5013333333333333, | |
| "grad_norm": 3.4738473892211914, | |
| "learning_rate": 4.761969555209394e-06, | |
| "logits/chosen": -1.7533724308013916, | |
| "logits/rejected": -1.5469913482666016, | |
| "logps/chosen": -203.04383850097656, | |
| "logps/rejected": -297.419921875, | |
| "loss": 0.1751, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.4458187520503998, | |
| "rewards/margins": 4.078585624694824, | |
| "rewards/rejected": -4.524405002593994, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.512, | |
| "grad_norm": 1.7620586156845093, | |
| "learning_rate": 4.7495591415709435e-06, | |
| "logits/chosen": -1.6817264556884766, | |
| "logits/rejected": -1.5172640085220337, | |
| "logps/chosen": -204.63270568847656, | |
| "logps/rejected": -305.5479431152344, | |
| "loss": 0.1515, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.46541064977645874, | |
| "rewards/margins": 4.120365142822266, | |
| "rewards/rejected": -4.585775852203369, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5226666666666666, | |
| "grad_norm": 1.4158145189285278, | |
| "learning_rate": 4.736850393329353e-06, | |
| "logits/chosen": -1.752340316772461, | |
| "logits/rejected": -1.503805160522461, | |
| "logps/chosen": -224.72238159179688, | |
| "logps/rejected": -302.3330078125, | |
| "loss": 0.2077, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.7848609685897827, | |
| "rewards/margins": 3.807910919189453, | |
| "rewards/rejected": -4.592772483825684, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5333333333333333, | |
| "grad_norm": 2.476516008377075, | |
| "learning_rate": 4.7238449959079095e-06, | |
| "logits/chosen": -1.7871164083480835, | |
| "logits/rejected": -1.5389041900634766, | |
| "logps/chosen": -208.1581268310547, | |
| "logps/rejected": -295.5426330566406, | |
| "loss": 0.1566, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.7614593505859375, | |
| "rewards/margins": 4.342654228210449, | |
| "rewards/rejected": -5.104113578796387, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.544, | |
| "grad_norm": 2.2022173404693604, | |
| "learning_rate": 4.710544674071259e-06, | |
| "logits/chosen": -1.7751401662826538, | |
| "logits/rejected": -1.5775682926177979, | |
| "logps/chosen": -266.8836669921875, | |
| "logps/rejected": -283.87957763671875, | |
| "loss": 0.2096, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.43025732040405273, | |
| "rewards/margins": 4.187695503234863, | |
| "rewards/rejected": -4.617953300476074, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5546666666666666, | |
| "grad_norm": 6.2148942947387695, | |
| "learning_rate": 4.696951191696673e-06, | |
| "logits/chosen": -1.8162314891815186, | |
| "logits/rejected": -1.5676801204681396, | |
| "logps/chosen": -198.07168579101562, | |
| "logps/rejected": -300.4014892578125, | |
| "loss": 0.2171, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.8536887168884277, | |
| "rewards/margins": 4.168699741363525, | |
| "rewards/rejected": -5.022388458251953, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5653333333333334, | |
| "grad_norm": 2.747276782989502, | |
| "learning_rate": 4.6830663515401196e-06, | |
| "logits/chosen": -1.810840368270874, | |
| "logits/rejected": -1.5849469900131226, | |
| "logps/chosen": -223.6453857421875, | |
| "logps/rejected": -289.7298278808594, | |
| "loss": 0.1779, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.4079108238220215, | |
| "rewards/margins": 4.084084987640381, | |
| "rewards/rejected": -4.491995811462402, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.576, | |
| "grad_norm": 2.140824794769287, | |
| "learning_rate": 4.6688919949971855e-06, | |
| "logits/chosen": -1.7053123712539673, | |
| "logits/rejected": -1.4852755069732666, | |
| "logps/chosen": -224.3308563232422, | |
| "logps/rejected": -324.8773498535156, | |
| "loss": 0.1714, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.6945469975471497, | |
| "rewards/margins": 4.143045902252197, | |
| "rewards/rejected": -4.837592601776123, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5866666666666667, | |
| "grad_norm": 1.720471739768982, | |
| "learning_rate": 4.654430001858874e-06, | |
| "logits/chosen": -1.5651119947433472, | |
| "logits/rejected": -1.4575681686401367, | |
| "logps/chosen": -230.16384887695312, | |
| "logps/rejected": -316.0377197265625, | |
| "loss": 0.1983, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.7454553842544556, | |
| "rewards/margins": 3.871629238128662, | |
| "rewards/rejected": -4.617084980010986, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.5973333333333334, | |
| "grad_norm": 1.851474404335022, | |
| "learning_rate": 4.6396822900623075e-06, | |
| "logits/chosen": -1.7224136590957642, | |
| "logits/rejected": -1.5002939701080322, | |
| "logps/chosen": -214.7126922607422, | |
| "logps/rejected": -302.85430908203125, | |
| "loss": 0.1892, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.7812764644622803, | |
| "rewards/margins": 4.019623756408691, | |
| "rewards/rejected": -4.800899982452393, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.608, | |
| "grad_norm": 2.1525115966796875, | |
| "learning_rate": 4.624650815436368e-06, | |
| "logits/chosen": -1.7956489324569702, | |
| "logits/rejected": -1.5618733167648315, | |
| "logps/chosen": -240.7410430908203, | |
| "logps/rejected": -317.7781982421875, | |
| "loss": 0.1822, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": -0.6186224222183228, | |
| "rewards/margins": 4.1468329429626465, | |
| "rewards/rejected": -4.765455722808838, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.6186666666666667, | |
| "grad_norm": 1.5195742845535278, | |
| "learning_rate": 4.609337571442324e-06, | |
| "logits/chosen": -1.8218319416046143, | |
| "logits/rejected": -1.6440036296844482, | |
| "logps/chosen": -199.41769409179688, | |
| "logps/rejected": -308.4180908203125, | |
| "loss": 0.1525, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.6784080266952515, | |
| "rewards/margins": 4.462456703186035, | |
| "rewards/rejected": -5.140865325927734, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6293333333333333, | |
| "grad_norm": 2.91068172454834, | |
| "learning_rate": 4.5937445889094535e-06, | |
| "logits/chosen": -1.7051767110824585, | |
| "logits/rejected": -1.537472128868103, | |
| "logps/chosen": -234.99716186523438, | |
| "logps/rejected": -309.03814697265625, | |
| "loss": 0.2014, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.6108633279800415, | |
| "rewards/margins": 3.835562229156494, | |
| "rewards/rejected": -4.446425437927246, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.7544635534286499, | |
| "learning_rate": 4.577873935765722e-06, | |
| "logits/chosen": -1.8061891794204712, | |
| "logits/rejected": -1.581192970275879, | |
| "logps/chosen": -193.08706665039062, | |
| "logps/rejected": -302.5655517578125, | |
| "loss": 0.1642, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.450061172246933, | |
| "rewards/margins": 4.481039524078369, | |
| "rewards/rejected": -4.931101322174072, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6506666666666666, | |
| "grad_norm": 3.741976022720337, | |
| "learning_rate": 4.561727716763532e-06, | |
| "logits/chosen": -1.750154733657837, | |
| "logits/rejected": -1.6348270177841187, | |
| "logps/chosen": -232.0447235107422, | |
| "logps/rejected": -299.55377197265625, | |
| "loss": 0.2224, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.28944095969200134, | |
| "rewards/margins": 4.048866271972656, | |
| "rewards/rejected": -4.3383073806762695, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.6613333333333333, | |
| "grad_norm": 2.625056028366089, | |
| "learning_rate": 4.545308073200591e-06, | |
| "logits/chosen": -1.7072067260742188, | |
| "logits/rejected": -1.515326738357544, | |
| "logps/chosen": -219.8765106201172, | |
| "logps/rejected": -305.98492431640625, | |
| "loss": 0.1724, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.3121286928653717, | |
| "rewards/margins": 4.423548698425293, | |
| "rewards/rejected": -4.7356767654418945, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.672, | |
| "grad_norm": 3.3447206020355225, | |
| "learning_rate": 4.528617182635943e-06, | |
| "logits/chosen": -1.7671434879302979, | |
| "logits/rejected": -1.5995731353759766, | |
| "logps/chosen": -213.7782745361328, | |
| "logps/rejected": -303.6085205078125, | |
| "loss": 0.2261, | |
| "rewards/accuracies": 0.90625, | |
| "rewards/chosen": -0.4999776780605316, | |
| "rewards/margins": 4.172996520996094, | |
| "rewards/rejected": -4.672974586486816, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.6826666666666666, | |
| "grad_norm": 3.0568952560424805, | |
| "learning_rate": 4.511657258601171e-06, | |
| "logits/chosen": -1.7790120840072632, | |
| "logits/rejected": -1.5551685094833374, | |
| "logps/chosen": -219.55514526367188, | |
| "logps/rejected": -298.2545471191406, | |
| "loss": 0.1924, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.366494745016098, | |
| "rewards/margins": 4.068981647491455, | |
| "rewards/rejected": -4.435476303100586, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6933333333333334, | |
| "grad_norm": 2.1040990352630615, | |
| "learning_rate": 4.494430550306848e-06, | |
| "logits/chosen": -1.5438107252120972, | |
| "logits/rejected": -1.4709008932113647, | |
| "logps/chosen": -226.4731903076172, | |
| "logps/rejected": -295.2748718261719, | |
| "loss": 0.1854, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -0.3337394595146179, | |
| "rewards/margins": 3.975902557373047, | |
| "rewards/rejected": -4.3096418380737305, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.704, | |
| "grad_norm": 2.9654102325439453, | |
| "learning_rate": 4.476939342344246e-06, | |
| "logits/chosen": -1.794424057006836, | |
| "logits/rejected": -1.5975098609924316, | |
| "logps/chosen": -220.2859344482422, | |
| "logps/rejected": -298.57977294921875, | |
| "loss": 0.1412, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.4857562482357025, | |
| "rewards/margins": 4.390166759490967, | |
| "rewards/rejected": -4.875923156738281, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7146666666666667, | |
| "grad_norm": 2.9396848678588867, | |
| "learning_rate": 4.45918595438236e-06, | |
| "logits/chosen": -1.595558762550354, | |
| "logits/rejected": -1.5090906620025635, | |
| "logps/chosen": -215.38565063476562, | |
| "logps/rejected": -291.2980041503906, | |
| "loss": 0.1736, | |
| "rewards/accuracies": 0.9125000238418579, | |
| "rewards/chosen": -0.33694586157798767, | |
| "rewards/margins": 4.20229434967041, | |
| "rewards/rejected": -4.53924036026001, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.7253333333333334, | |
| "grad_norm": 2.5800886154174805, | |
| "learning_rate": 4.441172740860267e-06, | |
| "logits/chosen": -1.6961015462875366, | |
| "logits/rejected": -1.5860854387283325, | |
| "logps/chosen": -208.26205444335938, | |
| "logps/rejected": -289.4010314941406, | |
| "loss": 0.2338, | |
| "rewards/accuracies": 0.918749988079071, | |
| "rewards/chosen": -0.3276976943016052, | |
| "rewards/margins": 3.7353787422180176, | |
| "rewards/rejected": -4.063076019287109, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.736, | |
| "grad_norm": 1.9638211727142334, | |
| "learning_rate": 4.422902090674886e-06, | |
| "logits/chosen": -1.658978819847107, | |
| "logits/rejected": -1.5040454864501953, | |
| "logps/chosen": -203.96173095703125, | |
| "logps/rejected": -306.2127990722656, | |
| "loss": 0.1547, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.2021322250366211, | |
| "rewards/margins": 4.683990955352783, | |
| "rewards/rejected": -4.886122703552246, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7466666666666667, | |
| "grad_norm": 1.643659234046936, | |
| "learning_rate": 4.4043764268641705e-06, | |
| "logits/chosen": -1.7135502099990845, | |
| "logits/rejected": -1.5774753093719482, | |
| "logps/chosen": -214.1969451904297, | |
| "logps/rejected": -296.029541015625, | |
| "loss": 0.1529, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.4376668930053711, | |
| "rewards/margins": 3.894376039505005, | |
| "rewards/rejected": -4.332043170928955, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7573333333333333, | |
| "grad_norm": 1.9113845825195312, | |
| "learning_rate": 4.385598206285758e-06, | |
| "logits/chosen": -1.7747424840927124, | |
| "logits/rejected": -1.6252682209014893, | |
| "logps/chosen": -212.4604034423828, | |
| "logps/rejected": -336.4494323730469, | |
| "loss": 0.1163, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -0.5154281258583069, | |
| "rewards/margins": 5.046199321746826, | |
| "rewards/rejected": -5.561627388000488, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.768, | |
| "grad_norm": 2.6899571418762207, | |
| "learning_rate": 4.3665699192911495e-06, | |
| "logits/chosen": -1.6771185398101807, | |
| "logits/rejected": -1.5176887512207031, | |
| "logps/chosen": -204.26272583007812, | |
| "logps/rejected": -299.89288330078125, | |
| "loss": 0.1435, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.373568058013916, | |
| "rewards/margins": 4.497839450836182, | |
| "rewards/rejected": -4.871407508850098, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7786666666666666, | |
| "grad_norm": 2.5716707706451416, | |
| "learning_rate": 4.347294089395438e-06, | |
| "logits/chosen": -1.7271244525909424, | |
| "logits/rejected": -1.5585415363311768, | |
| "logps/chosen": -234.6363525390625, | |
| "logps/rejected": -342.66668701171875, | |
| "loss": 0.1127, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -0.6868476271629333, | |
| "rewards/margins": 5.073822975158691, | |
| "rewards/rejected": -5.7606706619262695, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.7893333333333333, | |
| "grad_norm": 1.3837978839874268, | |
| "learning_rate": 4.327773272942647e-06, | |
| "logits/chosen": -1.6469981670379639, | |
| "logits/rejected": -1.5777738094329834, | |
| "logps/chosen": -203.450439453125, | |
| "logps/rejected": -331.1883544921875, | |
| "loss": 0.154, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.6987647414207458, | |
| "rewards/margins": 4.91748046875, | |
| "rewards/rejected": -5.616245746612549, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 2.062950849533081, | |
| "learning_rate": 4.308010058766703e-06, | |
| "logits/chosen": -1.829921007156372, | |
| "logits/rejected": -1.6271889209747314, | |
| "logps/chosen": -229.3686981201172, | |
| "logps/rejected": -320.89117431640625, | |
| "loss": 0.1308, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -0.9174984693527222, | |
| "rewards/margins": 5.067828178405762, | |
| "rewards/rejected": -5.985327243804932, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.8106666666666666, | |
| "grad_norm": 2.347057580947876, | |
| "learning_rate": 4.2880070678481135e-06, | |
| "logits/chosen": -1.7627887725830078, | |
| "logits/rejected": -1.5318098068237305, | |
| "logps/chosen": -199.64096069335938, | |
| "logps/rejected": -312.7142028808594, | |
| "loss": 0.1086, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.8036954998970032, | |
| "rewards/margins": 5.584313869476318, | |
| "rewards/rejected": -6.388009071350098, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8213333333333334, | |
| "grad_norm": 3.017650604248047, | |
| "learning_rate": 4.267766952966369e-06, | |
| "logits/chosen": -1.7800073623657227, | |
| "logits/rejected": -1.585495948791504, | |
| "logps/chosen": -227.3597412109375, | |
| "logps/rejected": -320.13653564453125, | |
| "loss": 0.1488, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -0.9181621670722961, | |
| "rewards/margins": 5.3811445236206055, | |
| "rewards/rejected": -6.299307346343994, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.832, | |
| "grad_norm": 3.2862038612365723, | |
| "learning_rate": 4.24729239834814e-06, | |
| "logits/chosen": -1.7047274112701416, | |
| "logits/rejected": -1.5511637926101685, | |
| "logps/chosen": -229.1103057861328, | |
| "logps/rejected": -305.8432312011719, | |
| "loss": 0.1743, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -1.0054670572280884, | |
| "rewards/margins": 4.895895957946777, | |
| "rewards/rejected": -5.901362895965576, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8426666666666667, | |
| "grad_norm": 2.632439136505127, | |
| "learning_rate": 4.226586119311291e-06, | |
| "logits/chosen": -1.7877633571624756, | |
| "logits/rejected": -1.6740882396697998, | |
| "logps/chosen": -217.3224334716797, | |
| "logps/rejected": -324.86737060546875, | |
| "loss": 0.155, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.9102714657783508, | |
| "rewards/margins": 5.167611598968506, | |
| "rewards/rejected": -6.077882766723633, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.8533333333333334, | |
| "grad_norm": 3.4450037479400635, | |
| "learning_rate": 4.205650861904782e-06, | |
| "logits/chosen": -1.7734384536743164, | |
| "logits/rejected": -1.5797088146209717, | |
| "logps/chosen": -212.7061309814453, | |
| "logps/rejected": -316.12127685546875, | |
| "loss": 0.1384, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -1.066833734512329, | |
| "rewards/margins": 5.305598258972168, | |
| "rewards/rejected": -6.372431755065918, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.864, | |
| "grad_norm": 0.9102749228477478, | |
| "learning_rate": 4.184489402544486e-06, | |
| "logits/chosen": -1.820951223373413, | |
| "logits/rejected": -1.6782753467559814, | |
| "logps/chosen": -207.24374389648438, | |
| "logps/rejected": -337.61700439453125, | |
| "loss": 0.1447, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -0.7356926202774048, | |
| "rewards/margins": 5.745200157165527, | |
| "rewards/rejected": -6.480892181396484, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.8746666666666667, | |
| "grad_norm": 2.6713948249816895, | |
| "learning_rate": 4.1631045476449865e-06, | |
| "logits/chosen": -1.8452482223510742, | |
| "logits/rejected": -1.6467907428741455, | |
| "logps/chosen": -204.97213745117188, | |
| "logps/rejected": -308.3583984375, | |
| "loss": 0.2012, | |
| "rewards/accuracies": 0.8999999761581421, | |
| "rewards/chosen": -0.972851574420929, | |
| "rewards/margins": 4.9895501136779785, | |
| "rewards/rejected": -5.962401866912842, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8853333333333333, | |
| "grad_norm": 1.5577032566070557, | |
| "learning_rate": 4.141499133247391e-06, | |
| "logits/chosen": -1.6295883655548096, | |
| "logits/rejected": -1.6040502786636353, | |
| "logps/chosen": -216.0922393798828, | |
| "logps/rejected": -305.439208984375, | |
| "loss": 0.1682, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -0.7497788667678833, | |
| "rewards/margins": 4.673377990722656, | |
| "rewards/rejected": -5.423156261444092, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.896, | |
| "grad_norm": 2.909951686859131, | |
| "learning_rate": 4.119676024643217e-06, | |
| "logits/chosen": -1.7234691381454468, | |
| "logits/rejected": -1.5772818326950073, | |
| "logps/chosen": -221.32339477539062, | |
| "logps/rejected": -318.9827880859375, | |
| "loss": 0.1345, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -0.6560640335083008, | |
| "rewards/margins": 5.7348504066467285, | |
| "rewards/rejected": -6.390913963317871, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9066666666666666, | |
| "grad_norm": 4.880727291107178, | |
| "learning_rate": 4.0976381159944055e-06, | |
| "logits/chosen": -1.7095870971679688, | |
| "logits/rejected": -1.5015006065368652, | |
| "logps/chosen": -239.50650024414062, | |
| "logps/rejected": -340.181884765625, | |
| "loss": 0.1615, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -0.8043699264526367, | |
| "rewards/margins": 5.589447975158691, | |
| "rewards/rejected": -6.393817901611328, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9173333333333333, | |
| "grad_norm": 1.4485254287719727, | |
| "learning_rate": 4.07538832994949e-06, | |
| "logits/chosen": -1.7408502101898193, | |
| "logits/rejected": -1.5344233512878418, | |
| "logps/chosen": -217.7578582763672, | |
| "logps/rejected": -321.7436218261719, | |
| "loss": 0.1404, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -1.0381317138671875, | |
| "rewards/margins": 5.289247512817383, | |
| "rewards/rejected": -6.3273797035217285, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.928, | |
| "grad_norm": 2.7532310485839844, | |
| "learning_rate": 4.0529296172560055e-06, | |
| "logits/chosen": -1.829403281211853, | |
| "logits/rejected": -1.6238079071044922, | |
| "logps/chosen": -212.47366333007812, | |
| "logps/rejected": -326.38494873046875, | |
| "loss": 0.1214, | |
| "rewards/accuracies": 0.925000011920929, | |
| "rewards/chosen": -1.2220343351364136, | |
| "rewards/margins": 5.693434715270996, | |
| "rewards/rejected": -6.915470123291016, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.9386666666666666, | |
| "grad_norm": 2.052518844604492, | |
| "learning_rate": 4.030264956369158e-06, | |
| "logits/chosen": -1.7937902212142944, | |
| "logits/rejected": -1.673724889755249, | |
| "logps/chosen": -213.76602172851562, | |
| "logps/rejected": -336.3763732910156, | |
| "loss": 0.1351, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -0.9977391958236694, | |
| "rewards/margins": 5.939627170562744, | |
| "rewards/rejected": -6.9373674392700195, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9493333333333334, | |
| "grad_norm": 5.171205043792725, | |
| "learning_rate": 4.0073973530568274e-06, | |
| "logits/chosen": -1.762075424194336, | |
| "logits/rejected": -1.5835630893707275, | |
| "logps/chosen": -238.7696990966797, | |
| "logps/rejected": -327.5137634277344, | |
| "loss": 0.153, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -1.2305481433868408, | |
| "rewards/margins": 5.5450639724731445, | |
| "rewards/rejected": -6.77561092376709, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 1.4903888702392578, | |
| "learning_rate": 3.9843298400009435e-06, | |
| "logits/chosen": -1.7811949253082275, | |
| "logits/rejected": -1.6663089990615845, | |
| "logps/chosen": -197.07852172851562, | |
| "logps/rejected": -334.981201171875, | |
| "loss": 0.0804, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -1.1236552000045776, | |
| "rewards/margins": 6.174225330352783, | |
| "rewards/rejected": -7.29788064956665, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9706666666666667, | |
| "grad_norm": 3.032621145248413, | |
| "learning_rate": 3.961065476395296e-06, | |
| "logits/chosen": -1.8615983724594116, | |
| "logits/rejected": -1.6502336263656616, | |
| "logps/chosen": -223.0669403076172, | |
| "logps/rejected": -337.3778991699219, | |
| "loss": 0.1113, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.9720472097396851, | |
| "rewards/margins": 6.413779258728027, | |
| "rewards/rejected": -7.385826110839844, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.9813333333333333, | |
| "grad_norm": 2.9131219387054443, | |
| "learning_rate": 3.937607347539823e-06, | |
| "logits/chosen": -1.839909553527832, | |
| "logits/rejected": -1.635088562965393, | |
| "logps/chosen": -208.74105834960938, | |
| "logps/rejected": -329.6622619628906, | |
| "loss": 0.17, | |
| "rewards/accuracies": 0.9312499761581421, | |
| "rewards/chosen": -1.350288987159729, | |
| "rewards/margins": 6.138724327087402, | |
| "rewards/rejected": -7.489013671875, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.992, | |
| "grad_norm": 1.870213508605957, | |
| "learning_rate": 3.913958564431447e-06, | |
| "logits/chosen": -1.8265674114227295, | |
| "logits/rejected": -1.642881155014038, | |
| "logps/chosen": -230.4730682373047, | |
| "logps/rejected": -343.09173583984375, | |
| "loss": 0.1161, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.3268934488296509, | |
| "rewards/margins": 6.083160877227783, | |
| "rewards/rejected": -7.410054683685303, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.0021333333333333, | |
| "grad_norm": 3.683117151260376, | |
| "learning_rate": 3.89012226335149e-06, | |
| "logits/chosen": -1.7430078983306885, | |
| "logits/rejected": -1.6155897378921509, | |
| "logps/chosen": -204.52291870117188, | |
| "logps/rejected": -320.3079833984375, | |
| "loss": 0.1491, | |
| "rewards/accuracies": 0.9276315569877625, | |
| "rewards/chosen": -1.002981424331665, | |
| "rewards/margins": 5.943866729736328, | |
| "rewards/rejected": -6.946847915649414, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0128, | |
| "grad_norm": 1.268686294555664, | |
| "learning_rate": 3.86610160544975e-06, | |
| "logits/chosen": -1.8006131649017334, | |
| "logits/rejected": -1.650631308555603, | |
| "logps/chosen": -208.288818359375, | |
| "logps/rejected": -330.7633972167969, | |
| "loss": 0.1072, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.0482348203659058, | |
| "rewards/margins": 6.511569023132324, | |
| "rewards/rejected": -7.5598039627075195, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0234666666666667, | |
| "grad_norm": 1.9249908924102783, | |
| "learning_rate": 3.841899776325267e-06, | |
| "logits/chosen": -1.7247737646102905, | |
| "logits/rejected": -1.6122996807098389, | |
| "logps/chosen": -229.8289337158203, | |
| "logps/rejected": -347.37017822265625, | |
| "loss": 0.1015, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -0.8600400686264038, | |
| "rewards/margins": 6.6198883056640625, | |
| "rewards/rejected": -7.479928493499756, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.0341333333333333, | |
| "grad_norm": 1.7193055152893066, | |
| "learning_rate": 3.817519985603855e-06, | |
| "logits/chosen": -1.8941304683685303, | |
| "logits/rejected": -1.7352015972137451, | |
| "logps/chosen": -212.7666778564453, | |
| "logps/rejected": -335.093994140625, | |
| "loss": 0.1174, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.8410255908966064, | |
| "rewards/margins": 6.038805961608887, | |
| "rewards/rejected": -7.8798322677612305, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.0448, | |
| "grad_norm": 3.5899593830108643, | |
| "learning_rate": 3.792965466512437e-06, | |
| "logits/chosen": -1.7666019201278687, | |
| "logits/rejected": -1.573142409324646, | |
| "logps/chosen": -232.1917266845703, | |
| "logps/rejected": -326.2380065917969, | |
| "loss": 0.1254, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.2813751697540283, | |
| "rewards/margins": 6.1387457847595215, | |
| "rewards/rejected": -7.4201202392578125, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.0554666666666668, | |
| "grad_norm": 3.8329415321350098, | |
| "learning_rate": 3.7682394754502687e-06, | |
| "logits/chosen": -1.7883208990097046, | |
| "logits/rejected": -1.6514673233032227, | |
| "logps/chosen": -220.38290405273438, | |
| "logps/rejected": -339.58355712890625, | |
| "loss": 0.1327, | |
| "rewards/accuracies": 0.9437500238418579, | |
| "rewards/chosen": -1.6085094213485718, | |
| "rewards/margins": 6.206122398376465, | |
| "rewards/rejected": -7.814632415771484, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.0661333333333334, | |
| "grad_norm": 2.348676919937134, | |
| "learning_rate": 3.7433452915570685e-06, | |
| "logits/chosen": -1.8436359167099, | |
| "logits/rejected": -1.6850477457046509, | |
| "logps/chosen": -236.15066528320312, | |
| "logps/rejected": -352.1568908691406, | |
| "loss": 0.0809, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.5462639331817627, | |
| "rewards/margins": 6.4064435958862305, | |
| "rewards/rejected": -7.952707767486572, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0768, | |
| "grad_norm": 0.5271938443183899, | |
| "learning_rate": 3.718286216278143e-06, | |
| "logits/chosen": -1.7932989597320557, | |
| "logits/rejected": -1.6435298919677734, | |
| "logps/chosen": -217.5049591064453, | |
| "logps/rejected": -361.5685729980469, | |
| "loss": 0.0622, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -1.248918056488037, | |
| "rewards/margins": 7.289629936218262, | |
| "rewards/rejected": -8.538548469543457, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.0874666666666666, | |
| "grad_norm": 2.5198798179626465, | |
| "learning_rate": 3.6930655729265557e-06, | |
| "logits/chosen": -1.8247435092926025, | |
| "logits/rejected": -1.648455023765564, | |
| "logps/chosen": -252.8979949951172, | |
| "logps/rejected": -347.34930419921875, | |
| "loss": 0.1646, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -1.7946258783340454, | |
| "rewards/margins": 6.223298072814941, | |
| "rewards/rejected": -8.017923355102539, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.0981333333333334, | |
| "grad_norm": 2.4615252017974854, | |
| "learning_rate": 3.667686706242386e-06, | |
| "logits/chosen": -1.748671293258667, | |
| "logits/rejected": -1.6229207515716553, | |
| "logps/chosen": -230.9926300048828, | |
| "logps/rejected": -341.594970703125, | |
| "loss": 0.1012, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.5733277797698975, | |
| "rewards/margins": 6.404208183288574, | |
| "rewards/rejected": -7.977537631988525, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.1088, | |
| "grad_norm": 1.5133553743362427, | |
| "learning_rate": 3.6421529819491576e-06, | |
| "logits/chosen": -1.8420203924179077, | |
| "logits/rejected": -1.71492600440979, | |
| "logps/chosen": -239.5989532470703, | |
| "logps/rejected": -357.04937744140625, | |
| "loss": 0.075, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -1.4472997188568115, | |
| "rewards/margins": 7.079667091369629, | |
| "rewards/rejected": -8.526968002319336, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.1194666666666666, | |
| "grad_norm": 1.1103506088256836, | |
| "learning_rate": 3.6164677863074783e-06, | |
| "logits/chosen": -1.7879250049591064, | |
| "logits/rejected": -1.6893844604492188, | |
| "logps/chosen": -215.9795684814453, | |
| "logps/rejected": -359.2721252441406, | |
| "loss": 0.0696, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -1.6059297323226929, | |
| "rewards/margins": 7.181890964508057, | |
| "rewards/rejected": -8.787820816040039, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.1301333333333332, | |
| "grad_norm": 1.1953598260879517, | |
| "learning_rate": 3.590634525665955e-06, | |
| "logits/chosen": -1.8406591415405273, | |
| "logits/rejected": -1.6219384670257568, | |
| "logps/chosen": -224.5021514892578, | |
| "logps/rejected": -349.89239501953125, | |
| "loss": 0.0765, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.9252121448516846, | |
| "rewards/margins": 6.995842933654785, | |
| "rewards/rejected": -8.921055793762207, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.1408, | |
| "grad_norm": 1.1147931814193726, | |
| "learning_rate": 3.564656626009446e-06, | |
| "logits/chosen": -1.8823022842407227, | |
| "logits/rejected": -1.8116251230239868, | |
| "logps/chosen": -210.24227905273438, | |
| "logps/rejected": -362.93145751953125, | |
| "loss": 0.1049, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.5180342197418213, | |
| "rewards/margins": 7.260135650634766, | |
| "rewards/rejected": -8.778169631958008, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.1514666666666666, | |
| "grad_norm": 2.6945362091064453, | |
| "learning_rate": 3.5385375325047167e-06, | |
| "logits/chosen": -1.9016218185424805, | |
| "logits/rejected": -1.729413628578186, | |
| "logps/chosen": -270.2596130371094, | |
| "logps/rejected": -332.03448486328125, | |
| "loss": 0.0759, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -1.4495543241500854, | |
| "rewards/margins": 6.9950737953186035, | |
| "rewards/rejected": -8.44462776184082, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.1621333333333332, | |
| "grad_norm": 3.8907995223999023, | |
| "learning_rate": 3.512280709043534e-06, | |
| "logits/chosen": -1.6742641925811768, | |
| "logits/rejected": -1.541598916053772, | |
| "logps/chosen": -226.13583374023438, | |
| "logps/rejected": -314.2048034667969, | |
| "loss": 0.1446, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.6109225749969482, | |
| "rewards/margins": 5.946639060974121, | |
| "rewards/rejected": -7.55756139755249, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.1728, | |
| "grad_norm": 5.299363613128662, | |
| "learning_rate": 3.4858896377832966e-06, | |
| "logits/chosen": -1.779280424118042, | |
| "logits/rejected": -1.6656166315078735, | |
| "logps/chosen": -223.9244384765625, | |
| "logps/rejected": -348.4911193847656, | |
| "loss": 0.1078, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.9482396841049194, | |
| "rewards/margins": 6.998904228210449, | |
| "rewards/rejected": -8.947145462036133, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.1834666666666667, | |
| "grad_norm": 3.3822855949401855, | |
| "learning_rate": 3.459367818685231e-06, | |
| "logits/chosen": -1.7240779399871826, | |
| "logits/rejected": -1.656288504600525, | |
| "logps/chosen": -236.51571655273438, | |
| "logps/rejected": -341.683837890625, | |
| "loss": 0.1403, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -1.5938842296600342, | |
| "rewards/margins": 6.741963863372803, | |
| "rewards/rejected": -8.335847854614258, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.1941333333333333, | |
| "grad_norm": 0.8533849716186523, | |
| "learning_rate": 3.432718769050228e-06, | |
| "logits/chosen": -1.6910005807876587, | |
| "logits/rejected": -1.5575242042541504, | |
| "logps/chosen": -234.0883026123047, | |
| "logps/rejected": -338.9890441894531, | |
| "loss": 0.1177, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.399706244468689, | |
| "rewards/margins": 6.219531059265137, | |
| "rewards/rejected": -7.619238376617432, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.2048, | |
| "grad_norm": 1.684748888015747, | |
| "learning_rate": 3.405946023052386e-06, | |
| "logits/chosen": -1.811579942703247, | |
| "logits/rejected": -1.6530382633209229, | |
| "logps/chosen": -210.05844116210938, | |
| "logps/rejected": -330.2632141113281, | |
| "loss": 0.0964, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.36553955078125, | |
| "rewards/margins": 6.456725120544434, | |
| "rewards/rejected": -7.822264671325684, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.2154666666666667, | |
| "grad_norm": 4.0320587158203125, | |
| "learning_rate": 3.3790531312703055e-06, | |
| "logits/chosen": -1.7837402820587158, | |
| "logits/rejected": -1.6148713827133179, | |
| "logps/chosen": -215.8035125732422, | |
| "logps/rejected": -337.56805419921875, | |
| "loss": 0.0946, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.377327561378479, | |
| "rewards/margins": 6.664053440093994, | |
| "rewards/rejected": -8.041379928588867, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.2261333333333333, | |
| "grad_norm": 2.986213445663452, | |
| "learning_rate": 3.35204366021622e-06, | |
| "logits/chosen": -1.7007958889007568, | |
| "logits/rejected": -1.554639220237732, | |
| "logps/chosen": -231.6057891845703, | |
| "logps/rejected": -361.81781005859375, | |
| "loss": 0.1045, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.2639760971069336, | |
| "rewards/margins": 6.448613166809082, | |
| "rewards/rejected": -7.712589263916016, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.2368000000000001, | |
| "grad_norm": 1.3645539283752441, | |
| "learning_rate": 3.3249211918630055e-06, | |
| "logits/chosen": -1.7643219232559204, | |
| "logits/rejected": -1.6710097789764404, | |
| "logps/chosen": -210.04476928710938, | |
| "logps/rejected": -327.1766662597656, | |
| "loss": 0.1177, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.2976925373077393, | |
| "rewards/margins": 5.873549461364746, | |
| "rewards/rejected": -7.171242713928223, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.2474666666666667, | |
| "grad_norm": 1.9278602600097656, | |
| "learning_rate": 3.2976893231691424e-06, | |
| "logits/chosen": -1.750998854637146, | |
| "logits/rejected": -1.67935049533844, | |
| "logps/chosen": -232.449462890625, | |
| "logps/rejected": -352.84014892578125, | |
| "loss": 0.1109, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.335108757019043, | |
| "rewards/margins": 6.393424987792969, | |
| "rewards/rejected": -7.728533744812012, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.2581333333333333, | |
| "grad_norm": 5.031718730926514, | |
| "learning_rate": 3.270351665601691e-06, | |
| "logits/chosen": -1.761778473854065, | |
| "logits/rejected": -1.628424048423767, | |
| "logps/chosen": -226.8080291748047, | |
| "logps/rejected": -330.947998046875, | |
| "loss": 0.1459, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.291100025177002, | |
| "rewards/margins": 6.421347141265869, | |
| "rewards/rejected": -7.712447166442871, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.2688, | |
| "grad_norm": 2.0690762996673584, | |
| "learning_rate": 3.2429118446573417e-06, | |
| "logits/chosen": -1.797255277633667, | |
| "logits/rejected": -1.6581494808197021, | |
| "logps/chosen": -232.0042724609375, | |
| "logps/rejected": -345.2641296386719, | |
| "loss": 0.1162, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.5298513174057007, | |
| "rewards/margins": 6.056611061096191, | |
| "rewards/rejected": -7.586462497711182, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.2794666666666665, | |
| "grad_norm": 2.8271262645721436, | |
| "learning_rate": 3.215373499381602e-06, | |
| "logits/chosen": -1.7107791900634766, | |
| "logits/rejected": -1.5847904682159424, | |
| "logps/chosen": -239.77529907226562, | |
| "logps/rejected": -327.42724609375, | |
| "loss": 0.0918, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.4036293029785156, | |
| "rewards/margins": 5.92539119720459, | |
| "rewards/rejected": -7.329021453857422, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.2901333333333334, | |
| "grad_norm": 1.7590172290802002, | |
| "learning_rate": 3.1877402818861954e-06, | |
| "logits/chosen": -1.8445212841033936, | |
| "logits/rejected": -1.6904475688934326, | |
| "logps/chosen": -243.2404327392578, | |
| "logps/rejected": -349.3418273925781, | |
| "loss": 0.079, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.8851490020751953, | |
| "rewards/margins": 6.981202602386475, | |
| "rewards/rejected": -8.866351127624512, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.3008, | |
| "grad_norm": 1.688004970550537, | |
| "learning_rate": 3.1600158568647145e-06, | |
| "logits/chosen": -1.782779335975647, | |
| "logits/rejected": -1.7313483953475952, | |
| "logps/chosen": -218.34713745117188, | |
| "logps/rejected": -319.5534362792969, | |
| "loss": 0.0996, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.3257147073745728, | |
| "rewards/margins": 6.426403045654297, | |
| "rewards/rejected": -7.752118110656738, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.3114666666666666, | |
| "grad_norm": 1.4718377590179443, | |
| "learning_rate": 3.1322039011066172e-06, | |
| "logits/chosen": -1.7013763189315796, | |
| "logits/rejected": -1.566279649734497, | |
| "logps/chosen": -247.23135375976562, | |
| "logps/rejected": -336.8480529785156, | |
| "loss": 0.0784, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -1.3434125185012817, | |
| "rewards/margins": 6.7102203369140625, | |
| "rewards/rejected": -8.053632736206055, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.3221333333333334, | |
| "grad_norm": 3.314124822616577, | |
| "learning_rate": 3.1043081030096125e-06, | |
| "logits/chosen": -1.6063953638076782, | |
| "logits/rejected": -1.5149648189544678, | |
| "logps/chosen": -228.9906463623047, | |
| "logps/rejected": -318.33660888671875, | |
| "loss": 0.1127, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.6745455265045166, | |
| "rewards/margins": 6.050885200500488, | |
| "rewards/rejected": -7.725429534912109, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.3328, | |
| "grad_norm": 2.55005145072937, | |
| "learning_rate": 3.0763321620905072e-06, | |
| "logits/chosen": -1.6070985794067383, | |
| "logits/rejected": -1.5515340566635132, | |
| "logps/chosen": -219.2987823486328, | |
| "logps/rejected": -357.78619384765625, | |
| "loss": 0.1139, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.6658519506454468, | |
| "rewards/margins": 6.835375785827637, | |
| "rewards/rejected": -8.501228332519531, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.3434666666666666, | |
| "grad_norm": 1.571263074874878, | |
| "learning_rate": 3.048279788494583e-06, | |
| "logits/chosen": -1.7900960445404053, | |
| "logits/rejected": -1.6767871379852295, | |
| "logps/chosen": -217.8803253173828, | |
| "logps/rejected": -332.90838623046875, | |
| "loss": 0.097, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.8096497058868408, | |
| "rewards/margins": 7.075921535491943, | |
| "rewards/rejected": -8.88557243347168, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.3541333333333334, | |
| "grad_norm": 1.4549236297607422, | |
| "learning_rate": 3.0201547025035553e-06, | |
| "logits/chosen": -1.6474237442016602, | |
| "logits/rejected": -1.5805208683013916, | |
| "logps/chosen": -222.01071166992188, | |
| "logps/rejected": -363.7785339355469, | |
| "loss": 0.1013, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.6705299615859985, | |
| "rewards/margins": 7.639910697937012, | |
| "rewards/rejected": -9.310441970825195, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.3648, | |
| "grad_norm": 4.120782852172852, | |
| "learning_rate": 2.9919606340421963e-06, | |
| "logits/chosen": -1.8172228336334229, | |
| "logits/rejected": -1.6626869440078735, | |
| "logps/chosen": -214.86422729492188, | |
| "logps/rejected": -349.9516296386719, | |
| "loss": 0.1206, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.9698278903961182, | |
| "rewards/margins": 7.807936191558838, | |
| "rewards/rejected": -9.777763366699219, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.3754666666666666, | |
| "grad_norm": 0.9963375926017761, | |
| "learning_rate": 2.9637013221836723e-06, | |
| "logits/chosen": -1.753147840499878, | |
| "logits/rejected": -1.6141316890716553, | |
| "logps/chosen": -239.36892700195312, | |
| "logps/rejected": -333.98956298828125, | |
| "loss": 0.1611, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.883836030960083, | |
| "rewards/margins": 6.4801025390625, | |
| "rewards/rejected": -8.363938331604004, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.3861333333333334, | |
| "grad_norm": 5.172098636627197, | |
| "learning_rate": 2.9353805146536762e-06, | |
| "logits/chosen": -1.6604881286621094, | |
| "logits/rejected": -1.6201698780059814, | |
| "logps/chosen": -220.64718627929688, | |
| "logps/rejected": -341.5318603515625, | |
| "loss": 0.148, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.7640655040740967, | |
| "rewards/margins": 6.463761806488037, | |
| "rewards/rejected": -8.227828025817871, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.3968, | |
| "grad_norm": 2.365213632583618, | |
| "learning_rate": 2.907001967333401e-06, | |
| "logits/chosen": -1.8392598628997803, | |
| "logits/rejected": -1.7236486673355103, | |
| "logps/chosen": -228.3126678466797, | |
| "logps/rejected": -362.47869873046875, | |
| "loss": 0.122, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.6854579448699951, | |
| "rewards/margins": 7.8559770584106445, | |
| "rewards/rejected": -9.541436195373535, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.4074666666666666, | |
| "grad_norm": 5.077343940734863, | |
| "learning_rate": 2.878569443761442e-06, | |
| "logits/chosen": -1.7001498937606812, | |
| "logits/rejected": -1.5788886547088623, | |
| "logps/chosen": -218.8361358642578, | |
| "logps/rejected": -336.6226501464844, | |
| "loss": 0.1438, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.9872394800186157, | |
| "rewards/margins": 6.80609130859375, | |
| "rewards/rejected": -8.793330192565918, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.4181333333333335, | |
| "grad_norm": 3.6902966499328613, | |
| "learning_rate": 2.85008671463468e-06, | |
| "logits/chosen": -1.6842687129974365, | |
| "logits/rejected": -1.6067196130752563, | |
| "logps/chosen": -217.57186889648438, | |
| "logps/rejected": -340.4504089355469, | |
| "loss": 0.1503, | |
| "rewards/accuracies": 0.9375, | |
| "rewards/chosen": -1.7623138427734375, | |
| "rewards/margins": 6.442144870758057, | |
| "rewards/rejected": -8.204458236694336, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.4288, | |
| "grad_norm": 2.927492380142212, | |
| "learning_rate": 2.8215575573082127e-06, | |
| "logits/chosen": -1.7495641708374023, | |
| "logits/rejected": -1.673855185508728, | |
| "logps/chosen": -225.540771484375, | |
| "logps/rejected": -351.88665771484375, | |
| "loss": 0.0832, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.921282410621643, | |
| "rewards/margins": 6.98816442489624, | |
| "rewards/rejected": -8.909446716308594, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.4394666666666667, | |
| "grad_norm": 1.9164410829544067, | |
| "learning_rate": 2.792985755294404e-06, | |
| "logits/chosen": -1.841391921043396, | |
| "logits/rejected": -1.6707547903060913, | |
| "logps/chosen": -209.87161254882812, | |
| "logps/rejected": -349.9462890625, | |
| "loss": 0.0892, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.7684261798858643, | |
| "rewards/margins": 7.643846035003662, | |
| "rewards/rejected": -9.412271499633789, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.4501333333333333, | |
| "grad_norm": 2.0672764778137207, | |
| "learning_rate": 2.764375097761122e-06, | |
| "logits/chosen": -1.6338927745819092, | |
| "logits/rejected": -1.4459412097930908, | |
| "logps/chosen": -221.2202606201172, | |
| "logps/rejected": -330.55242919921875, | |
| "loss": 0.0878, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -1.3578730821609497, | |
| "rewards/margins": 7.062647819519043, | |
| "rewards/rejected": -8.420520782470703, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.4607999999999999, | |
| "grad_norm": 1.4110078811645508, | |
| "learning_rate": 2.73572937902922e-06, | |
| "logits/chosen": -1.7500721216201782, | |
| "logits/rejected": -1.5899924039840698, | |
| "logps/chosen": -234.5608673095703, | |
| "logps/rejected": -351.0326232910156, | |
| "loss": 0.0846, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.7048619985580444, | |
| "rewards/margins": 7.528687953948975, | |
| "rewards/rejected": -9.233550071716309, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.4714666666666667, | |
| "grad_norm": NaN, | |
| "learning_rate": 2.7070523980693362e-06, | |
| "logits/chosen": -1.5613092184066772, | |
| "logits/rejected": -1.4905552864074707, | |
| "logps/chosen": -243.5371856689453, | |
| "logps/rejected": -338.37060546875, | |
| "loss": 0.0945, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.6675678491592407, | |
| "rewards/margins": 7.0091729164123535, | |
| "rewards/rejected": -8.676740646362305, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.4821333333333333, | |
| "grad_norm": 3.837553024291992, | |
| "learning_rate": 2.678347957998081e-06, | |
| "logits/chosen": -1.741687536239624, | |
| "logits/rejected": -1.612644910812378, | |
| "logps/chosen": -207.15847778320312, | |
| "logps/rejected": -344.6006774902344, | |
| "loss": 0.0866, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.7650225162506104, | |
| "rewards/margins": 7.5840888023376465, | |
| "rewards/rejected": -9.349112510681152, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.4928, | |
| "grad_norm": 1.987453818321228, | |
| "learning_rate": 2.649619865573666e-06, | |
| "logits/chosen": -1.6311708688735962, | |
| "logits/rejected": -1.6032650470733643, | |
| "logps/chosen": -224.74313354492188, | |
| "logps/rejected": -356.2095642089844, | |
| "loss": 0.0758, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.1680212020874023, | |
| "rewards/margins": 7.786881446838379, | |
| "rewards/rejected": -9.954902648925781, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.5034666666666667, | |
| "grad_norm": 2.1548123359680176, | |
| "learning_rate": 2.620871930691058e-06, | |
| "logits/chosen": -1.7811607122421265, | |
| "logits/rejected": -1.7126588821411133, | |
| "logps/chosen": -275.605712890625, | |
| "logps/rejected": -376.85308837890625, | |
| "loss": 0.0698, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -1.9767110347747803, | |
| "rewards/margins": 7.755047798156738, | |
| "rewards/rejected": -9.731759071350098, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.5141333333333333, | |
| "grad_norm": 3.925725221633911, | |
| "learning_rate": 2.5921079658767152e-06, | |
| "logits/chosen": -1.7364921569824219, | |
| "logits/rejected": -1.5872316360473633, | |
| "logps/chosen": -219.77951049804688, | |
| "logps/rejected": -333.9054870605469, | |
| "loss": 0.095, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -1.9969074726104736, | |
| "rewards/margins": 7.254309177398682, | |
| "rewards/rejected": -9.251215934753418, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.5248, | |
| "grad_norm": 3.2879045009613037, | |
| "learning_rate": 2.56333178578297e-06, | |
| "logits/chosen": -1.744199514389038, | |
| "logits/rejected": -1.5770639181137085, | |
| "logps/chosen": -217.22317504882812, | |
| "logps/rejected": -348.82867431640625, | |
| "loss": 0.1105, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.2446813583374023, | |
| "rewards/margins": 7.685030460357666, | |
| "rewards/rejected": -9.92971134185791, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.5354666666666668, | |
| "grad_norm": 4.384338855743408, | |
| "learning_rate": 2.534547206682136e-06, | |
| "logits/chosen": -1.691664457321167, | |
| "logits/rejected": -1.6016451120376587, | |
| "logps/chosen": -232.01016235351562, | |
| "logps/rejected": -377.1441345214844, | |
| "loss": 0.0828, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.3139617443084717, | |
| "rewards/margins": 8.008916854858398, | |
| "rewards/rejected": -10.32287883758545, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.5461333333333334, | |
| "grad_norm": 1.2283564805984497, | |
| "learning_rate": 2.5057580459603947e-06, | |
| "logits/chosen": -1.7561960220336914, | |
| "logits/rejected": -1.623084306716919, | |
| "logps/chosen": -205.0697784423828, | |
| "logps/rejected": -363.44268798828125, | |
| "loss": 0.1036, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.9934345483779907, | |
| "rewards/margins": 8.515266418457031, | |
| "rewards/rejected": -10.508700370788574, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.5568, | |
| "grad_norm": 2.0477917194366455, | |
| "learning_rate": 2.4769681216115384e-06, | |
| "logits/chosen": -1.770697832107544, | |
| "logits/rejected": -1.6845451593399048, | |
| "logps/chosen": -223.17129516601562, | |
| "logps/rejected": -366.52569580078125, | |
| "loss": 0.0925, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.7600250244140625, | |
| "rewards/margins": 8.27499771118164, | |
| "rewards/rejected": -10.03502368927002, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.5674666666666668, | |
| "grad_norm": 3.279738187789917, | |
| "learning_rate": 2.448181251730632e-06, | |
| "logits/chosen": -1.71505606174469, | |
| "logits/rejected": -1.6390784978866577, | |
| "logps/chosen": -233.17410278320312, | |
| "logps/rejected": -355.71905517578125, | |
| "loss": 0.1536, | |
| "rewards/accuracies": 0.949999988079071, | |
| "rewards/chosen": -2.3069491386413574, | |
| "rewards/margins": 7.33617639541626, | |
| "rewards/rejected": -9.643125534057617, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.5781333333333334, | |
| "grad_norm": 3.533376693725586, | |
| "learning_rate": 2.419401254007658e-06, | |
| "logits/chosen": -1.6508073806762695, | |
| "logits/rejected": -1.5678499937057495, | |
| "logps/chosen": -246.8570556640625, | |
| "logps/rejected": -379.45648193359375, | |
| "loss": 0.0884, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.69010329246521, | |
| "rewards/margins": 7.795050621032715, | |
| "rewards/rejected": -10.48515510559082, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.5888, | |
| "grad_norm": 1.8679414987564087, | |
| "learning_rate": 2.390631945221221e-06, | |
| "logits/chosen": -1.6903434991836548, | |
| "logits/rejected": -1.565431833267212, | |
| "logps/chosen": -242.61788940429688, | |
| "logps/rejected": -357.1767578125, | |
| "loss": 0.0628, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.191249370574951, | |
| "rewards/margins": 8.514483451843262, | |
| "rewards/rejected": -10.705734252929688, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.5994666666666668, | |
| "grad_norm": 2.0882701873779297, | |
| "learning_rate": 2.361877140732366e-06, | |
| "logits/chosen": -1.691928505897522, | |
| "logits/rejected": -1.5991172790527344, | |
| "logps/chosen": -222.71511840820312, | |
| "logps/rejected": -371.5699462890625, | |
| "loss": 0.0904, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.2086446285247803, | |
| "rewards/margins": 8.20213508605957, | |
| "rewards/rejected": -10.410778999328613, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.6101333333333332, | |
| "grad_norm": 3.39593505859375, | |
| "learning_rate": 2.333140653978592e-06, | |
| "logits/chosen": -1.7964928150177002, | |
| "logits/rejected": -1.6038188934326172, | |
| "logps/chosen": -221.6672821044922, | |
| "logps/rejected": -335.6654357910156, | |
| "loss": 0.0802, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.082045316696167, | |
| "rewards/margins": 8.015445709228516, | |
| "rewards/rejected": -10.097490310668945, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.6208, | |
| "grad_norm": 5.544408798217773, | |
| "learning_rate": 2.3044262959681156e-06, | |
| "logits/chosen": -1.7598270177841187, | |
| "logits/rejected": -1.5819107294082642, | |
| "logps/chosen": -222.6882781982422, | |
| "logps/rejected": -347.28985595703125, | |
| "loss": 0.1282, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -1.9872537851333618, | |
| "rewards/margins": 7.4105939865112305, | |
| "rewards/rejected": -9.397848129272461, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.6314666666666666, | |
| "grad_norm": 1.4111205339431763, | |
| "learning_rate": 2.2757378747744575e-06, | |
| "logits/chosen": -1.763664960861206, | |
| "logits/rejected": -1.5994547605514526, | |
| "logps/chosen": -218.54043579101562, | |
| "logps/rejected": -361.8656921386719, | |
| "loss": 0.0953, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -1.9402878284454346, | |
| "rewards/margins": 8.229757308959961, | |
| "rewards/rejected": -10.170045852661133, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.6421333333333332, | |
| "grad_norm": 4.1563215255737305, | |
| "learning_rate": 2.24707919503142e-06, | |
| "logits/chosen": -1.773432970046997, | |
| "logits/rejected": -1.6846001148223877, | |
| "logps/chosen": -217.88540649414062, | |
| "logps/rejected": -382.12823486328125, | |
| "loss": 0.0899, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -2.1007978916168213, | |
| "rewards/margins": 8.273159980773926, | |
| "rewards/rejected": -10.373956680297852, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.6528, | |
| "grad_norm": 2.1946141719818115, | |
| "learning_rate": 2.2184540574285213e-06, | |
| "logits/chosen": -1.786668062210083, | |
| "logits/rejected": -1.6237224340438843, | |
| "logps/chosen": -258.10015869140625, | |
| "logps/rejected": -369.31195068359375, | |
| "loss": 0.1073, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.0337066650390625, | |
| "rewards/margins": 7.8874945640563965, | |
| "rewards/rejected": -9.921201705932617, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.6634666666666666, | |
| "grad_norm": 3.9432976245880127, | |
| "learning_rate": 2.1898662582069486e-06, | |
| "logits/chosen": -1.7747735977172852, | |
| "logits/rejected": -1.6449607610702515, | |
| "logps/chosen": -211.96719360351562, | |
| "logps/rejected": -374.8441467285156, | |
| "loss": 0.0624, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.5460779666900635, | |
| "rewards/margins": 8.801756858825684, | |
| "rewards/rejected": -10.347834587097168, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.6741333333333333, | |
| "grad_norm": 2.114940643310547, | |
| "learning_rate": 2.1613195886561044e-06, | |
| "logits/chosen": -1.797666311264038, | |
| "logits/rejected": -1.6107561588287354, | |
| "logps/chosen": -219.36013793945312, | |
| "logps/rejected": -354.28033447265625, | |
| "loss": 0.0811, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.8233768939971924, | |
| "rewards/margins": 8.105354309082031, | |
| "rewards/rejected": -9.928730964660645, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.6848, | |
| "grad_norm": 1.5351241827011108, | |
| "learning_rate": 2.132817834610811e-06, | |
| "logits/chosen": -1.751610517501831, | |
| "logits/rejected": -1.5730657577514648, | |
| "logps/chosen": -222.4546356201172, | |
| "logps/rejected": -355.1824035644531, | |
| "loss": 0.0647, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -1.9879817962646484, | |
| "rewards/margins": 8.064549446105957, | |
| "rewards/rejected": -10.052530288696289, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.6954666666666667, | |
| "grad_norm": 3.84106183052063, | |
| "learning_rate": 2.1043647759492338e-06, | |
| "logits/chosen": -1.5872787237167358, | |
| "logits/rejected": -1.5767438411712646, | |
| "logps/chosen": -239.85684204101562, | |
| "logps/rejected": -348.74920654296875, | |
| "loss": 0.0847, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.5143554210662842, | |
| "rewards/margins": 7.725428581237793, | |
| "rewards/rejected": -9.239785194396973, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.7061333333333333, | |
| "grad_norm": 3.106780767440796, | |
| "learning_rate": 2.075964186091599e-06, | |
| "logits/chosen": -1.7316697835922241, | |
| "logits/rejected": -1.650206208229065, | |
| "logps/chosen": -226.59616088867188, | |
| "logps/rejected": -354.37823486328125, | |
| "loss": 0.093, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.121168613433838, | |
| "rewards/margins": 7.989222526550293, | |
| "rewards/rejected": -10.110391616821289, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.7168, | |
| "grad_norm": 1.7571396827697754, | |
| "learning_rate": 2.047619831499765e-06, | |
| "logits/chosen": -1.7232033014297485, | |
| "logits/rejected": -1.568366289138794, | |
| "logps/chosen": -234.1296844482422, | |
| "logps/rejected": -346.7981872558594, | |
| "loss": 0.1052, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.1577060222625732, | |
| "rewards/margins": 7.079756736755371, | |
| "rewards/rejected": -9.237462997436523, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.7274666666666667, | |
| "grad_norm": 2.546520948410034, | |
| "learning_rate": 2.0193354711777155e-06, | |
| "logits/chosen": -1.7943166494369507, | |
| "logits/rejected": -1.6009585857391357, | |
| "logps/chosen": -238.33779907226562, | |
| "logps/rejected": -359.30426025390625, | |
| "loss": 0.0945, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -1.741201639175415, | |
| "rewards/margins": 8.233468055725098, | |
| "rewards/rejected": -9.974669456481934, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.7381333333333333, | |
| "grad_norm": 4.782870769500732, | |
| "learning_rate": 1.991114856173048e-06, | |
| "logits/chosen": -1.7833083868026733, | |
| "logits/rejected": -1.6827278137207031, | |
| "logps/chosen": -235.49661254882812, | |
| "logps/rejected": -378.20562744140625, | |
| "loss": 0.0973, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.7793164253234863, | |
| "rewards/margins": 8.449636459350586, | |
| "rewards/rejected": -11.228952407836914, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.7488000000000001, | |
| "grad_norm": 1.997058391571045, | |
| "learning_rate": 1.9629617290795115e-06, | |
| "logits/chosen": -1.6820003986358643, | |
| "logits/rejected": -1.6001437902450562, | |
| "logps/chosen": -212.8640594482422, | |
| "logps/rejected": -348.04241943359375, | |
| "loss": 0.0829, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.0805208683013916, | |
| "rewards/margins": 7.856542110443115, | |
| "rewards/rejected": -9.937063217163086, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.7594666666666665, | |
| "grad_norm": 1.1324129104614258, | |
| "learning_rate": 1.934879823540663e-06, | |
| "logits/chosen": -1.7404649257659912, | |
| "logits/rejected": -1.5713565349578857, | |
| "logps/chosen": -217.12887573242188, | |
| "logps/rejected": -357.5457458496094, | |
| "loss": 0.0789, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.7752182483673096, | |
| "rewards/margins": 8.235331535339355, | |
| "rewards/rejected": -10.01054859161377, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.7701333333333333, | |
| "grad_norm": 0.8319897651672363, | |
| "learning_rate": 1.9068728637547195e-06, | |
| "logits/chosen": -1.713891625404358, | |
| "logits/rejected": -1.6125208139419556, | |
| "logps/chosen": -236.43612670898438, | |
| "logps/rejected": -374.36236572265625, | |
| "loss": 0.0656, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.294593572616577, | |
| "rewards/margins": 8.278243064880371, | |
| "rewards/rejected": -10.572836875915527, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.7808000000000002, | |
| "grad_norm": 1.1885546445846558, | |
| "learning_rate": 1.878944563980658e-06, | |
| "logits/chosen": -1.659051537513733, | |
| "logits/rejected": -1.5556640625, | |
| "logps/chosen": -217.66622924804688, | |
| "logps/rejected": -363.8323669433594, | |
| "loss": 0.0698, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.268317461013794, | |
| "rewards/margins": 7.98989200592041, | |
| "rewards/rejected": -10.258210182189941, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.7914666666666665, | |
| "grad_norm": 3.575385332107544, | |
| "learning_rate": 1.85109862804563e-06, | |
| "logits/chosen": -1.717116117477417, | |
| "logits/rejected": -1.5746225118637085, | |
| "logps/chosen": -245.6321258544922, | |
| "logps/rejected": -334.73565673828125, | |
| "loss": 0.0874, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.0019404888153076, | |
| "rewards/margins": 7.676904201507568, | |
| "rewards/rejected": -9.678845405578613, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.8021333333333334, | |
| "grad_norm": 1.076490044593811, | |
| "learning_rate": 1.823338748853766e-06, | |
| "logits/chosen": -1.7594982385635376, | |
| "logits/rejected": -1.6361639499664307, | |
| "logps/chosen": -230.13330078125, | |
| "logps/rejected": -371.546875, | |
| "loss": 0.0682, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.2079403400421143, | |
| "rewards/margins": 8.9056396484375, | |
| "rewards/rejected": -11.113580703735352, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.8128, | |
| "grad_norm": 1.023444414138794, | |
| "learning_rate": 1.7956686078964257e-06, | |
| "logits/chosen": -1.785599946975708, | |
| "logits/rejected": -1.6359217166900635, | |
| "logps/chosen": -234.2008819580078, | |
| "logps/rejected": -333.7929992675781, | |
| "loss": 0.1263, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.184248447418213, | |
| "rewards/margins": 7.763758659362793, | |
| "rewards/rejected": -9.948007583618164, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.8234666666666666, | |
| "grad_norm": 5.103050231933594, | |
| "learning_rate": 1.7680918747639608e-06, | |
| "logits/chosen": -1.7238714694976807, | |
| "logits/rejected": -1.6119329929351807, | |
| "logps/chosen": -235.1346893310547, | |
| "logps/rejected": -385.647705078125, | |
| "loss": 0.0734, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.530133008956909, | |
| "rewards/margins": 8.610017776489258, | |
| "rewards/rejected": -11.14015007019043, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.8341333333333334, | |
| "grad_norm": 2.734606981277466, | |
| "learning_rate": 1.740612206659057e-06, | |
| "logits/chosen": -1.798112154006958, | |
| "logits/rejected": -1.5648561716079712, | |
| "logps/chosen": -229.7447052001953, | |
| "logps/rejected": -367.8794860839844, | |
| "loss": 0.0698, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.2181127071380615, | |
| "rewards/margins": 8.9027681350708, | |
| "rewards/rejected": -11.120881080627441, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.8448, | |
| "grad_norm": 1.9527674913406372, | |
| "learning_rate": 1.7132332479117204e-06, | |
| "logits/chosen": -1.780371904373169, | |
| "logits/rejected": -1.5965863466262817, | |
| "logps/chosen": -245.5487518310547, | |
| "logps/rejected": -359.4418029785156, | |
| "loss": 0.1013, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.1059577465057373, | |
| "rewards/margins": 8.449846267700195, | |
| "rewards/rejected": -10.555803298950195, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.8554666666666666, | |
| "grad_norm": 6.878162384033203, | |
| "learning_rate": 1.6859586294959666e-06, | |
| "logits/chosen": -1.7246229648590088, | |
| "logits/rejected": -1.6141942739486694, | |
| "logps/chosen": -234.10302734375, | |
| "logps/rejected": -347.4426574707031, | |
| "loss": 0.0942, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.0066280364990234, | |
| "rewards/margins": 8.30981731414795, | |
| "rewards/rejected": -10.316446304321289, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.8661333333333334, | |
| "grad_norm": 2.037459135055542, | |
| "learning_rate": 1.6587919685482884e-06, | |
| "logits/chosen": -1.8677127361297607, | |
| "logits/rejected": -1.7134828567504883, | |
| "logps/chosen": -212.8134002685547, | |
| "logps/rejected": -393.59124755859375, | |
| "loss": 0.0808, | |
| "rewards/accuracies": 0.956250011920929, | |
| "rewards/chosen": -2.3060128688812256, | |
| "rewards/margins": 8.69517993927002, | |
| "rewards/rejected": -11.001192092895508, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.8768, | |
| "grad_norm": 1.5264317989349365, | |
| "learning_rate": 1.6317368678879497e-06, | |
| "logits/chosen": -1.6491420269012451, | |
| "logits/rejected": -1.5571231842041016, | |
| "logps/chosen": -231.6708526611328, | |
| "logps/rejected": -384.9656677246094, | |
| "loss": 0.0894, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.254824161529541, | |
| "rewards/margins": 9.314870834350586, | |
| "rewards/rejected": -11.569695472717285, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.8874666666666666, | |
| "grad_norm": 2.8329501152038574, | |
| "learning_rate": 1.6047969155391856e-06, | |
| "logits/chosen": -1.6241254806518555, | |
| "logits/rejected": -1.6059930324554443, | |
| "logps/chosen": -242.1722412109375, | |
| "logps/rejected": -375.7530517578125, | |
| "loss": 0.0708, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.215430498123169, | |
| "rewards/margins": 8.22697639465332, | |
| "rewards/rejected": -10.442407608032227, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.8981333333333335, | |
| "grad_norm": 1.2483875751495361, | |
| "learning_rate": 1.5779756842553617e-06, | |
| "logits/chosen": -1.695774793624878, | |
| "logits/rejected": -1.667642593383789, | |
| "logps/chosen": -236.163330078125, | |
| "logps/rejected": -370.6156311035156, | |
| "loss": 0.0599, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.2344157695770264, | |
| "rewards/margins": 8.541747093200684, | |
| "rewards/rejected": -10.776163101196289, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.9088, | |
| "grad_norm": 3.0992672443389893, | |
| "learning_rate": 1.551276731045157e-06, | |
| "logits/chosen": -1.7390620708465576, | |
| "logits/rejected": -1.6380354166030884, | |
| "logps/chosen": -248.9411163330078, | |
| "logps/rejected": -380.0399475097656, | |
| "loss": 0.0641, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.2495522499084473, | |
| "rewards/margins": 8.640344619750977, | |
| "rewards/rejected": -10.889896392822266, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.9194666666666667, | |
| "grad_norm": 1.3069320917129517, | |
| "learning_rate": 1.5247035967008369e-06, | |
| "logits/chosen": -1.6581281423568726, | |
| "logits/rejected": -1.498859167098999, | |
| "logps/chosen": -232.68618774414062, | |
| "logps/rejected": -349.40374755859375, | |
| "loss": 0.0631, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.9037091732025146, | |
| "rewards/margins": 8.36854362487793, | |
| "rewards/rejected": -10.272253036499023, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.9301333333333335, | |
| "grad_norm": 3.359713077545166, | |
| "learning_rate": 1.498259805328678e-06, | |
| "logits/chosen": -1.6838054656982422, | |
| "logits/rejected": -1.529866337776184, | |
| "logps/chosen": -230.3268585205078, | |
| "logps/rejected": -340.8519287109375, | |
| "loss": 0.0802, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.370974063873291, | |
| "rewards/margins": 8.133573532104492, | |
| "rewards/rejected": -10.504546165466309, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 1.9407999999999999, | |
| "grad_norm": 1.0852340459823608, | |
| "learning_rate": 1.471948863881601e-06, | |
| "logits/chosen": -1.7539818286895752, | |
| "logits/rejected": -1.6175880432128906, | |
| "logps/chosen": -244.2122039794922, | |
| "logps/rejected": -387.53924560546875, | |
| "loss": 0.0534, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.2629380226135254, | |
| "rewards/margins": 8.641889572143555, | |
| "rewards/rejected": -10.904828071594238, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 1.9514666666666667, | |
| "grad_norm": 3.436288356781006, | |
| "learning_rate": 1.4457742616940864e-06, | |
| "logits/chosen": -1.6151090860366821, | |
| "logits/rejected": -1.5289099216461182, | |
| "logps/chosen": -245.4002227783203, | |
| "logps/rejected": -371.73565673828125, | |
| "loss": 0.0941, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.361675500869751, | |
| "rewards/margins": 7.663311958312988, | |
| "rewards/rejected": -10.02498722076416, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 1.9621333333333333, | |
| "grad_norm": 3.603616237640381, | |
| "learning_rate": 1.4197394700194165e-06, | |
| "logits/chosen": -1.7797037363052368, | |
| "logits/rejected": -1.6602210998535156, | |
| "logps/chosen": -229.91262817382812, | |
| "logps/rejected": -347.8226013183594, | |
| "loss": 0.0738, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.6596853733062744, | |
| "rewards/margins": 7.811383247375488, | |
| "rewards/rejected": -10.471068382263184, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 1.9727999999999999, | |
| "grad_norm": 5.25838565826416, | |
| "learning_rate": 1.3938479415693244e-06, | |
| "logits/chosen": -1.7488892078399658, | |
| "logits/rejected": -1.6513652801513672, | |
| "logps/chosen": -237.98727416992188, | |
| "logps/rejected": -373.8163757324219, | |
| "loss": 0.0797, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.413729667663574, | |
| "rewards/margins": 9.135477066040039, | |
| "rewards/rejected": -11.54920768737793, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.9834666666666667, | |
| "grad_norm": 4.8256425857543945, | |
| "learning_rate": 1.368103110056096e-06, | |
| "logits/chosen": -1.7451817989349365, | |
| "logits/rejected": -1.6328808069229126, | |
| "logps/chosen": -253.07363891601562, | |
| "logps/rejected": -370.066650390625, | |
| "loss": 0.1011, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.8300654888153076, | |
| "rewards/margins": 8.426507949829102, | |
| "rewards/rejected": -11.256572723388672, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 1.9941333333333333, | |
| "grad_norm": 0.881809413433075, | |
| "learning_rate": 1.3425083897371983e-06, | |
| "logits/chosen": -1.5809837579727173, | |
| "logits/rejected": -1.5179001092910767, | |
| "logps/chosen": -246.60287475585938, | |
| "logps/rejected": -365.4212341308594, | |
| "loss": 0.0904, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.0594959259033203, | |
| "rewards/margins": 8.06749153137207, | |
| "rewards/rejected": -10.126986503601074, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.0042666666666666, | |
| "grad_norm": 3.4227020740509033, | |
| "learning_rate": 1.317067174962478e-06, | |
| "logits/chosen": -1.6698054075241089, | |
| "logits/rejected": -1.4948031902313232, | |
| "logps/chosen": -241.87635803222656, | |
| "logps/rejected": -372.1206970214844, | |
| "loss": 0.0836, | |
| "rewards/accuracies": 0.9868420958518982, | |
| "rewards/chosen": -2.477794647216797, | |
| "rewards/margins": 8.646845817565918, | |
| "rewards/rejected": -11.124640464782715, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.0149333333333335, | |
| "grad_norm": 2.4874966144561768, | |
| "learning_rate": 1.291782839724009e-06, | |
| "logits/chosen": -1.6582422256469727, | |
| "logits/rejected": -1.6185009479522705, | |
| "logps/chosen": -228.345947265625, | |
| "logps/rejected": -355.91973876953125, | |
| "loss": 0.073, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.1200289726257324, | |
| "rewards/margins": 8.30976390838623, | |
| "rewards/rejected": -10.429792404174805, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.0256, | |
| "grad_norm": 1.4161961078643799, | |
| "learning_rate": 1.2666587372086352e-06, | |
| "logits/chosen": -1.6435474157333374, | |
| "logits/rejected": -1.5627238750457764, | |
| "logps/chosen": -209.0333709716797, | |
| "logps/rejected": -350.1138916015625, | |
| "loss": 0.0693, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.8323615789413452, | |
| "rewards/margins": 8.273909568786621, | |
| "rewards/rejected": -10.106269836425781, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.0362666666666667, | |
| "grad_norm": 3.069528102874756, | |
| "learning_rate": 1.241698199353275e-06, | |
| "logits/chosen": -1.7319631576538086, | |
| "logits/rejected": -1.5979344844818115, | |
| "logps/chosen": -216.0243377685547, | |
| "logps/rejected": -368.0318298339844, | |
| "loss": 0.0599, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.209562301635742, | |
| "rewards/margins": 8.545756340026855, | |
| "rewards/rejected": -10.755319595336914, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.0469333333333335, | |
| "grad_norm": 5.300919532775879, | |
| "learning_rate": 1.2169045364030404e-06, | |
| "logits/chosen": -1.7645591497421265, | |
| "logits/rejected": -1.62359619140625, | |
| "logps/chosen": -227.6970672607422, | |
| "logps/rejected": -369.8218078613281, | |
| "loss": 0.0689, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.6951305866241455, | |
| "rewards/margins": 8.503260612487793, | |
| "rewards/rejected": -11.19839096069336, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.0576, | |
| "grad_norm": 2.6272425651550293, | |
| "learning_rate": 1.1922810364722402e-06, | |
| "logits/chosen": -1.7294490337371826, | |
| "logits/rejected": -1.5723257064819336, | |
| "logps/chosen": -243.034423828125, | |
| "logps/rejected": -356.21685791015625, | |
| "loss": 0.0824, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.0579140186309814, | |
| "rewards/margins": 8.599225044250488, | |
| "rewards/rejected": -10.65713882446289, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.0682666666666667, | |
| "grad_norm": 3.7131705284118652, | |
| "learning_rate": 1.1678309651083055e-06, | |
| "logits/chosen": -1.7538963556289673, | |
| "logits/rejected": -1.6433712244033813, | |
| "logps/chosen": -242.6889190673828, | |
| "logps/rejected": -358.15277099609375, | |
| "loss": 0.0524, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.461589813232422, | |
| "rewards/margins": 8.619664192199707, | |
| "rewards/rejected": -11.081254959106445, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.0789333333333335, | |
| "grad_norm": 6.3900909423828125, | |
| "learning_rate": 1.1435575648587224e-06, | |
| "logits/chosen": -1.6953847408294678, | |
| "logits/rejected": -1.621117353439331, | |
| "logps/chosen": -233.9971923828125, | |
| "logps/rejected": -375.391845703125, | |
| "loss": 0.0777, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.4303104877471924, | |
| "rewards/margins": 9.11983585357666, | |
| "rewards/rejected": -11.550145149230957, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.0896, | |
| "grad_norm": 0.9288052916526794, | |
| "learning_rate": 1.1194640548410039e-06, | |
| "logits/chosen": -1.6072397232055664, | |
| "logits/rejected": -1.5468966960906982, | |
| "logps/chosen": -229.78286743164062, | |
| "logps/rejected": -353.2090759277344, | |
| "loss": 0.0697, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.165677547454834, | |
| "rewards/margins": 8.486506462097168, | |
| "rewards/rejected": -10.65218448638916, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.1002666666666667, | |
| "grad_norm": 1.0610967874526978, | |
| "learning_rate": 1.095553630315774e-06, | |
| "logits/chosen": -1.7196861505508423, | |
| "logits/rejected": -1.606482744216919, | |
| "logps/chosen": -221.6151580810547, | |
| "logps/rejected": -366.8529052734375, | |
| "loss": 0.0894, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -1.9320646524429321, | |
| "rewards/margins": 8.738572120666504, | |
| "rewards/rejected": -10.670637130737305, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.1109333333333336, | |
| "grad_norm": 1.3458142280578613, | |
| "learning_rate": 1.0718294622630188e-06, | |
| "logits/chosen": -1.7000640630722046, | |
| "logits/rejected": -1.6278724670410156, | |
| "logps/chosen": -229.53878784179688, | |
| "logps/rejected": -381.9530029296875, | |
| "loss": 0.0627, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.1144888401031494, | |
| "rewards/margins": 8.574979782104492, | |
| "rewards/rejected": -10.689468383789062, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.1216, | |
| "grad_norm": 2.989312171936035, | |
| "learning_rate": 1.0482946969615495e-06, | |
| "logits/chosen": -1.7484781742095947, | |
| "logits/rejected": -1.588270664215088, | |
| "logps/chosen": -246.27987670898438, | |
| "logps/rejected": -367.8683166503906, | |
| "loss": 0.0533, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.7404978275299072, | |
| "rewards/margins": 8.548757553100586, | |
| "rewards/rejected": -11.289255142211914, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.1322666666666668, | |
| "grad_norm": 1.4040266275405884, | |
| "learning_rate": 1.0249524555717461e-06, | |
| "logits/chosen": -1.7742398977279663, | |
| "logits/rejected": -1.6097848415374756, | |
| "logps/chosen": -223.71377563476562, | |
| "logps/rejected": -370.94305419921875, | |
| "loss": 0.0799, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.451437473297119, | |
| "rewards/margins": 8.23726749420166, | |
| "rewards/rejected": -10.688704490661621, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.142933333333333, | |
| "grad_norm": 1.852662444114685, | |
| "learning_rate": 1.0018058337216327e-06, | |
| "logits/chosen": -1.6842467784881592, | |
| "logits/rejected": -1.6574596166610718, | |
| "logps/chosen": -249.32254028320312, | |
| "logps/rejected": -369.68927001953125, | |
| "loss": 0.0656, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.577439785003662, | |
| "rewards/margins": 8.032537460327148, | |
| "rewards/rejected": -10.609979629516602, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.1536, | |
| "grad_norm": 3.422244071960449, | |
| "learning_rate": 9.788579010963387e-07, | |
| "logits/chosen": -1.7132316827774048, | |
| "logits/rejected": -1.5610625743865967, | |
| "logps/chosen": -246.59152221679688, | |
| "logps/rejected": -363.26898193359375, | |
| "loss": 0.0596, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.5583107471466064, | |
| "rewards/margins": 8.616121292114258, | |
| "rewards/rejected": -11.174431800842285, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.164266666666667, | |
| "grad_norm": 3.149967908859253, | |
| "learning_rate": 9.56111701030998e-07, | |
| "logits/chosen": -1.8254941701889038, | |
| "logits/rejected": -1.6014741659164429, | |
| "logps/chosen": -257.9132080078125, | |
| "logps/rejected": -370.2958984375, | |
| "loss": 0.0748, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.586059093475342, | |
| "rewards/margins": 8.62619400024414, | |
| "rewards/rejected": -11.212254524230957, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 2.174933333333333, | |
| "grad_norm": 0.670924723148346, | |
| "learning_rate": 9.335702501071459e-07, | |
| "logits/chosen": -1.6369680166244507, | |
| "logits/rejected": -1.4811519384384155, | |
| "logps/chosen": -254.5886993408203, | |
| "logps/rejected": -366.7772216796875, | |
| "loss": 0.0637, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.6869049072265625, | |
| "rewards/margins": 8.372596740722656, | |
| "rewards/rejected": -11.059501647949219, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.1856, | |
| "grad_norm": 0.4992627203464508, | |
| "learning_rate": 9.112365377526636e-07, | |
| "logits/chosen": -1.7114737033843994, | |
| "logits/rejected": -1.6579517126083374, | |
| "logps/chosen": -232.11563110351562, | |
| "logps/rejected": -392.6571350097656, | |
| "loss": 0.0379, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.052781581878662, | |
| "rewards/margins": 9.346757888793945, | |
| "rewards/rejected": -12.39953899383545, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.196266666666667, | |
| "grad_norm": 0.7529945373535156, | |
| "learning_rate": 8.891135258453195e-07, | |
| "logits/chosen": -1.6109974384307861, | |
| "logits/rejected": -1.518336534500122, | |
| "logps/chosen": -221.02468872070312, | |
| "logps/rejected": -362.6771240234375, | |
| "loss": 0.0525, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.2659616470336914, | |
| "rewards/margins": 9.202993392944336, | |
| "rewards/rejected": -11.468955993652344, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.206933333333333, | |
| "grad_norm": 2.34568190574646, | |
| "learning_rate": 8.672041483199709e-07, | |
| "logits/chosen": -1.7083330154418945, | |
| "logits/rejected": -1.5656239986419678, | |
| "logps/chosen": -244.4073486328125, | |
| "logps/rejected": -343.1373291015625, | |
| "loss": 0.0595, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.3901219367980957, | |
| "rewards/margins": 7.6386895179748535, | |
| "rewards/rejected": -10.028810501098633, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 2.2176, | |
| "grad_norm": 0.49317407608032227, | |
| "learning_rate": 8.455113107794652e-07, | |
| "logits/chosen": -1.6537469625473022, | |
| "logits/rejected": -1.6144192218780518, | |
| "logps/chosen": -217.38223266601562, | |
| "logps/rejected": -389.8377685546875, | |
| "loss": 0.0596, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.900716781616211, | |
| "rewards/margins": 9.58027458190918, | |
| "rewards/rejected": -12.480993270874023, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.228266666666667, | |
| "grad_norm": 2.7996599674224854, | |
| "learning_rate": 8.240378901093035e-07, | |
| "logits/chosen": -1.7323932647705078, | |
| "logits/rejected": -1.5816843509674072, | |
| "logps/chosen": -245.81143188476562, | |
| "logps/rejected": -368.0496826171875, | |
| "loss": 0.0609, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.495245933532715, | |
| "rewards/margins": 8.47819995880127, | |
| "rewards/rejected": -10.973443984985352, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 2.238933333333333, | |
| "grad_norm": 0.7402082681655884, | |
| "learning_rate": 8.027867340961113e-07, | |
| "logits/chosen": -1.5743385553359985, | |
| "logits/rejected": -1.4896413087844849, | |
| "logps/chosen": -235.98117065429688, | |
| "logps/rejected": -354.88763427734375, | |
| "loss": 0.0608, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.446408987045288, | |
| "rewards/margins": 8.018033981323242, | |
| "rewards/rejected": -10.464442253112793, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.2496, | |
| "grad_norm": 1.0578789710998535, | |
| "learning_rate": 7.817606610499639e-07, | |
| "logits/chosen": -1.7072864770889282, | |
| "logits/rejected": -1.6435352563858032, | |
| "logps/chosen": -240.1691436767578, | |
| "logps/rejected": -382.712158203125, | |
| "loss": 0.0511, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.5651392936706543, | |
| "rewards/margins": 8.454736709594727, | |
| "rewards/rejected": -11.019875526428223, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 2.2602666666666664, | |
| "grad_norm": 0.5394600629806519, | |
| "learning_rate": 7.609624594306278e-07, | |
| "logits/chosen": -1.7428966760635376, | |
| "logits/rejected": -1.643012285232544, | |
| "logps/chosen": -247.66360473632812, | |
| "logps/rejected": -390.7719421386719, | |
| "loss": 0.0475, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.6858017444610596, | |
| "rewards/margins": 8.323606491088867, | |
| "rewards/rejected": -11.009407043457031, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.2709333333333332, | |
| "grad_norm": 7.077690124511719, | |
| "learning_rate": 7.403948874777556e-07, | |
| "logits/chosen": -1.6925182342529297, | |
| "logits/rejected": -1.5964515209197998, | |
| "logps/chosen": -229.00881958007812, | |
| "logps/rejected": -373.58740234375, | |
| "loss": 0.0589, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.457432508468628, | |
| "rewards/margins": 8.91236686706543, | |
| "rewards/rejected": -11.369799613952637, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 2.2816, | |
| "grad_norm": 1.9213427305221558, | |
| "learning_rate": 7.20060672845091e-07, | |
| "logits/chosen": -1.6789592504501343, | |
| "logits/rejected": -1.6316735744476318, | |
| "logps/chosen": -226.3343505859375, | |
| "logps/rejected": -381.14398193359375, | |
| "loss": 0.0501, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.6402697563171387, | |
| "rewards/margins": 9.172629356384277, | |
| "rewards/rejected": -11.812898635864258, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.2922666666666665, | |
| "grad_norm": 2.5269217491149902, | |
| "learning_rate": 6.99962512238733e-07, | |
| "logits/chosen": -1.657263994216919, | |
| "logits/rejected": -1.5200378894805908, | |
| "logps/chosen": -226.5382537841797, | |
| "logps/rejected": -372.1704406738281, | |
| "loss": 0.1022, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.4905364513397217, | |
| "rewards/margins": 8.78575325012207, | |
| "rewards/rejected": -11.276289939880371, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.3029333333333333, | |
| "grad_norm": 2.181729555130005, | |
| "learning_rate": 6.801030710594949e-07, | |
| "logits/chosen": -1.628305435180664, | |
| "logits/rejected": -1.5248959064483643, | |
| "logps/chosen": -246.23812866210938, | |
| "logps/rejected": -402.61614990234375, | |
| "loss": 0.0546, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.856438159942627, | |
| "rewards/margins": 9.355535507202148, | |
| "rewards/rejected": -12.211973190307617, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.3136, | |
| "grad_norm": 1.0526963472366333, | |
| "learning_rate": 6.604849830494264e-07, | |
| "logits/chosen": -1.700221061706543, | |
| "logits/rejected": -1.608945608139038, | |
| "logps/chosen": -239.19741821289062, | |
| "logps/rejected": -369.9862976074219, | |
| "loss": 0.0432, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.78136944770813, | |
| "rewards/margins": 8.838476181030273, | |
| "rewards/rejected": -11.619847297668457, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.3242666666666665, | |
| "grad_norm": 0.7994614839553833, | |
| "learning_rate": 6.411108499425253e-07, | |
| "logits/chosen": -1.7244764566421509, | |
| "logits/rejected": -1.5778989791870117, | |
| "logps/chosen": -234.7858123779297, | |
| "logps/rejected": -374.0656433105469, | |
| "loss": 0.0539, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.5700016021728516, | |
| "rewards/margins": 9.354166984558105, | |
| "rewards/rejected": -11.924168586730957, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.3349333333333333, | |
| "grad_norm": 0.7055251598358154, | |
| "learning_rate": 6.219832411197013e-07, | |
| "logits/chosen": -1.678105354309082, | |
| "logits/rejected": -1.5513595342636108, | |
| "logps/chosen": -240.69384765625, | |
| "logps/rejected": -375.53179931640625, | |
| "loss": 0.0666, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.769808053970337, | |
| "rewards/margins": 9.200688362121582, | |
| "rewards/rejected": -11.970497131347656, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.3456, | |
| "grad_norm": 0.7503064870834351, | |
| "learning_rate": 6.031046932680229e-07, | |
| "logits/chosen": -1.69338059425354, | |
| "logits/rejected": -1.5614005327224731, | |
| "logps/chosen": -256.23828125, | |
| "logps/rejected": -392.1869201660156, | |
| "loss": 0.0492, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.936000108718872, | |
| "rewards/margins": 9.314740180969238, | |
| "rewards/rejected": -12.250740051269531, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.3562666666666665, | |
| "grad_norm": 0.8023340702056885, | |
| "learning_rate": 5.844777100443091e-07, | |
| "logits/chosen": -1.7291017770767212, | |
| "logits/rejected": -1.5592310428619385, | |
| "logps/chosen": -237.56771850585938, | |
| "logps/rejected": -390.6042785644531, | |
| "loss": 0.062, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.023751735687256, | |
| "rewards/margins": 9.082342147827148, | |
| "rewards/rejected": -12.106094360351562, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.3669333333333333, | |
| "grad_norm": 0.3595770299434662, | |
| "learning_rate": 5.66104761743092e-07, | |
| "logits/chosen": -1.7586097717285156, | |
| "logits/rejected": -1.6678698062896729, | |
| "logps/chosen": -252.323974609375, | |
| "logps/rejected": -401.20147705078125, | |
| "loss": 0.042, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.7489242553710938, | |
| "rewards/margins": 9.122079849243164, | |
| "rewards/rejected": -12.871004104614258, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.3776, | |
| "grad_norm": 3.4164505004882812, | |
| "learning_rate": 5.479882849690102e-07, | |
| "logits/chosen": -1.732012391090393, | |
| "logits/rejected": -1.5598955154418945, | |
| "logps/chosen": -234.59298706054688, | |
| "logps/rejected": -378.66162109375, | |
| "loss": 0.0873, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.783665895462036, | |
| "rewards/margins": 9.570892333984375, | |
| "rewards/rejected": -12.354558944702148, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.3882666666666665, | |
| "grad_norm": 1.439928650856018, | |
| "learning_rate": 5.301306823136687e-07, | |
| "logits/chosen": -1.664751410484314, | |
| "logits/rejected": -1.590052604675293, | |
| "logps/chosen": -214.80111694335938, | |
| "logps/rejected": -394.08221435546875, | |
| "loss": 0.0645, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.2062828540802, | |
| "rewards/margins": 9.442559242248535, | |
| "rewards/rejected": -11.648841857910156, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.3989333333333334, | |
| "grad_norm": 1.5526204109191895, | |
| "learning_rate": 5.125343220370088e-07, | |
| "logits/chosen": -1.6302855014801025, | |
| "logits/rejected": -1.553674578666687, | |
| "logps/chosen": -257.54541015625, | |
| "logps/rejected": -374.77142333984375, | |
| "loss": 0.0523, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.0847949981689453, | |
| "rewards/margins": 8.612579345703125, | |
| "rewards/rejected": -11.69737434387207, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.4096, | |
| "grad_norm": 0.7676222324371338, | |
| "learning_rate": 4.952015377532329e-07, | |
| "logits/chosen": -1.8063156604766846, | |
| "logits/rejected": -1.6408159732818604, | |
| "logps/chosen": -240.57730102539062, | |
| "logps/rejected": -386.2501220703125, | |
| "loss": 0.0612, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.309523820877075, | |
| "rewards/margins": 9.168109893798828, | |
| "rewards/rejected": -12.47763442993164, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.4202666666666666, | |
| "grad_norm": 2.877019166946411, | |
| "learning_rate": 4.781346281213204e-07, | |
| "logits/chosen": -1.8087190389633179, | |
| "logits/rejected": -1.6685491800308228, | |
| "logps/chosen": -243.5975799560547, | |
| "logps/rejected": -416.5228576660156, | |
| "loss": 0.0539, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.3564605712890625, | |
| "rewards/margins": 9.94178581237793, | |
| "rewards/rejected": -13.298245429992676, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.4309333333333334, | |
| "grad_norm": 3.171461582183838, | |
| "learning_rate": 4.613358565401818e-07, | |
| "logits/chosen": -1.6455013751983643, | |
| "logits/rejected": -1.5212013721466064, | |
| "logps/chosen": -240.3773193359375, | |
| "logps/rejected": -380.5497131347656, | |
| "loss": 0.076, | |
| "rewards/accuracies": 0.9624999761581421, | |
| "rewards/chosen": -2.7457993030548096, | |
| "rewards/margins": 8.988884925842285, | |
| "rewards/rejected": -11.7346830368042, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.4416, | |
| "grad_norm": 0.8925107717514038, | |
| "learning_rate": 4.448074508484906e-07, | |
| "logits/chosen": -1.6455951929092407, | |
| "logits/rejected": -1.4987901449203491, | |
| "logps/chosen": -257.70379638671875, | |
| "logps/rejected": -374.9892272949219, | |
| "loss": 0.0572, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.067699909210205, | |
| "rewards/margins": 9.026286125183105, | |
| "rewards/rejected": -12.093984603881836, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.4522666666666666, | |
| "grad_norm": 1.1913326978683472, | |
| "learning_rate": 4.2855160302922777e-07, | |
| "logits/chosen": -1.690572738647461, | |
| "logits/rejected": -1.6463066339492798, | |
| "logps/chosen": -234.546630859375, | |
| "logps/rejected": -398.32763671875, | |
| "loss": 0.0459, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -3.1086277961730957, | |
| "rewards/margins": 9.187312126159668, | |
| "rewards/rejected": -12.295941352844238, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.4629333333333334, | |
| "grad_norm": 2.599015235900879, | |
| "learning_rate": 4.125704689189819e-07, | |
| "logits/chosen": -1.5713093280792236, | |
| "logits/rejected": -1.5037789344787598, | |
| "logps/chosen": -241.7811737060547, | |
| "logps/rejected": -394.9588623046875, | |
| "loss": 0.066, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.8472790718078613, | |
| "rewards/margins": 9.292223930358887, | |
| "rewards/rejected": -12.139504432678223, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.4736000000000002, | |
| "grad_norm": 3.4746813774108887, | |
| "learning_rate": 3.9686616792204677e-07, | |
| "logits/chosen": -1.671574354171753, | |
| "logits/rejected": -1.6419601440429688, | |
| "logps/chosen": -243.90811157226562, | |
| "logps/rejected": -380.66265869140625, | |
| "loss": 0.0644, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -2.9403159618377686, | |
| "rewards/margins": 8.807522773742676, | |
| "rewards/rejected": -11.747838020324707, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.4842666666666666, | |
| "grad_norm": 1.217083215713501, | |
| "learning_rate": 3.8144078272934676e-07, | |
| "logits/chosen": -1.616916298866272, | |
| "logits/rejected": -1.5705225467681885, | |
| "logps/chosen": -246.2116241455078, | |
| "logps/rejected": -400.27520751953125, | |
| "loss": 0.054, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.046478748321533, | |
| "rewards/margins": 9.513192176818848, | |
| "rewards/rejected": -12.559671401977539, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.4949333333333334, | |
| "grad_norm": 0.5213105082511902, | |
| "learning_rate": 3.6629635904222943e-07, | |
| "logits/chosen": -1.632601022720337, | |
| "logits/rejected": -1.580075740814209, | |
| "logps/chosen": -242.1397705078125, | |
| "logps/rejected": -392.51708984375, | |
| "loss": 0.035, | |
| "rewards/accuracies": 1.0, | |
| "rewards/chosen": -2.779862642288208, | |
| "rewards/margins": 9.45389461517334, | |
| "rewards/rejected": -12.233755111694336, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.5056000000000003, | |
| "grad_norm": 4.385895729064941, | |
| "learning_rate": 3.514349053011698e-07, | |
| "logits/chosen": -1.5838334560394287, | |
| "logits/rejected": -1.48439621925354, | |
| "logps/chosen": -240.0968475341797, | |
| "logps/rejected": -380.6497802734375, | |
| "loss": 0.0528, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.24969482421875, | |
| "rewards/margins": 9.019821166992188, | |
| "rewards/rejected": -12.269515991210938, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.5162666666666667, | |
| "grad_norm": 1.1202504634857178, | |
| "learning_rate": 3.3685839241941133e-07, | |
| "logits/chosen": -1.6621793508529663, | |
| "logits/rejected": -1.585744857788086, | |
| "logps/chosen": -241.1162872314453, | |
| "logps/rejected": -379.06610107421875, | |
| "loss": 0.053, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.787099838256836, | |
| "rewards/margins": 9.54288101196289, | |
| "rewards/rejected": -12.329981803894043, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.5269333333333335, | |
| "grad_norm": 0.898692786693573, | |
| "learning_rate": 3.2256875352158435e-07, | |
| "logits/chosen": -1.6413805484771729, | |
| "logits/rejected": -1.5792820453643799, | |
| "logps/chosen": -241.3943634033203, | |
| "logps/rejected": -397.4687805175781, | |
| "loss": 0.0829, | |
| "rewards/accuracies": 0.96875, | |
| "rewards/chosen": -3.0518360137939453, | |
| "rewards/margins": 9.231449127197266, | |
| "rewards/rejected": -12.283285140991211, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.5376, | |
| "grad_norm": 1.140761375427246, | |
| "learning_rate": 3.085678836873399e-07, | |
| "logits/chosen": -1.6138441562652588, | |
| "logits/rejected": -1.5633118152618408, | |
| "logps/chosen": -232.3341827392578, | |
| "logps/rejected": -381.39111328125, | |
| "loss": 0.0851, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.112748146057129, | |
| "rewards/margins": 8.837594985961914, | |
| "rewards/rejected": -11.95034408569336, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.5482666666666667, | |
| "grad_norm": 4.43955135345459, | |
| "learning_rate": 2.9485763970002106e-07, | |
| "logits/chosen": -1.7343018054962158, | |
| "logits/rejected": -1.5757277011871338, | |
| "logps/chosen": -242.49642944335938, | |
| "logps/rejected": -405.9391784667969, | |
| "loss": 0.0528, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.864837408065796, | |
| "rewards/margins": 9.605855941772461, | |
| "rewards/rejected": -12.470693588256836, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.558933333333333, | |
| "grad_norm": 0.7420125603675842, | |
| "learning_rate": 2.8143983980042174e-07, | |
| "logits/chosen": -1.6621475219726562, | |
| "logits/rejected": -1.5564953088760376, | |
| "logps/chosen": -231.9013214111328, | |
| "logps/rejected": -376.8731689453125, | |
| "loss": 0.0673, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.134626865386963, | |
| "rewards/margins": 8.874670028686523, | |
| "rewards/rejected": -12.009298324584961, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.5696, | |
| "grad_norm": 0.8060418963432312, | |
| "learning_rate": 2.683162634456515e-07, | |
| "logits/chosen": -1.6644542217254639, | |
| "logits/rejected": -1.5664650201797485, | |
| "logps/chosen": -258.3770751953125, | |
| "logps/rejected": -404.95135498046875, | |
| "loss": 0.0543, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.6065478324890137, | |
| "rewards/margins": 9.685090065002441, | |
| "rewards/rejected": -12.29163646697998, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.5802666666666667, | |
| "grad_norm": 0.8783431053161621, | |
| "learning_rate": 2.5548865107314606e-07, | |
| "logits/chosen": -1.6931167840957642, | |
| "logits/rejected": -1.5787171125411987, | |
| "logps/chosen": -248.09957885742188, | |
| "logps/rejected": -387.348388671875, | |
| "loss": 0.0508, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.2612154483795166, | |
| "rewards/margins": 10.170672416687012, | |
| "rewards/rejected": -12.431889533996582, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.590933333333333, | |
| "grad_norm": 2.2933311462402344, | |
| "learning_rate": 2.429587038698511e-07, | |
| "logits/chosen": -1.5951554775238037, | |
| "logits/rejected": -1.5162432193756104, | |
| "logps/chosen": -224.2011260986328, | |
| "logps/rejected": -361.96673583984375, | |
| "loss": 0.0685, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.2784221172332764, | |
| "rewards/margins": 8.822007179260254, | |
| "rewards/rejected": -12.100428581237793, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.6016, | |
| "grad_norm": 0.8401004076004028, | |
| "learning_rate": 2.307280835466144e-07, | |
| "logits/chosen": -1.6584374904632568, | |
| "logits/rejected": -1.5249509811401367, | |
| "logps/chosen": -243.69009399414062, | |
| "logps/rejected": -373.2393798828125, | |
| "loss": 0.068, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.3214409351348877, | |
| "rewards/margins": 9.552881240844727, | |
| "rewards/rejected": -12.874322891235352, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.6122666666666667, | |
| "grad_norm": 2.254423141479492, | |
| "learning_rate": 2.1879841211780856e-07, | |
| "logits/chosen": -1.6165263652801514, | |
| "logits/rejected": -1.5144685506820679, | |
| "logps/chosen": -231.9131317138672, | |
| "logps/rejected": -352.79345703125, | |
| "loss": 0.0671, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.164689302444458, | |
| "rewards/margins": 8.177474975585938, | |
| "rewards/rejected": -11.342164993286133, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.622933333333333, | |
| "grad_norm": 4.380727767944336, | |
| "learning_rate": 2.0717127168622202e-07, | |
| "logits/chosen": -1.6678218841552734, | |
| "logits/rejected": -1.5193488597869873, | |
| "logps/chosen": -248.5584259033203, | |
| "logps/rejected": -370.3858947753906, | |
| "loss": 0.0618, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.2485344409942627, | |
| "rewards/margins": 9.039493560791016, | |
| "rewards/rejected": -12.2880277633667, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.6336, | |
| "grad_norm": 0.5853209495544434, | |
| "learning_rate": 1.958482042332424e-07, | |
| "logits/chosen": -1.6132605075836182, | |
| "logits/rejected": -1.4864953756332397, | |
| "logps/chosen": -249.97372436523438, | |
| "logps/rejected": -382.3087463378906, | |
| "loss": 0.0471, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.8786864280700684, | |
| "rewards/margins": 9.31507396697998, | |
| "rewards/rejected": -12.193760871887207, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.6442666666666668, | |
| "grad_norm": 0.6446689367294312, | |
| "learning_rate": 1.8483071141435937e-07, | |
| "logits/chosen": -1.5566718578338623, | |
| "logits/rejected": -1.482312560081482, | |
| "logps/chosen": -243.97195434570312, | |
| "logps/rejected": -386.7613830566406, | |
| "loss": 0.0738, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.7106494903564453, | |
| "rewards/margins": 9.434020042419434, | |
| "rewards/rejected": -12.144668579101562, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.654933333333333, | |
| "grad_norm": 4.565989017486572, | |
| "learning_rate": 1.741202543600176e-07, | |
| "logits/chosen": -1.6722030639648438, | |
| "logits/rejected": -1.593416452407837, | |
| "logps/chosen": -229.83444213867188, | |
| "logps/rejected": -398.0259704589844, | |
| "loss": 0.0547, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.4450182914733887, | |
| "rewards/margins": 9.658830642700195, | |
| "rewards/rejected": -13.103849411010742, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.6656, | |
| "grad_norm": 0.8811970949172974, | |
| "learning_rate": 1.6371825348184155e-07, | |
| "logits/chosen": -1.6740272045135498, | |
| "logits/rejected": -1.5639140605926514, | |
| "logps/chosen": -237.81863403320312, | |
| "logps/rejected": -390.57489013671875, | |
| "loss": 0.0788, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.0801661014556885, | |
| "rewards/margins": 8.766717910766602, | |
| "rewards/rejected": -11.846883773803711, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.676266666666667, | |
| "grad_norm": 0.733801007270813, | |
| "learning_rate": 1.5362608828426246e-07, | |
| "logits/chosen": -1.6724077463150024, | |
| "logits/rejected": -1.5936537981033325, | |
| "logps/chosen": -244.7788848876953, | |
| "logps/rejected": -397.16827392578125, | |
| "loss": 0.06, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.0939319133758545, | |
| "rewards/margins": 9.266348838806152, | |
| "rewards/rejected": -12.360280990600586, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.686933333333333, | |
| "grad_norm": 1.226430892944336, | |
| "learning_rate": 1.438450971815686e-07, | |
| "logits/chosen": -1.6890594959259033, | |
| "logits/rejected": -1.644351601600647, | |
| "logps/chosen": -225.499267578125, | |
| "logps/rejected": -397.11492919921875, | |
| "loss": 0.0462, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.3902297019958496, | |
| "rewards/margins": 9.87806510925293, | |
| "rewards/rejected": -13.268295288085938, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.6976, | |
| "grad_norm": 1.1510404348373413, | |
| "learning_rate": 1.3437657732040783e-07, | |
| "logits/chosen": -1.6579421758651733, | |
| "logits/rejected": -1.5118043422698975, | |
| "logps/chosen": -248.87814331054688, | |
| "logps/rejected": -364.0889587402344, | |
| "loss": 0.0622, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.0955281257629395, | |
| "rewards/margins": 8.875585556030273, | |
| "rewards/rejected": -11.971113204956055, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.708266666666667, | |
| "grad_norm": 0.9455292820930481, | |
| "learning_rate": 1.2522178440776022e-07, | |
| "logits/chosen": -1.6150785684585571, | |
| "logits/rejected": -1.5976345539093018, | |
| "logps/chosen": -249.5107421875, | |
| "logps/rejected": -389.8973083496094, | |
| "loss": 0.0803, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.392077922821045, | |
| "rewards/margins": 9.01916217803955, | |
| "rewards/rejected": -12.41124153137207, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.718933333333333, | |
| "grad_norm": 1.740357756614685, | |
| "learning_rate": 1.1638193254440655e-07, | |
| "logits/chosen": -1.6622123718261719, | |
| "logits/rejected": -1.534313678741455, | |
| "logps/chosen": -238.5977325439453, | |
| "logps/rejected": -380.42108154296875, | |
| "loss": 0.0555, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.4845123291015625, | |
| "rewards/margins": 9.121153831481934, | |
| "rewards/rejected": -12.60566520690918, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.7296, | |
| "grad_norm": 1.714776635169983, | |
| "learning_rate": 1.0785819406391562e-07, | |
| "logits/chosen": -1.5962209701538086, | |
| "logits/rejected": -1.5493957996368408, | |
| "logps/chosen": -271.0741271972656, | |
| "logps/rejected": -369.3987731933594, | |
| "loss": 0.042, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.6246178150177, | |
| "rewards/margins": 9.22441577911377, | |
| "rewards/rejected": -11.849031448364258, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.740266666666667, | |
| "grad_norm": 2.9277751445770264, | |
| "learning_rate": 9.965169937717095e-08, | |
| "logits/chosen": -1.64768385887146, | |
| "logits/rejected": -1.5789892673492432, | |
| "logps/chosen": -234.07608032226562, | |
| "logps/rejected": -388.78961181640625, | |
| "loss": 0.0615, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.3321361541748047, | |
| "rewards/margins": 9.532690048217773, | |
| "rewards/rejected": -12.864827156066895, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.7509333333333332, | |
| "grad_norm": 4.05463981628418, | |
| "learning_rate": 9.176353682245676e-08, | |
| "logits/chosen": -1.7111181020736694, | |
| "logits/rejected": -1.5939412117004395, | |
| "logps/chosen": -242.18759155273438, | |
| "logps/rejected": -376.65948486328125, | |
| "loss": 0.0931, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.3087191581726074, | |
| "rewards/margins": 9.18929672241211, | |
| "rewards/rejected": -12.498015403747559, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.7616, | |
| "grad_norm": 4.355930805206299, | |
| "learning_rate": 8.41947525211223e-08, | |
| "logits/chosen": -1.528749704360962, | |
| "logits/rejected": -1.5081889629364014, | |
| "logps/chosen": -232.96383666992188, | |
| "logps/rejected": -353.97906494140625, | |
| "loss": 0.1119, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.941885471343994, | |
| "rewards/margins": 8.040489196777344, | |
| "rewards/rejected": -10.98237419128418, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.772266666666667, | |
| "grad_norm": 1.2377660274505615, | |
| "learning_rate": 7.694635023884789e-08, | |
| "logits/chosen": -1.596066951751709, | |
| "logits/rejected": -1.5349980592727661, | |
| "logps/chosen": -244.0199737548828, | |
| "logps/rejected": -365.12384033203125, | |
| "loss": 0.0548, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.9959444999694824, | |
| "rewards/margins": 8.654120445251465, | |
| "rewards/rejected": -11.650065422058105, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.7829333333333333, | |
| "grad_norm": 4.812592506408691, | |
| "learning_rate": 7.00192912525241e-08, | |
| "logits/chosen": -1.6420648097991943, | |
| "logits/rejected": -1.5604394674301147, | |
| "logps/chosen": -229.23388671875, | |
| "logps/rejected": -394.1202697753906, | |
| "loss": 0.0746, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -3.189704418182373, | |
| "rewards/margins": 9.603094100952148, | |
| "rewards/rejected": -12.79279899597168, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.7936, | |
| "grad_norm": 2.0514466762542725, | |
| "learning_rate": 6.341449422277047e-08, | |
| "logits/chosen": -1.5872209072113037, | |
| "logits/rejected": -1.4994614124298096, | |
| "logps/chosen": -246.319091796875, | |
| "logps/rejected": -378.219970703125, | |
| "loss": 0.0499, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.6841087341308594, | |
| "rewards/margins": 10.14326286315918, | |
| "rewards/rejected": -12.827372550964355, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.804266666666667, | |
| "grad_norm": 3.7857437133789062, | |
| "learning_rate": 5.7132835072101486e-08, | |
| "logits/chosen": -1.6783745288848877, | |
| "logits/rejected": -1.5917742252349854, | |
| "logps/chosen": -280.02191162109375, | |
| "logps/rejected": -345.45831298828125, | |
| "loss": 0.0696, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.7913777828216553, | |
| "rewards/margins": 8.280149459838867, | |
| "rewards/rejected": -11.071527481079102, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.8149333333333333, | |
| "grad_norm": 2.1288628578186035, | |
| "learning_rate": 5.117514686876379e-08, | |
| "logits/chosen": -1.5824289321899414, | |
| "logits/rejected": -1.4594101905822754, | |
| "logps/chosen": -239.4272003173828, | |
| "logps/rejected": -340.62200927734375, | |
| "loss": 0.0832, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.9561493396759033, | |
| "rewards/margins": 7.562721252441406, | |
| "rewards/rejected": -10.51887035369873, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.8256, | |
| "grad_norm": 5.216054439544678, | |
| "learning_rate": 4.5542219716255274e-08, | |
| "logits/chosen": -1.6920030117034912, | |
| "logits/rejected": -1.5826241970062256, | |
| "logps/chosen": -247.36904907226562, | |
| "logps/rejected": -388.71051025390625, | |
| "loss": 0.0521, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.8860912322998047, | |
| "rewards/margins": 9.873689651489258, | |
| "rewards/rejected": -12.75977897644043, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.836266666666667, | |
| "grad_norm": 1.0527479648590088, | |
| "learning_rate": 4.023480064854174e-08, | |
| "logits/chosen": -1.712881088256836, | |
| "logits/rejected": -1.6000264883041382, | |
| "logps/chosen": -236.73312377929688, | |
| "logps/rejected": -387.08599853515625, | |
| "loss": 0.0456, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.8067479133605957, | |
| "rewards/margins": 10.028037071228027, | |
| "rewards/rejected": -12.834785461425781, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.8469333333333333, | |
| "grad_norm": 3.546607494354248, | |
| "learning_rate": 3.525359353098584e-08, | |
| "logits/chosen": -1.6619724035263062, | |
| "logits/rejected": -1.5283750295639038, | |
| "logps/chosen": -236.0811309814453, | |
| "logps/rejected": -376.9278259277344, | |
| "loss": 0.0546, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.1373705863952637, | |
| "rewards/margins": 9.576279640197754, | |
| "rewards/rejected": -12.713650703430176, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.8576, | |
| "grad_norm": 1.9862865209579468, | |
| "learning_rate": 3.059925896700094e-08, | |
| "logits/chosen": -1.5844295024871826, | |
| "logits/rejected": -1.4807649850845337, | |
| "logps/chosen": -267.7585144042969, | |
| "logps/rejected": -370.53790283203125, | |
| "loss": 0.0592, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.931180477142334, | |
| "rewards/margins": 8.453359603881836, | |
| "rewards/rejected": -11.384538650512695, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.8682666666666665, | |
| "grad_norm": 3.8812670707702637, | |
| "learning_rate": 2.6272414210443964e-08, | |
| "logits/chosen": -1.6309881210327148, | |
| "logits/rejected": -1.5491442680358887, | |
| "logps/chosen": -225.6214599609375, | |
| "logps/rejected": -383.81243896484375, | |
| "loss": 0.0824, | |
| "rewards/accuracies": 0.9750000238418579, | |
| "rewards/chosen": -2.9007303714752197, | |
| "rewards/margins": 9.299989700317383, | |
| "rewards/rejected": -12.200719833374023, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 2.8789333333333333, | |
| "grad_norm": 4.220300197601318, | |
| "learning_rate": 2.2273633083753632e-08, | |
| "logits/chosen": -1.6484569311141968, | |
| "logits/rejected": -1.5259983539581299, | |
| "logps/chosen": -221.6566619873047, | |
| "logps/rejected": -352.6129455566406, | |
| "loss": 0.0606, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.176244020462036, | |
| "rewards/margins": 8.576908111572266, | |
| "rewards/rejected": -11.753152847290039, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 2.8895999999999997, | |
| "grad_norm": 1.0096988677978516, | |
| "learning_rate": 1.8603445901851647e-08, | |
| "logits/chosen": -1.6377588510513306, | |
| "logits/rejected": -1.559996485710144, | |
| "logps/chosen": -252.84579467773438, | |
| "logps/rejected": -386.17755126953125, | |
| "loss": 0.0955, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.2294230461120605, | |
| "rewards/margins": 8.582195281982422, | |
| "rewards/rejected": -11.811617851257324, | |
| "step": 1355 | |
| }, | |
| { | |
| "epoch": 2.9002666666666665, | |
| "grad_norm": 1.8424628973007202, | |
| "learning_rate": 1.5262339401813375e-08, | |
| "logits/chosen": -1.7368948459625244, | |
| "logits/rejected": -1.6223742961883545, | |
| "logps/chosen": -211.1990203857422, | |
| "logps/rejected": -368.0191650390625, | |
| "loss": 0.0704, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.6657638549804688, | |
| "rewards/margins": 9.301887512207031, | |
| "rewards/rejected": -11.9676513671875, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 2.9109333333333334, | |
| "grad_norm": 3.6020452976226807, | |
| "learning_rate": 1.2250756678315601e-08, | |
| "logits/chosen": -1.6663650274276733, | |
| "logits/rejected": -1.57073974609375, | |
| "logps/chosen": -233.00131225585938, | |
| "logps/rejected": -400.0315856933594, | |
| "loss": 0.0579, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.8992886543273926, | |
| "rewards/margins": 10.07255744934082, | |
| "rewards/rejected": -12.971845626831055, | |
| "step": 1365 | |
| }, | |
| { | |
| "epoch": 2.9215999999999998, | |
| "grad_norm": 1.2735259532928467, | |
| "learning_rate": 9.569097124875193e-09, | |
| "logits/chosen": -1.6023895740509033, | |
| "logits/rejected": -1.4932624101638794, | |
| "logps/chosen": -224.2072296142578, | |
| "logps/rejected": -374.86883544921875, | |
| "loss": 0.0506, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.029714584350586, | |
| "rewards/margins": 9.75147533416748, | |
| "rewards/rejected": -12.781190872192383, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 2.9322666666666666, | |
| "grad_norm": 3.791727066040039, | |
| "learning_rate": 7.217716380881479e-09, | |
| "logits/chosen": -1.6325794458389282, | |
| "logits/rejected": -1.6049026250839233, | |
| "logps/chosen": -258.8813171386719, | |
| "logps/rejected": -384.8707580566406, | |
| "loss": 0.0517, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -3.0506906509399414, | |
| "rewards/margins": 9.173063278198242, | |
| "rewards/rejected": -12.223753929138184, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 2.9429333333333334, | |
| "grad_norm": 4.857158660888672, | |
| "learning_rate": 5.196926284430359e-09, | |
| "logits/chosen": -1.6092033386230469, | |
| "logits/rejected": -1.5137813091278076, | |
| "logps/chosen": -244.2127685546875, | |
| "logps/rejected": -382.6106872558594, | |
| "loss": 0.0808, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -3.1281540393829346, | |
| "rewards/margins": 9.14082145690918, | |
| "rewards/rejected": -12.268976211547852, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 2.9536, | |
| "grad_norm": 3.6009581089019775, | |
| "learning_rate": 3.5069948309712775e-09, | |
| "logits/chosen": -1.5844300985336304, | |
| "logits/rejected": -1.5796252489089966, | |
| "logps/chosen": -220.5863800048828, | |
| "logps/rejected": -371.11956787109375, | |
| "loss": 0.0649, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.970752716064453, | |
| "rewards/margins": 8.276488304138184, | |
| "rewards/rejected": -11.247241020202637, | |
| "step": 1385 | |
| }, | |
| { | |
| "epoch": 2.9642666666666666, | |
| "grad_norm": 3.476202964782715, | |
| "learning_rate": 2.1481461377634294e-09, | |
| "logits/chosen": -1.631340742111206, | |
| "logits/rejected": -1.5940053462982178, | |
| "logps/chosen": -231.7400360107422, | |
| "logps/rejected": -374.5093078613281, | |
| "loss": 0.0465, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -3.1524415016174316, | |
| "rewards/margins": 9.017885208129883, | |
| "rewards/rejected": -12.170326232910156, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 2.9749333333333334, | |
| "grad_norm": 1.2247248888015747, | |
| "learning_rate": 1.120560414154259e-09, | |
| "logits/chosen": -1.642568588256836, | |
| "logits/rejected": -1.479210615158081, | |
| "logps/chosen": -236.7920379638672, | |
| "logps/rejected": -384.9251708984375, | |
| "loss": 0.0431, | |
| "rewards/accuracies": 0.9937499761581421, | |
| "rewards/chosen": -2.9622087478637695, | |
| "rewards/margins": 9.579465866088867, | |
| "rewards/rejected": -12.541674613952637, | |
| "step": 1395 | |
| }, | |
| { | |
| "epoch": 2.9856, | |
| "grad_norm": 1.0781242847442627, | |
| "learning_rate": 4.2437393768079983e-10, | |
| "logits/chosen": -1.670657753944397, | |
| "logits/rejected": -1.5803184509277344, | |
| "logps/chosen": -252.31094360351562, | |
| "logps/rejected": -374.4371032714844, | |
| "loss": 0.0526, | |
| "rewards/accuracies": 0.981249988079071, | |
| "rewards/chosen": -2.9776816368103027, | |
| "rewards/margins": 8.582575798034668, | |
| "rewards/rejected": -11.560258865356445, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 2.9962666666666666, | |
| "grad_norm": 1.7229729890823364, | |
| "learning_rate": 5.967903599635171e-11, | |
| "logits/chosen": -1.6550830602645874, | |
| "logits/rejected": -1.4953168630599976, | |
| "logps/chosen": -239.16610717773438, | |
| "logps/rejected": -392.11444091796875, | |
| "loss": 0.0611, | |
| "rewards/accuracies": 0.987500011920929, | |
| "rewards/chosen": -2.644193172454834, | |
| "rewards/margins": 9.434123039245605, | |
| "rewards/rejected": -12.078317642211914, | |
| "step": 1405 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 1407, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |