| { |
| "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": 3.2248189449310303, |
| "learning_rate": 4.651162790697675e-07, |
| "logits/chosen": -1.417102575302124, |
| "logits/rejected": -1.3289285898208618, |
| "logps/chosen": -223.09201049804688, |
| "logps/rejected": -281.4899597167969, |
| "loss": 0.6927, |
| "rewards/accuracies": 0.33125001192092896, |
| "rewards/chosen": 8.828159479890019e-05, |
| "rewards/margins": 0.0009968661470338702, |
| "rewards/rejected": -0.0009085845085792243, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.021333333333333333, |
| "grad_norm": 3.0620968341827393, |
| "learning_rate": 1.0465116279069768e-06, |
| "logits/chosen": -1.4176433086395264, |
| "logits/rejected": -1.4581596851348877, |
| "logps/chosen": -227.9161834716797, |
| "logps/rejected": -281.8757629394531, |
| "loss": 0.6906, |
| "rewards/accuracies": 0.643750011920929, |
| "rewards/chosen": -0.0013590670423582196, |
| "rewards/margins": 0.005104787182062864, |
| "rewards/rejected": -0.006463854573667049, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.032, |
| "grad_norm": 2.624357223510742, |
| "learning_rate": 1.6279069767441862e-06, |
| "logits/chosen": -1.3286405801773071, |
| "logits/rejected": -1.2039461135864258, |
| "logps/chosen": -220.56112670898438, |
| "logps/rejected": -255.6830596923828, |
| "loss": 0.6861, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": -0.004264736548066139, |
| "rewards/margins": 0.014199944213032722, |
| "rewards/rejected": -0.01846468076109886, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.042666666666666665, |
| "grad_norm": 2.8290464878082275, |
| "learning_rate": 2.2093023255813954e-06, |
| "logits/chosen": -1.3126474618911743, |
| "logits/rejected": -1.2001457214355469, |
| "logps/chosen": -231.92227172851562, |
| "logps/rejected": -240.58273315429688, |
| "loss": 0.6805, |
| "rewards/accuracies": 0.731249988079071, |
| "rewards/chosen": -0.007818732410669327, |
| "rewards/margins": 0.02588127925992012, |
| "rewards/rejected": -0.03370000794529915, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.05333333333333334, |
| "grad_norm": 2.9549717903137207, |
| "learning_rate": 2.790697674418605e-06, |
| "logits/chosen": -1.4152873754501343, |
| "logits/rejected": -1.34293532371521, |
| "logps/chosen": -228.6177215576172, |
| "logps/rejected": -274.3415222167969, |
| "loss": 0.67, |
| "rewards/accuracies": 0.762499988079071, |
| "rewards/chosen": -0.017063328996300697, |
| "rewards/margins": 0.04786022752523422, |
| "rewards/rejected": -0.06492356210947037, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.064, |
| "grad_norm": 3.4597318172454834, |
| "learning_rate": 3.372093023255814e-06, |
| "logits/chosen": -1.397507667541504, |
| "logits/rejected": -1.3085404634475708, |
| "logps/chosen": -207.57083129882812, |
| "logps/rejected": -246.46914672851562, |
| "loss": 0.6596, |
| "rewards/accuracies": 0.7749999761581421, |
| "rewards/chosen": -0.028314024209976196, |
| "rewards/margins": 0.0706639513373375, |
| "rewards/rejected": -0.09897797554731369, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.07466666666666667, |
| "grad_norm": 2.903716802597046, |
| "learning_rate": 3.953488372093024e-06, |
| "logits/chosen": -1.4894464015960693, |
| "logits/rejected": -1.3991295099258423, |
| "logps/chosen": -218.21859741210938, |
| "logps/rejected": -259.07818603515625, |
| "loss": 0.6396, |
| "rewards/accuracies": 0.7875000238418579, |
| "rewards/chosen": -0.03908306732773781, |
| "rewards/margins": 0.11545582115650177, |
| "rewards/rejected": -0.15453888475894928, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.08533333333333333, |
| "grad_norm": 2.7245819568634033, |
| "learning_rate": 4.5348837209302326e-06, |
| "logits/chosen": -1.4166643619537354, |
| "logits/rejected": -1.2851589918136597, |
| "logps/chosen": -237.89370727539062, |
| "logps/rejected": -270.0038757324219, |
| "loss": 0.6095, |
| "rewards/accuracies": 0.831250011920929, |
| "rewards/chosen": -0.08339570462703705, |
| "rewards/margins": 0.18969933688640594, |
| "rewards/rejected": -0.273095041513443, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.096, |
| "grad_norm": 3.0906941890716553, |
| "learning_rate": 4.99999336897255e-06, |
| "logits/chosen": -1.4081467390060425, |
| "logits/rejected": -1.3213756084442139, |
| "logps/chosen": -245.74490356445312, |
| "logps/rejected": -272.46575927734375, |
| "loss": 0.6033, |
| "rewards/accuracies": 0.7875000238418579, |
| "rewards/chosen": -0.1594085991382599, |
| "rewards/margins": 0.21051104366779327, |
| "rewards/rejected": -0.36991962790489197, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.10666666666666667, |
| "grad_norm": 2.9989840984344482, |
| "learning_rate": 4.999761286705285e-06, |
| "logits/chosen": -1.425335168838501, |
| "logits/rejected": -1.4387575387954712, |
| "logps/chosen": -219.0963592529297, |
| "logps/rejected": -262.6189270019531, |
| "loss": 0.5662, |
| "rewards/accuracies": 0.8187500238418579, |
| "rewards/chosen": -0.1677021086215973, |
| "rewards/margins": 0.3115597367286682, |
| "rewards/rejected": -0.4792618155479431, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.11733333333333333, |
| "grad_norm": 3.8567123413085938, |
| "learning_rate": 4.999197688241076e-06, |
| "logits/chosen": -1.4213941097259521, |
| "logits/rejected": -1.2188364267349243, |
| "logps/chosen": -282.511962890625, |
| "logps/rejected": -260.6012268066406, |
| "loss": 0.5587, |
| "rewards/accuracies": 0.8062499761581421, |
| "rewards/chosen": -0.2968081831932068, |
| "rewards/margins": 0.33465075492858887, |
| "rewards/rejected": -0.6314589381217957, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.128, |
| "grad_norm": 2.7673559188842773, |
| "learning_rate": 4.9983026483238685e-06, |
| "logits/chosen": -1.3809436559677124, |
| "logits/rejected": -1.3605387210845947, |
| "logps/chosen": -227.60287475585938, |
| "logps/rejected": -268.99884033203125, |
| "loss": 0.5575, |
| "rewards/accuracies": 0.7875000238418579, |
| "rewards/chosen": -0.32013896107673645, |
| "rewards/margins": 0.38853272795677185, |
| "rewards/rejected": -0.7086717486381531, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.13866666666666666, |
| "grad_norm": 2.7526016235351562, |
| "learning_rate": 4.997076285653087e-06, |
| "logits/chosen": -1.500636339187622, |
| "logits/rejected": -1.3579481840133667, |
| "logps/chosen": -251.8033447265625, |
| "logps/rejected": -282.87042236328125, |
| "loss": 0.4822, |
| "rewards/accuracies": 0.831250011920929, |
| "rewards/chosen": -0.3049132227897644, |
| "rewards/margins": 0.5729039907455444, |
| "rewards/rejected": -0.8778173327445984, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.14933333333333335, |
| "grad_norm": 2.671469211578369, |
| "learning_rate": 4.995518762867897e-06, |
| "logits/chosen": -1.466827630996704, |
| "logits/rejected": -1.3935381174087524, |
| "logps/chosen": -193.24647521972656, |
| "logps/rejected": -263.76239013671875, |
| "loss": 0.4588, |
| "rewards/accuracies": 0.862500011920929, |
| "rewards/chosen": -0.3088843524456024, |
| "rewards/margins": 0.6813338994979858, |
| "rewards/rejected": -0.9902183413505554, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.16, |
| "grad_norm": 3.234593629837036, |
| "learning_rate": 4.993630286525634e-06, |
| "logits/chosen": -1.5930967330932617, |
| "logits/rejected": -1.4225651025772095, |
| "logps/chosen": -220.28488159179688, |
| "logps/rejected": -262.43218994140625, |
| "loss": 0.4809, |
| "rewards/accuracies": 0.78125, |
| "rewards/chosen": -0.4655492901802063, |
| "rewards/margins": 0.7154077291488647, |
| "rewards/rejected": -1.1809570789337158, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.17066666666666666, |
| "grad_norm": 3.725409746170044, |
| "learning_rate": 4.99141110707441e-06, |
| "logits/chosen": -1.5412805080413818, |
| "logits/rejected": -1.3626563549041748, |
| "logps/chosen": -234.2522735595703, |
| "logps/rejected": -259.58172607421875, |
| "loss": 0.4561, |
| "rewards/accuracies": 0.8062499761581421, |
| "rewards/chosen": -0.5094849467277527, |
| "rewards/margins": 0.8051780462265015, |
| "rewards/rejected": -1.3146629333496094, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.18133333333333335, |
| "grad_norm": 3.086479902267456, |
| "learning_rate": 4.9888615188199e-06, |
| "logits/chosen": -1.6061999797821045, |
| "logits/rejected": -1.5315011739730835, |
| "logps/chosen": -214.01416015625, |
| "logps/rejected": -277.1407165527344, |
| "loss": 0.3968, |
| "rewards/accuracies": 0.8062499761581421, |
| "rewards/chosen": -0.3647157847881317, |
| "rewards/margins": 1.1234705448150635, |
| "rewards/rejected": -1.488186240196228, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.192, |
| "grad_norm": 2.2907378673553467, |
| "learning_rate": 4.9859818598863104e-06, |
| "logits/chosen": -1.617488145828247, |
| "logits/rejected": -1.4285557270050049, |
| "logps/chosen": -215.92794799804688, |
| "logps/rejected": -293.9332580566406, |
| "loss": 0.3555, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": -0.4863402247428894, |
| "rewards/margins": 1.2553744316101074, |
| "rewards/rejected": -1.7417147159576416, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.20266666666666666, |
| "grad_norm": 2.879300117492676, |
| "learning_rate": 4.982772512171538e-06, |
| "logits/chosen": -1.4808334112167358, |
| "logits/rejected": -1.4129270315170288, |
| "logps/chosen": -224.61282348632812, |
| "logps/rejected": -287.599609375, |
| "loss": 0.3854, |
| "rewards/accuracies": 0.8374999761581421, |
| "rewards/chosen": -0.5822515487670898, |
| "rewards/margins": 1.2435760498046875, |
| "rewards/rejected": -1.8258273601531982, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.21333333333333335, |
| "grad_norm": 2.7722413539886475, |
| "learning_rate": 4.979233901296523e-06, |
| "logits/chosen": -1.5995476245880127, |
| "logits/rejected": -1.5469409227371216, |
| "logps/chosen": -218.4732666015625, |
| "logps/rejected": -274.4031982421875, |
| "loss": 0.3723, |
| "rewards/accuracies": 0.8687499761581421, |
| "rewards/chosen": -0.623435914516449, |
| "rewards/margins": 1.2485257387161255, |
| "rewards/rejected": -1.8719615936279297, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.224, |
| "grad_norm": 2.1027016639709473, |
| "learning_rate": 4.975366496548804e-06, |
| "logits/chosen": -1.5575580596923828, |
| "logits/rejected": -1.418176293373108, |
| "logps/chosen": -238.68490600585938, |
| "logps/rejected": -300.532470703125, |
| "loss": 0.3361, |
| "rewards/accuracies": 0.8812500238418579, |
| "rewards/chosen": -0.5908123254776001, |
| "rewards/margins": 1.523303747177124, |
| "rewards/rejected": -2.1141159534454346, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.23466666666666666, |
| "grad_norm": 3.443408727645874, |
| "learning_rate": 4.971170810820279e-06, |
| "logits/chosen": -1.5500075817108154, |
| "logits/rejected": -1.5055526494979858, |
| "logps/chosen": -234.66940307617188, |
| "logps/rejected": -267.90087890625, |
| "loss": 0.4005, |
| "rewards/accuracies": 0.824999988079071, |
| "rewards/chosen": -0.7276199460029602, |
| "rewards/margins": 1.2876373529434204, |
| "rewards/rejected": -2.0152573585510254, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.24533333333333332, |
| "grad_norm": 3.1174283027648926, |
| "learning_rate": 4.96664740053919e-06, |
| "logits/chosen": -1.5923100709915161, |
| "logits/rejected": -1.4889472723007202, |
| "logps/chosen": -225.01779174804688, |
| "logps/rejected": -288.71173095703125, |
| "loss": 0.3272, |
| "rewards/accuracies": 0.8500000238418579, |
| "rewards/chosen": -0.6660363078117371, |
| "rewards/margins": 1.6434768438339233, |
| "rewards/rejected": -2.3095130920410156, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.256, |
| "grad_norm": 3.5097219944000244, |
| "learning_rate": 4.961796865596328e-06, |
| "logits/chosen": -1.5631916522979736, |
| "logits/rejected": -1.5810332298278809, |
| "logps/chosen": -216.34115600585938, |
| "logps/rejected": -309.64642333984375, |
| "loss": 0.2956, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -0.607854962348938, |
| "rewards/margins": 1.7155084609985352, |
| "rewards/rejected": -2.3233633041381836, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.26666666666666666, |
| "grad_norm": 3.2788164615631104, |
| "learning_rate": 4.956619849265473e-06, |
| "logits/chosen": -1.5768873691558838, |
| "logits/rejected": -1.4464285373687744, |
| "logps/chosen": -227.64956665039062, |
| "logps/rejected": -267.4158935546875, |
| "loss": 0.3828, |
| "rewards/accuracies": 0.8500000238418579, |
| "rewards/chosen": -0.6830345392227173, |
| "rewards/margins": 1.5734184980392456, |
| "rewards/rejected": -2.256453037261963, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.2773333333333333, |
| "grad_norm": 2.1826467514038086, |
| "learning_rate": 4.951117038118091e-06, |
| "logits/chosen": -1.6085306406021118, |
| "logits/rejected": -1.5910682678222656, |
| "logps/chosen": -232.57266235351562, |
| "logps/rejected": -304.3570556640625, |
| "loss": 0.294, |
| "rewards/accuracies": 0.8812500238418579, |
| "rewards/chosen": -0.5465613603591919, |
| "rewards/margins": 1.9157283306121826, |
| "rewards/rejected": -2.462289571762085, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.288, |
| "grad_norm": 2.313305377960205, |
| "learning_rate": 4.945289161932273e-06, |
| "logits/chosen": -1.6053348779678345, |
| "logits/rejected": -1.5949219465255737, |
| "logps/chosen": -215.5905303955078, |
| "logps/rejected": -289.11212158203125, |
| "loss": 0.2507, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.41833844780921936, |
| "rewards/margins": 2.0826306343078613, |
| "rewards/rejected": -2.500969409942627, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.2986666666666667, |
| "grad_norm": 3.754728078842163, |
| "learning_rate": 4.93913699359596e-06, |
| "logits/chosen": -1.5684372186660767, |
| "logits/rejected": -1.4641824960708618, |
| "logps/chosen": -210.9490203857422, |
| "logps/rejected": -281.46337890625, |
| "loss": 0.3506, |
| "rewards/accuracies": 0.831250011920929, |
| "rewards/chosen": -0.6244282722473145, |
| "rewards/margins": 1.8182014226913452, |
| "rewards/rejected": -2.442629814147949, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.30933333333333335, |
| "grad_norm": 4.167830467224121, |
| "learning_rate": 4.9326613490044364e-06, |
| "logits/chosen": -1.5899982452392578, |
| "logits/rejected": -1.486648678779602, |
| "logps/chosen": -225.3086700439453, |
| "logps/rejected": -274.19500732421875, |
| "loss": 0.3288, |
| "rewards/accuracies": 0.8500000238418579, |
| "rewards/chosen": -0.6446894407272339, |
| "rewards/margins": 1.9515396356582642, |
| "rewards/rejected": -2.596228837966919, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.32, |
| "grad_norm": 4.168672561645508, |
| "learning_rate": 4.925863086952132e-06, |
| "logits/chosen": -1.5442506074905396, |
| "logits/rejected": -1.4385302066802979, |
| "logps/chosen": -218.75851440429688, |
| "logps/rejected": -283.3717041015625, |
| "loss": 0.3998, |
| "rewards/accuracies": 0.793749988079071, |
| "rewards/chosen": -0.570462703704834, |
| "rewards/margins": 1.811073899269104, |
| "rewards/rejected": -2.3815362453460693, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.33066666666666666, |
| "grad_norm": 2.5970113277435303, |
| "learning_rate": 4.918743109018727e-06, |
| "logits/chosen": -1.6545788049697876, |
| "logits/rejected": -1.5757300853729248, |
| "logps/chosen": -226.00436401367188, |
| "logps/rejected": -298.34307861328125, |
| "loss": 0.3149, |
| "rewards/accuracies": 0.8812500238418579, |
| "rewards/chosen": -0.5687687397003174, |
| "rewards/margins": 2.034900426864624, |
| "rewards/rejected": -2.6036694049835205, |
| "step": 155 |
| }, |
| { |
| "epoch": 0.3413333333333333, |
| "grad_norm": 2.196605682373047, |
| "learning_rate": 4.911302359449585e-06, |
| "logits/chosen": -1.725408911705017, |
| "logits/rejected": -1.5622869729995728, |
| "logps/chosen": -224.3115692138672, |
| "logps/rejected": -299.6641540527344, |
| "loss": 0.2456, |
| "rewards/accuracies": 0.918749988079071, |
| "rewards/chosen": -0.4611068367958069, |
| "rewards/margins": 2.2169716358184814, |
| "rewards/rejected": -2.6780784130096436, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.352, |
| "grad_norm": 2.2756359577178955, |
| "learning_rate": 4.9035418250305314e-06, |
| "logits/chosen": -1.5784146785736084, |
| "logits/rejected": -1.459627628326416, |
| "logps/chosen": -237.4366912841797, |
| "logps/rejected": -282.8692321777344, |
| "loss": 0.3194, |
| "rewards/accuracies": 0.8812500238418579, |
| "rewards/chosen": -0.42128247022628784, |
| "rewards/margins": 1.9330406188964844, |
| "rewards/rejected": -2.354323148727417, |
| "step": 165 |
| }, |
| { |
| "epoch": 0.3626666666666667, |
| "grad_norm": 2.7691073417663574, |
| "learning_rate": 4.895462534956981e-06, |
| "logits/chosen": -1.6254955530166626, |
| "logits/rejected": -1.5783421993255615, |
| "logps/chosen": -210.7351837158203, |
| "logps/rejected": -305.00836181640625, |
| "loss": 0.2511, |
| "rewards/accuracies": 0.925000011920929, |
| "rewards/chosen": -0.3136264681816101, |
| "rewards/margins": 2.3627917766571045, |
| "rewards/rejected": -2.6764187812805176, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.37333333333333335, |
| "grad_norm": 2.728076934814453, |
| "learning_rate": 4.887065560697449e-06, |
| "logits/chosen": -1.6129728555679321, |
| "logits/rejected": -1.4741400480270386, |
| "logps/chosen": -242.5968475341797, |
| "logps/rejected": -298.1058654785156, |
| "loss": 0.2798, |
| "rewards/accuracies": 0.8999999761581421, |
| "rewards/chosen": -0.3989323079586029, |
| "rewards/margins": 2.153937816619873, |
| "rewards/rejected": -2.552870035171509, |
| "step": 175 |
| }, |
| { |
| "epoch": 0.384, |
| "grad_norm": 3.1587653160095215, |
| "learning_rate": 4.878352015851459e-06, |
| "logits/chosen": -1.6577237844467163, |
| "logits/rejected": -1.5179884433746338, |
| "logps/chosen": -258.1352233886719, |
| "logps/rejected": -287.0603942871094, |
| "loss": 0.3108, |
| "rewards/accuracies": 0.8687499761581421, |
| "rewards/chosen": -0.6214097142219543, |
| "rewards/margins": 2.040191650390625, |
| "rewards/rejected": -2.6616015434265137, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.39466666666666667, |
| "grad_norm": 3.1338894367218018, |
| "learning_rate": 4.869323056001852e-06, |
| "logits/chosen": -1.5607925653457642, |
| "logits/rejected": -1.4988266229629517, |
| "logps/chosen": -211.45278930664062, |
| "logps/rejected": -289.1008605957031, |
| "loss": 0.2638, |
| "rewards/accuracies": 0.887499988079071, |
| "rewards/chosen": -0.3999144434928894, |
| "rewards/margins": 2.4039969444274902, |
| "rewards/rejected": -2.8039114475250244, |
| "step": 185 |
| }, |
| { |
| "epoch": 0.4053333333333333, |
| "grad_norm": 2.9660542011260986, |
| "learning_rate": 4.859979878561536e-06, |
| "logits/chosen": -1.680594801902771, |
| "logits/rejected": -1.4900007247924805, |
| "logps/chosen": -219.1302947998047, |
| "logps/rejected": -285.2054443359375, |
| "loss": 0.2477, |
| "rewards/accuracies": 0.9312499761581421, |
| "rewards/chosen": -0.34523534774780273, |
| "rewards/margins": 2.4674582481384277, |
| "rewards/rejected": -2.8126938343048096, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.416, |
| "grad_norm": 3.570490598678589, |
| "learning_rate": 4.850323722614685e-06, |
| "logits/chosen": -1.6982581615447998, |
| "logits/rejected": -1.5957467555999756, |
| "logps/chosen": -224.43197631835938, |
| "logps/rejected": -306.5550842285156, |
| "loss": 0.2465, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.5012459754943848, |
| "rewards/margins": 2.4313809871673584, |
| "rewards/rejected": -2.932626724243164, |
| "step": 195 |
| }, |
| { |
| "epoch": 0.4266666666666667, |
| "grad_norm": 2.77095627784729, |
| "learning_rate": 4.840355868752416e-06, |
| "logits/chosen": -1.5608937740325928, |
| "logits/rejected": -1.448979377746582, |
| "logps/chosen": -285.95556640625, |
| "logps/rejected": -297.87982177734375, |
| "loss": 0.2754, |
| "rewards/accuracies": 0.8812500238418579, |
| "rewards/chosen": -0.3715810179710388, |
| "rewards/margins": 2.369892120361328, |
| "rewards/rejected": -2.7414729595184326, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.43733333333333335, |
| "grad_norm": 2.7541775703430176, |
| "learning_rate": 4.830077638902955e-06, |
| "logits/chosen": -1.6883554458618164, |
| "logits/rejected": -1.579251766204834, |
| "logps/chosen": -193.3550567626953, |
| "logps/rejected": -294.20953369140625, |
| "loss": 0.2067, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.4765893518924713, |
| "rewards/margins": 2.9404797554016113, |
| "rewards/rejected": -3.4170689582824707, |
| "step": 205 |
| }, |
| { |
| "epoch": 0.448, |
| "grad_norm": 3.730454206466675, |
| "learning_rate": 4.8194903961563255e-06, |
| "logits/chosen": -1.7941776514053345, |
| "logits/rejected": -1.643109679222107, |
| "logps/chosen": -230.3562469482422, |
| "logps/rejected": -287.34381103515625, |
| "loss": 0.2412, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": -0.6958181858062744, |
| "rewards/margins": 2.7433438301086426, |
| "rewards/rejected": -3.439162015914917, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.45866666666666667, |
| "grad_norm": 2.533442497253418, |
| "learning_rate": 4.808595544583578e-06, |
| "logits/chosen": -1.6759326457977295, |
| "logits/rejected": -1.5923807621002197, |
| "logps/chosen": -238.96255493164062, |
| "logps/rejected": -312.38726806640625, |
| "loss": 0.2548, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.8674060702323914, |
| "rewards/margins": 2.7441043853759766, |
| "rewards/rejected": -3.6115105152130127, |
| "step": 215 |
| }, |
| { |
| "epoch": 0.4693333333333333, |
| "grad_norm": 2.9548909664154053, |
| "learning_rate": 4.797394529050577e-06, |
| "logits/chosen": -1.6703212261199951, |
| "logits/rejected": -1.5990310907363892, |
| "logps/chosen": -249.1092071533203, |
| "logps/rejected": -305.9916687011719, |
| "loss": 0.2558, |
| "rewards/accuracies": 0.887499988079071, |
| "rewards/chosen": -0.7586711645126343, |
| "rewards/margins": 2.6208558082580566, |
| "rewards/rejected": -3.3795273303985596, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.48, |
| "grad_norm": 3.1150755882263184, |
| "learning_rate": 4.785888835026393e-06, |
| "logits/chosen": -1.6585559844970703, |
| "logits/rejected": -1.5935546159744263, |
| "logps/chosen": -226.49783325195312, |
| "logps/rejected": -299.1880187988281, |
| "loss": 0.2518, |
| "rewards/accuracies": 0.887499988079071, |
| "rewards/chosen": -0.6116968393325806, |
| "rewards/margins": 2.668179750442505, |
| "rewards/rejected": -3.279876232147217, |
| "step": 225 |
| }, |
| { |
| "epoch": 0.49066666666666664, |
| "grad_norm": 2.324882745742798, |
| "learning_rate": 4.7740799883862966e-06, |
| "logits/chosen": -1.648912787437439, |
| "logits/rejected": -1.658570647239685, |
| "logps/chosen": -215.8841094970703, |
| "logps/rejected": -304.59271240234375, |
| "loss": 0.2392, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -0.4058478772640228, |
| "rewards/margins": 2.8351776599884033, |
| "rewards/rejected": -3.241025447845459, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.5013333333333333, |
| "grad_norm": 3.683439016342163, |
| "learning_rate": 4.761969555209394e-06, |
| "logits/chosen": -1.8577114343643188, |
| "logits/rejected": -1.7553380727767944, |
| "logps/chosen": -217.13571166992188, |
| "logps/rejected": -311.47308349609375, |
| "loss": 0.2889, |
| "rewards/accuracies": 0.8687499761581421, |
| "rewards/chosen": -0.7178010940551758, |
| "rewards/margins": 2.7824273109436035, |
| "rewards/rejected": -3.5002281665802, |
| "step": 235 |
| }, |
| { |
| "epoch": 0.512, |
| "grad_norm": 2.6882176399230957, |
| "learning_rate": 4.7495591415709435e-06, |
| "logits/chosen": -1.641305685043335, |
| "logits/rejected": -1.5148091316223145, |
| "logps/chosen": -220.66104125976562, |
| "logps/rejected": -293.9461669921875, |
| "loss": 0.2823, |
| "rewards/accuracies": 0.887499988079071, |
| "rewards/chosen": -0.7442353963851929, |
| "rewards/margins": 2.766519546508789, |
| "rewards/rejected": -3.5107548236846924, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.5226666666666666, |
| "grad_norm": 2.4825658798217773, |
| "learning_rate": 4.736850393329353e-06, |
| "logits/chosen": -1.8355581760406494, |
| "logits/rejected": -1.5800683498382568, |
| "logps/chosen": -246.59066772460938, |
| "logps/rejected": -295.7894287109375, |
| "loss": 0.2453, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.5356179475784302, |
| "rewards/margins": 2.7746822834014893, |
| "rewards/rejected": -3.31030011177063, |
| "step": 245 |
| }, |
| { |
| "epoch": 0.5333333333333333, |
| "grad_norm": 3.213561773300171, |
| "learning_rate": 4.7238449959079095e-06, |
| "logits/chosen": -1.7159204483032227, |
| "logits/rejected": -1.582465410232544, |
| "logps/chosen": -226.40658569335938, |
| "logps/rejected": -287.4487609863281, |
| "loss": 0.251, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -0.7098070383071899, |
| "rewards/margins": 2.7901222705841064, |
| "rewards/rejected": -3.499929428100586, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.544, |
| "grad_norm": 2.457315444946289, |
| "learning_rate": 4.710544674071259e-06, |
| "logits/chosen": -1.7356235980987549, |
| "logits/rejected": -1.568500280380249, |
| "logps/chosen": -234.6002655029297, |
| "logps/rejected": -301.8827209472656, |
| "loss": 0.2264, |
| "rewards/accuracies": 0.918749988079071, |
| "rewards/chosen": -0.7143124938011169, |
| "rewards/margins": 2.9733080863952637, |
| "rewards/rejected": -3.6876206398010254, |
| "step": 255 |
| }, |
| { |
| "epoch": 0.5546666666666666, |
| "grad_norm": 2.9479100704193115, |
| "learning_rate": 4.696951191696673e-06, |
| "logits/chosen": -1.7711416482925415, |
| "logits/rejected": -1.6228008270263672, |
| "logps/chosen": -216.1881561279297, |
| "logps/rejected": -294.84771728515625, |
| "loss": 0.2206, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -0.7808042764663696, |
| "rewards/margins": 3.0704331398010254, |
| "rewards/rejected": -3.8512375354766846, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.5653333333333334, |
| "grad_norm": 2.788898468017578, |
| "learning_rate": 4.6830663515401196e-06, |
| "logits/chosen": -1.7666418552398682, |
| "logits/rejected": -1.5968812704086304, |
| "logps/chosen": -238.59384155273438, |
| "logps/rejected": -305.6246643066406, |
| "loss": 0.2035, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -0.8097629547119141, |
| "rewards/margins": 3.0955402851104736, |
| "rewards/rejected": -3.9053032398223877, |
| "step": 265 |
| }, |
| { |
| "epoch": 0.576, |
| "grad_norm": 3.132816791534424, |
| "learning_rate": 4.6688919949971855e-06, |
| "logits/chosen": -1.7248319387435913, |
| "logits/rejected": -1.5983197689056396, |
| "logps/chosen": -229.5932159423828, |
| "logps/rejected": -309.0335388183594, |
| "loss": 0.2052, |
| "rewards/accuracies": 0.925000011920929, |
| "rewards/chosen": -0.7570894956588745, |
| "rewards/margins": 3.117053747177124, |
| "rewards/rejected": -3.874143600463867, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.5866666666666667, |
| "grad_norm": 4.418814182281494, |
| "learning_rate": 4.654430001858874e-06, |
| "logits/chosen": -1.6472892761230469, |
| "logits/rejected": -1.5750362873077393, |
| "logps/chosen": -235.33743286132812, |
| "logps/rejected": -306.19622802734375, |
| "loss": 0.2581, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -0.7988941669464111, |
| "rewards/margins": 3.0710582733154297, |
| "rewards/rejected": -3.869952440261841, |
| "step": 275 |
| }, |
| { |
| "epoch": 0.5973333333333334, |
| "grad_norm": 3.9372878074645996, |
| "learning_rate": 4.6396822900623075e-06, |
| "logits/chosen": -1.8032338619232178, |
| "logits/rejected": -1.6216585636138916, |
| "logps/chosen": -225.59542846679688, |
| "logps/rejected": -291.0912780761719, |
| "loss": 0.2022, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.7199317216873169, |
| "rewards/margins": 3.431600570678711, |
| "rewards/rejected": -4.151532173156738, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.608, |
| "grad_norm": 3.708203077316284, |
| "learning_rate": 4.624650815436368e-06, |
| "logits/chosen": -1.7838048934936523, |
| "logits/rejected": -1.6193605661392212, |
| "logps/chosen": -247.69082641601562, |
| "logps/rejected": -310.2096862792969, |
| "loss": 0.2188, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -0.9647334814071655, |
| "rewards/margins": 3.321147918701172, |
| "rewards/rejected": -4.285881519317627, |
| "step": 285 |
| }, |
| { |
| "epoch": 0.6186666666666667, |
| "grad_norm": 1.9725751876831055, |
| "learning_rate": 4.609337571442324e-06, |
| "logits/chosen": -1.813089370727539, |
| "logits/rejected": -1.6678119897842407, |
| "logps/chosen": -224.18002319335938, |
| "logps/rejected": -306.02935791015625, |
| "loss": 0.2495, |
| "rewards/accuracies": 0.887499988079071, |
| "rewards/chosen": -1.1165226697921753, |
| "rewards/margins": 3.2529945373535156, |
| "rewards/rejected": -4.369516849517822, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.6293333333333333, |
| "grad_norm": 5.2026777267456055, |
| "learning_rate": 4.5937445889094535e-06, |
| "logits/chosen": -1.7238123416900635, |
| "logits/rejected": -1.6830676794052124, |
| "logps/chosen": -245.8369598388672, |
| "logps/rejected": -311.4858093261719, |
| "loss": 0.2428, |
| "rewards/accuracies": 0.8999999761581421, |
| "rewards/chosen": -0.6427221894264221, |
| "rewards/margins": 3.463367462158203, |
| "rewards/rejected": -4.1060895919799805, |
| "step": 295 |
| }, |
| { |
| "epoch": 0.64, |
| "grad_norm": 2.9667296409606934, |
| "learning_rate": 4.577873935765722e-06, |
| "logits/chosen": -1.8424475193023682, |
| "logits/rejected": -1.6727721691131592, |
| "logps/chosen": -232.18850708007812, |
| "logps/rejected": -304.9006042480469, |
| "loss": 0.199, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.9507037401199341, |
| "rewards/margins": 3.463369846343994, |
| "rewards/rejected": -4.4140729904174805, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.6506666666666666, |
| "grad_norm": 2.3393447399139404, |
| "learning_rate": 4.561727716763532e-06, |
| "logits/chosen": -1.702309012413025, |
| "logits/rejected": -1.6070747375488281, |
| "logps/chosen": -245.9197540283203, |
| "logps/rejected": -308.6512756347656, |
| "loss": 0.2258, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -0.8237674832344055, |
| "rewards/margins": 3.2607333660125732, |
| "rewards/rejected": -4.084500789642334, |
| "step": 305 |
| }, |
| { |
| "epoch": 0.6613333333333333, |
| "grad_norm": 4.266489505767822, |
| "learning_rate": 4.545308073200591e-06, |
| "logits/chosen": -1.7092502117156982, |
| "logits/rejected": -1.588419795036316, |
| "logps/chosen": -240.1862030029297, |
| "logps/rejected": -300.18084716796875, |
| "loss": 0.2137, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.6268893480300903, |
| "rewards/margins": 3.2068610191345215, |
| "rewards/rejected": -3.8337504863739014, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.672, |
| "grad_norm": 2.109783172607422, |
| "learning_rate": 4.528617182635943e-06, |
| "logits/chosen": -1.8387340307235718, |
| "logits/rejected": -1.728690505027771, |
| "logps/chosen": -226.1668243408203, |
| "logps/rejected": -312.61566162109375, |
| "loss": 0.2581, |
| "rewards/accuracies": 0.8812500238418579, |
| "rewards/chosen": -0.7394665479660034, |
| "rewards/margins": 3.5360705852508545, |
| "rewards/rejected": -4.27553653717041, |
| "step": 315 |
| }, |
| { |
| "epoch": 0.6826666666666666, |
| "grad_norm": 2.6758790016174316, |
| "learning_rate": 4.511657258601171e-06, |
| "logits/chosen": -1.7539641857147217, |
| "logits/rejected": -1.6158876419067383, |
| "logps/chosen": -239.16726684570312, |
| "logps/rejected": -306.02581787109375, |
| "loss": 0.2369, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -0.7840451002120972, |
| "rewards/margins": 3.146629810333252, |
| "rewards/rejected": -3.9306750297546387, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.6933333333333334, |
| "grad_norm": 1.7696614265441895, |
| "learning_rate": 4.494430550306848e-06, |
| "logits/chosen": -1.5977143049240112, |
| "logits/rejected": -1.527240514755249, |
| "logps/chosen": -252.3458709716797, |
| "logps/rejected": -297.1427307128906, |
| "loss": 0.2069, |
| "rewards/accuracies": 0.9312499761581421, |
| "rewards/chosen": -0.7750762701034546, |
| "rewards/margins": 3.285799741744995, |
| "rewards/rejected": -4.06087589263916, |
| "step": 325 |
| }, |
| { |
| "epoch": 0.704, |
| "grad_norm": 2.7009758949279785, |
| "learning_rate": 4.476939342344246e-06, |
| "logits/chosen": -1.8222010135650635, |
| "logits/rejected": -1.6697711944580078, |
| "logps/chosen": -236.01950073242188, |
| "logps/rejected": -309.9307861328125, |
| "loss": 0.2076, |
| "rewards/accuracies": 0.925000011920929, |
| "rewards/chosen": -0.6853605508804321, |
| "rewards/margins": 3.5107035636901855, |
| "rewards/rejected": -4.196064472198486, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.7146666666666667, |
| "grad_norm": 2.219874143600464, |
| "learning_rate": 4.45918595438236e-06, |
| "logits/chosen": -1.7028766870498657, |
| "logits/rejected": -1.6242986917495728, |
| "logps/chosen": -228.4136505126953, |
| "logps/rejected": -323.2743835449219, |
| "loss": 0.1681, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -0.4335177540779114, |
| "rewards/margins": 3.8759121894836426, |
| "rewards/rejected": -4.309430122375488, |
| "step": 335 |
| }, |
| { |
| "epoch": 0.7253333333333334, |
| "grad_norm": 3.1199467182159424, |
| "learning_rate": 4.441172740860267e-06, |
| "logits/chosen": -1.7407314777374268, |
| "logits/rejected": -1.668487548828125, |
| "logps/chosen": -231.2909698486328, |
| "logps/rejected": -320.69512939453125, |
| "loss": 0.1919, |
| "rewards/accuracies": 0.918749988079071, |
| "rewards/chosen": -0.6203645467758179, |
| "rewards/margins": 3.9102706909179688, |
| "rewards/rejected": -4.530634880065918, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.736, |
| "grad_norm": 1.6120609045028687, |
| "learning_rate": 4.422902090674886e-06, |
| "logits/chosen": -1.7064342498779297, |
| "logits/rejected": -1.6468336582183838, |
| "logps/chosen": -221.8558349609375, |
| "logps/rejected": -309.10345458984375, |
| "loss": 0.1907, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.9249537587165833, |
| "rewards/margins": 3.7362067699432373, |
| "rewards/rejected": -4.661160945892334, |
| "step": 345 |
| }, |
| { |
| "epoch": 0.7466666666666667, |
| "grad_norm": 3.919717788696289, |
| "learning_rate": 4.4043764268641705e-06, |
| "logits/chosen": -1.7625980377197266, |
| "logits/rejected": -1.5392252206802368, |
| "logps/chosen": -247.5630645751953, |
| "logps/rejected": -319.32684326171875, |
| "loss": 0.197, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -1.0334270000457764, |
| "rewards/margins": 4.253495216369629, |
| "rewards/rejected": -5.286921977996826, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.7573333333333333, |
| "grad_norm": 3.917741537094116, |
| "learning_rate": 4.385598206285758e-06, |
| "logits/chosen": -1.8463985919952393, |
| "logits/rejected": -1.6996173858642578, |
| "logps/chosen": -250.5867156982422, |
| "logps/rejected": -340.89422607421875, |
| "loss": 0.2088, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -1.037157654762268, |
| "rewards/margins": 4.255171298980713, |
| "rewards/rejected": -5.292328834533691, |
| "step": 355 |
| }, |
| { |
| "epoch": 0.768, |
| "grad_norm": 3.468600273132324, |
| "learning_rate": 4.3665699192911495e-06, |
| "logits/chosen": -1.6923949718475342, |
| "logits/rejected": -1.5864944458007812, |
| "logps/chosen": -227.09860229492188, |
| "logps/rejected": -320.20343017578125, |
| "loss": 0.203, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -1.0783360004425049, |
| "rewards/margins": 3.753375291824341, |
| "rewards/rejected": -4.831711769104004, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.7786666666666666, |
| "grad_norm": 3.8932225704193115, |
| "learning_rate": 4.347294089395438e-06, |
| "logits/chosen": -1.6695184707641602, |
| "logits/rejected": -1.5590105056762695, |
| "logps/chosen": -247.1300048828125, |
| "logps/rejected": -310.8145446777344, |
| "loss": 0.1774, |
| "rewards/accuracies": 0.918749988079071, |
| "rewards/chosen": -0.9608127474784851, |
| "rewards/margins": 3.990257978439331, |
| "rewards/rejected": -4.951070785522461, |
| "step": 365 |
| }, |
| { |
| "epoch": 0.7893333333333333, |
| "grad_norm": 3.3788342475891113, |
| "learning_rate": 4.327773272942647e-06, |
| "logits/chosen": -1.618805170059204, |
| "logits/rejected": -1.5945888757705688, |
| "logps/chosen": -225.7582244873047, |
| "logps/rejected": -319.0153503417969, |
| "loss": 0.1866, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -0.9858270883560181, |
| "rewards/margins": 3.8777613639831543, |
| "rewards/rejected": -4.863588809967041, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.8, |
| "grad_norm": 2.828134059906006, |
| "learning_rate": 4.308010058766703e-06, |
| "logits/chosen": -1.7046676874160767, |
| "logits/rejected": -1.5746122598648071, |
| "logps/chosen": -300.2704162597656, |
| "logps/rejected": -307.11297607421875, |
| "loss": 0.1834, |
| "rewards/accuracies": 0.918749988079071, |
| "rewards/chosen": -1.043919324874878, |
| "rewards/margins": 3.8452529907226562, |
| "rewards/rejected": -4.889172554016113, |
| "step": 375 |
| }, |
| { |
| "epoch": 0.8106666666666666, |
| "grad_norm": 2.781174659729004, |
| "learning_rate": 4.2880070678481135e-06, |
| "logits/chosen": -1.7193530797958374, |
| "logits/rejected": -1.5741509199142456, |
| "logps/chosen": -222.92428588867188, |
| "logps/rejected": -311.06689453125, |
| "loss": 0.1894, |
| "rewards/accuracies": 0.925000011920929, |
| "rewards/chosen": -1.1129997968673706, |
| "rewards/margins": 3.535658597946167, |
| "rewards/rejected": -4.648658275604248, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.8213333333333334, |
| "grad_norm": 2.395315170288086, |
| "learning_rate": 4.267766952966369e-06, |
| "logits/chosen": -1.6871780157089233, |
| "logits/rejected": -1.5884290933609009, |
| "logps/chosen": -221.77444458007812, |
| "logps/rejected": -321.648681640625, |
| "loss": 0.1822, |
| "rewards/accuracies": 0.9312499761581421, |
| "rewards/chosen": -0.9971541166305542, |
| "rewards/margins": 4.199106216430664, |
| "rewards/rejected": -5.196260452270508, |
| "step": 385 |
| }, |
| { |
| "epoch": 0.832, |
| "grad_norm": 3.117591142654419, |
| "learning_rate": 4.24729239834814e-06, |
| "logits/chosen": -1.5949792861938477, |
| "logits/rejected": -1.5942808389663696, |
| "logps/chosen": -230.6989288330078, |
| "logps/rejected": -321.4884033203125, |
| "loss": 0.205, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -1.0500239133834839, |
| "rewards/margins": 3.972008466720581, |
| "rewards/rejected": -5.022032737731934, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.8426666666666667, |
| "grad_norm": 2.8519601821899414, |
| "learning_rate": 4.226586119311291e-06, |
| "logits/chosen": -1.7170889377593994, |
| "logits/rejected": -1.6525356769561768, |
| "logps/chosen": -238.529296875, |
| "logps/rejected": -329.4853820800781, |
| "loss": 0.2118, |
| "rewards/accuracies": 0.925000011920929, |
| "rewards/chosen": -1.1062920093536377, |
| "rewards/margins": 3.961061954498291, |
| "rewards/rejected": -5.06735372543335, |
| "step": 395 |
| }, |
| { |
| "epoch": 0.8533333333333334, |
| "grad_norm": 1.4091684818267822, |
| "learning_rate": 4.205650861904782e-06, |
| "logits/chosen": -1.681089162826538, |
| "logits/rejected": -1.5169637203216553, |
| "logps/chosen": -227.4136199951172, |
| "logps/rejected": -329.40985107421875, |
| "loss": 0.1699, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.3499810695648193, |
| "rewards/margins": 4.325152397155762, |
| "rewards/rejected": -5.67513370513916, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.864, |
| "grad_norm": 2.204573392868042, |
| "learning_rate": 4.184489402544486e-06, |
| "logits/chosen": -1.7952913045883179, |
| "logits/rejected": -1.6552798748016357, |
| "logps/chosen": -227.94210815429688, |
| "logps/rejected": -321.5538635253906, |
| "loss": 0.1467, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.0351591110229492, |
| "rewards/margins": 4.314890384674072, |
| "rewards/rejected": -5.350049018859863, |
| "step": 405 |
| }, |
| { |
| "epoch": 0.8746666666666667, |
| "grad_norm": 4.7300639152526855, |
| "learning_rate": 4.1631045476449865e-06, |
| "logits/chosen": -1.7709486484527588, |
| "logits/rejected": -1.5174936056137085, |
| "logps/chosen": -234.28475952148438, |
| "logps/rejected": -306.0198059082031, |
| "loss": 0.2281, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -1.6678578853607178, |
| "rewards/margins": 3.790971040725708, |
| "rewards/rejected": -5.458829402923584, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.8853333333333333, |
| "grad_norm": 2.627563953399658, |
| "learning_rate": 4.141499133247391e-06, |
| "logits/chosen": -1.6847022771835327, |
| "logits/rejected": -1.5950194597244263, |
| "logps/chosen": -238.90029907226562, |
| "logps/rejected": -310.26214599609375, |
| "loss": 0.1812, |
| "rewards/accuracies": 0.9312499761581421, |
| "rewards/chosen": -1.1100462675094604, |
| "rewards/margins": 4.378572463989258, |
| "rewards/rejected": -5.488618850708008, |
| "step": 415 |
| }, |
| { |
| "epoch": 0.896, |
| "grad_norm": 3.2908780574798584, |
| "learning_rate": 4.119676024643217e-06, |
| "logits/chosen": -1.6498327255249023, |
| "logits/rejected": -1.6628739833831787, |
| "logps/chosen": -232.87203979492188, |
| "logps/rejected": -300.0521240234375, |
| "loss": 0.2159, |
| "rewards/accuracies": 0.887499988079071, |
| "rewards/chosen": -1.1272834539413452, |
| "rewards/margins": 3.792628765106201, |
| "rewards/rejected": -4.919912815093994, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.9066666666666666, |
| "grad_norm": 1.855165958404541, |
| "learning_rate": 4.0976381159944055e-06, |
| "logits/chosen": -1.6586782932281494, |
| "logits/rejected": -1.517372965812683, |
| "logps/chosen": -258.5635986328125, |
| "logps/rejected": -344.59893798828125, |
| "loss": 0.1408, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.5040361881256104, |
| "rewards/margins": 4.680717468261719, |
| "rewards/rejected": -6.184753894805908, |
| "step": 425 |
| }, |
| { |
| "epoch": 0.9173333333333333, |
| "grad_norm": 5.024399757385254, |
| "learning_rate": 4.07538832994949e-06, |
| "logits/chosen": -1.7532565593719482, |
| "logits/rejected": -1.5877439975738525, |
| "logps/chosen": -236.11898803710938, |
| "logps/rejected": -334.73052978515625, |
| "loss": 0.2139, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -1.5054124593734741, |
| "rewards/margins": 4.465746879577637, |
| "rewards/rejected": -5.971159934997559, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.928, |
| "grad_norm": 2.292858362197876, |
| "learning_rate": 4.0529296172560055e-06, |
| "logits/chosen": -1.8473304510116577, |
| "logits/rejected": -1.7024672031402588, |
| "logps/chosen": -237.39791870117188, |
| "logps/rejected": -326.8873291015625, |
| "loss": 0.2034, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -1.3960918188095093, |
| "rewards/margins": 4.522124290466309, |
| "rewards/rejected": -5.918216705322266, |
| "step": 435 |
| }, |
| { |
| "epoch": 0.9386666666666666, |
| "grad_norm": 3.1267998218536377, |
| "learning_rate": 4.030264956369158e-06, |
| "logits/chosen": -1.7679741382598877, |
| "logits/rejected": -1.6751598119735718, |
| "logps/chosen": -217.69375610351562, |
| "logps/rejected": -321.40911865234375, |
| "loss": 0.2045, |
| "rewards/accuracies": 0.8999999761581421, |
| "rewards/chosen": -1.2226085662841797, |
| "rewards/margins": 4.901395320892334, |
| "rewards/rejected": -6.1240034103393555, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.9493333333333334, |
| "grad_norm": 2.8430871963500977, |
| "learning_rate": 4.0073973530568274e-06, |
| "logits/chosen": -1.7301061153411865, |
| "logits/rejected": -1.5988894701004028, |
| "logps/chosen": -218.8601837158203, |
| "logps/rejected": -315.28485107421875, |
| "loss": 0.21, |
| "rewards/accuracies": 0.893750011920929, |
| "rewards/chosen": -1.3717520236968994, |
| "rewards/margins": 4.2111992835998535, |
| "rewards/rejected": -5.582951545715332, |
| "step": 445 |
| }, |
| { |
| "epoch": 0.96, |
| "grad_norm": 2.1232240200042725, |
| "learning_rate": 3.9843298400009435e-06, |
| "logits/chosen": -1.818627119064331, |
| "logits/rejected": -1.711194396018982, |
| "logps/chosen": -238.9432373046875, |
| "logps/rejected": -327.90936279296875, |
| "loss": 0.17, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -1.2995021343231201, |
| "rewards/margins": 4.501131534576416, |
| "rewards/rejected": -5.800633430480957, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.9706666666666667, |
| "grad_norm": 4.588263511657715, |
| "learning_rate": 3.961065476395296e-06, |
| "logits/chosen": -1.737735390663147, |
| "logits/rejected": -1.6649131774902344, |
| "logps/chosen": -250.17367553710938, |
| "logps/rejected": -330.4911193847656, |
| "loss": 0.2198, |
| "rewards/accuracies": 0.918749988079071, |
| "rewards/chosen": -1.7452892065048218, |
| "rewards/margins": 4.413934230804443, |
| "rewards/rejected": -6.159224033355713, |
| "step": 455 |
| }, |
| { |
| "epoch": 0.9813333333333333, |
| "grad_norm": 2.611341714859009, |
| "learning_rate": 3.937607347539823e-06, |
| "logits/chosen": -1.69869065284729, |
| "logits/rejected": -1.5674892663955688, |
| "logps/chosen": -231.4336700439453, |
| "logps/rejected": -311.5548400878906, |
| "loss": 0.192, |
| "rewards/accuracies": 0.918749988079071, |
| "rewards/chosen": -1.5492182970046997, |
| "rewards/margins": 4.648384094238281, |
| "rewards/rejected": -6.19760274887085, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.992, |
| "grad_norm": 4.72195291519165, |
| "learning_rate": 3.913958564431447e-06, |
| "logits/chosen": -1.8486287593841553, |
| "logits/rejected": -1.6863996982574463, |
| "logps/chosen": -240.7964324951172, |
| "logps/rejected": -322.6090393066406, |
| "loss": 0.1649, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.4922518730163574, |
| "rewards/margins": 4.469428539276123, |
| "rewards/rejected": -5.961680889129639, |
| "step": 465 |
| }, |
| { |
| "epoch": 1.0021333333333333, |
| "grad_norm": 4.31331205368042, |
| "learning_rate": 3.89012226335149e-06, |
| "logits/chosen": -1.733007550239563, |
| "logits/rejected": -1.6687922477722168, |
| "logps/chosen": -235.17596435546875, |
| "logps/rejected": -336.431884765625, |
| "loss": 0.1628, |
| "rewards/accuracies": 0.9276315569877625, |
| "rewards/chosen": -1.2398635149002075, |
| "rewards/margins": 4.856743812561035, |
| "rewards/rejected": -6.096607208251953, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.0128, |
| "grad_norm": 0.9625500440597534, |
| "learning_rate": 3.86610160544975e-06, |
| "logits/chosen": -1.7513107061386108, |
| "logits/rejected": -1.7320438623428345, |
| "logps/chosen": -236.9391326904297, |
| "logps/rejected": -353.11468505859375, |
| "loss": 0.119, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.2717801332473755, |
| "rewards/margins": 5.247875690460205, |
| "rewards/rejected": -6.519656181335449, |
| "step": 475 |
| }, |
| { |
| "epoch": 1.0234666666666667, |
| "grad_norm": 1.7186386585235596, |
| "learning_rate": 3.841899776325267e-06, |
| "logits/chosen": -1.6451976299285889, |
| "logits/rejected": -1.5232036113739014, |
| "logps/chosen": -248.01620483398438, |
| "logps/rejected": -315.8173522949219, |
| "loss": 0.1408, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.2212826013565063, |
| "rewards/margins": 4.718570232391357, |
| "rewards/rejected": -5.939852714538574, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.0341333333333333, |
| "grad_norm": 2.955115795135498, |
| "learning_rate": 3.817519985603855e-06, |
| "logits/chosen": -1.7891451120376587, |
| "logits/rejected": -1.6652504205703735, |
| "logps/chosen": -222.29953002929688, |
| "logps/rejected": -318.55938720703125, |
| "loss": 0.1148, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.5009971857070923, |
| "rewards/margins": 5.109112739562988, |
| "rewards/rejected": -6.610108852386475, |
| "step": 485 |
| }, |
| { |
| "epoch": 1.0448, |
| "grad_norm": 1.8832951784133911, |
| "learning_rate": 3.792965466512437e-06, |
| "logits/chosen": -1.7436507940292358, |
| "logits/rejected": -1.6489429473876953, |
| "logps/chosen": -251.84146118164062, |
| "logps/rejected": -335.59271240234375, |
| "loss": 0.1024, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -1.1944636106491089, |
| "rewards/margins": 5.200083255767822, |
| "rewards/rejected": -6.3945465087890625, |
| "step": 490 |
| }, |
| { |
| "epoch": 1.0554666666666668, |
| "grad_norm": 3.0533058643341064, |
| "learning_rate": 3.7682394754502687e-06, |
| "logits/chosen": -1.7651679515838623, |
| "logits/rejected": -1.6751089096069336, |
| "logps/chosen": -241.49221801757812, |
| "logps/rejected": -335.492431640625, |
| "loss": 0.123, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -1.5157369375228882, |
| "rewards/margins": 5.0457444190979, |
| "rewards/rejected": -6.561480522155762, |
| "step": 495 |
| }, |
| { |
| "epoch": 1.0661333333333334, |
| "grad_norm": 2.936387777328491, |
| "learning_rate": 3.7433452915570685e-06, |
| "logits/chosen": -1.6875702142715454, |
| "logits/rejected": -1.6360981464385986, |
| "logps/chosen": -230.61770629882812, |
| "logps/rejected": -336.7130432128906, |
| "loss": 0.1503, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.8132297992706299, |
| "rewards/margins": 5.069985866546631, |
| "rewards/rejected": -6.883215427398682, |
| "step": 500 |
| }, |
| { |
| "epoch": 1.0768, |
| "grad_norm": 3.192586660385132, |
| "learning_rate": 3.718286216278143e-06, |
| "logits/chosen": -1.817999243736267, |
| "logits/rejected": -1.712794542312622, |
| "logps/chosen": -217.67965698242188, |
| "logps/rejected": -315.28875732421875, |
| "loss": 0.1817, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.406536340713501, |
| "rewards/margins": 5.129839897155762, |
| "rewards/rejected": -6.536375999450684, |
| "step": 505 |
| }, |
| { |
| "epoch": 1.0874666666666666, |
| "grad_norm": 4.433254718780518, |
| "learning_rate": 3.6930655729265557e-06, |
| "logits/chosen": -1.7239211797714233, |
| "logits/rejected": -1.6257041692733765, |
| "logps/chosen": -254.9601287841797, |
| "logps/rejected": -342.13629150390625, |
| "loss": 0.1264, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -1.459402322769165, |
| "rewards/margins": 5.333830833435059, |
| "rewards/rejected": -6.793233394622803, |
| "step": 510 |
| }, |
| { |
| "epoch": 1.0981333333333334, |
| "grad_norm": 3.8896450996398926, |
| "learning_rate": 3.667686706242386e-06, |
| "logits/chosen": -1.6991560459136963, |
| "logits/rejected": -1.5692527294158936, |
| "logps/chosen": -231.50454711914062, |
| "logps/rejected": -332.5694885253906, |
| "loss": 0.117, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.4644403457641602, |
| "rewards/margins": 5.550719261169434, |
| "rewards/rejected": -7.015158653259277, |
| "step": 515 |
| }, |
| { |
| "epoch": 1.1088, |
| "grad_norm": 2.798954963684082, |
| "learning_rate": 3.6421529819491576e-06, |
| "logits/chosen": -1.740290641784668, |
| "logits/rejected": -1.6251468658447266, |
| "logps/chosen": -247.70217895507812, |
| "logps/rejected": -324.9705810546875, |
| "loss": 0.1123, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.6576683521270752, |
| "rewards/margins": 5.431394100189209, |
| "rewards/rejected": -7.089062690734863, |
| "step": 520 |
| }, |
| { |
| "epoch": 1.1194666666666666, |
| "grad_norm": 3.2726736068725586, |
| "learning_rate": 3.6164677863074783e-06, |
| "logits/chosen": -1.7375444173812866, |
| "logits/rejected": -1.7175487279891968, |
| "logps/chosen": -254.1574249267578, |
| "logps/rejected": -331.61798095703125, |
| "loss": 0.1336, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -2.0355722904205322, |
| "rewards/margins": 4.999141216278076, |
| "rewards/rejected": -7.0347137451171875, |
| "step": 525 |
| }, |
| { |
| "epoch": 1.1301333333333332, |
| "grad_norm": 3.4253458976745605, |
| "learning_rate": 3.590634525665955e-06, |
| "logits/chosen": -1.838283896446228, |
| "logits/rejected": -1.6944366693496704, |
| "logps/chosen": -230.379150390625, |
| "logps/rejected": -352.56005859375, |
| "loss": 0.1093, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.7831109762191772, |
| "rewards/margins": 5.638839244842529, |
| "rewards/rejected": -7.421949863433838, |
| "step": 530 |
| }, |
| { |
| "epoch": 1.1408, |
| "grad_norm": 1.8465055227279663, |
| "learning_rate": 3.564656626009446e-06, |
| "logits/chosen": -1.765180230140686, |
| "logits/rejected": -1.7436046600341797, |
| "logps/chosen": -231.00894165039062, |
| "logps/rejected": -354.35552978515625, |
| "loss": 0.1105, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -2.084205150604248, |
| "rewards/margins": 6.009603500366211, |
| "rewards/rejected": -8.093809127807617, |
| "step": 535 |
| }, |
| { |
| "epoch": 1.1514666666666666, |
| "grad_norm": 2.5956075191497803, |
| "learning_rate": 3.5385375325047167e-06, |
| "logits/chosen": -1.775918960571289, |
| "logits/rejected": -1.6911274194717407, |
| "logps/chosen": -240.2334747314453, |
| "logps/rejected": -335.20184326171875, |
| "loss": 0.1565, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.8987281322479248, |
| "rewards/margins": 5.527867794036865, |
| "rewards/rejected": -7.426596164703369, |
| "step": 540 |
| }, |
| { |
| "epoch": 1.1621333333333332, |
| "grad_norm": 2.830385684967041, |
| "learning_rate": 3.512280709043534e-06, |
| "logits/chosen": -1.6608412265777588, |
| "logits/rejected": -1.5417561531066895, |
| "logps/chosen": -261.1434326171875, |
| "logps/rejected": -343.75775146484375, |
| "loss": 0.1444, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.7868309020996094, |
| "rewards/margins": 5.3465094566345215, |
| "rewards/rejected": -7.133339881896973, |
| "step": 545 |
| }, |
| { |
| "epoch": 1.1728, |
| "grad_norm": 1.6458228826522827, |
| "learning_rate": 3.4858896377832966e-06, |
| "logits/chosen": -1.682835340499878, |
| "logits/rejected": -1.682253122329712, |
| "logps/chosen": -228.0749969482422, |
| "logps/rejected": -350.91064453125, |
| "loss": 0.1117, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.8750747442245483, |
| "rewards/margins": 5.591343879699707, |
| "rewards/rejected": -7.466418266296387, |
| "step": 550 |
| }, |
| { |
| "epoch": 1.1834666666666667, |
| "grad_norm": 4.969726085662842, |
| "learning_rate": 3.459367818685231e-06, |
| "logits/chosen": -1.7055584192276, |
| "logits/rejected": -1.6184914112091064, |
| "logps/chosen": -246.9654998779297, |
| "logps/rejected": -335.02557373046875, |
| "loss": 0.1431, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.870309829711914, |
| "rewards/margins": 5.379855155944824, |
| "rewards/rejected": -7.250164031982422, |
| "step": 555 |
| }, |
| { |
| "epoch": 1.1941333333333333, |
| "grad_norm": 4.240958213806152, |
| "learning_rate": 3.432718769050228e-06, |
| "logits/chosen": -1.6397478580474854, |
| "logits/rejected": -1.5480825901031494, |
| "logps/chosen": -238.30917358398438, |
| "logps/rejected": -341.9546813964844, |
| "loss": 0.1158, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -1.5620613098144531, |
| "rewards/margins": 5.2377424240112305, |
| "rewards/rejected": -6.799803256988525, |
| "step": 560 |
| }, |
| { |
| "epoch": 1.2048, |
| "grad_norm": 3.5481252670288086, |
| "learning_rate": 3.405946023052386e-06, |
| "logits/chosen": -1.758758544921875, |
| "logits/rejected": -1.6746900081634521, |
| "logps/chosen": -235.3990478515625, |
| "logps/rejected": -341.65557861328125, |
| "loss": 0.1671, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -2.0450544357299805, |
| "rewards/margins": 5.749896049499512, |
| "rewards/rejected": -7.794950008392334, |
| "step": 565 |
| }, |
| { |
| "epoch": 1.2154666666666667, |
| "grad_norm": 2.327967882156372, |
| "learning_rate": 3.3790531312703055e-06, |
| "logits/chosen": -1.6582815647125244, |
| "logits/rejected": -1.707456350326538, |
| "logps/chosen": -242.3777313232422, |
| "logps/rejected": -363.1072998046875, |
| "loss": 0.1118, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -2.587329387664795, |
| "rewards/margins": 5.6046342849731445, |
| "rewards/rejected": -8.191963195800781, |
| "step": 570 |
| }, |
| { |
| "epoch": 1.2261333333333333, |
| "grad_norm": 1.739214539527893, |
| "learning_rate": 3.35204366021622e-06, |
| "logits/chosen": -1.6814908981323242, |
| "logits/rejected": -1.575382947921753, |
| "logps/chosen": -249.67098999023438, |
| "logps/rejected": -339.2182922363281, |
| "loss": 0.1331, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.7393226623535156, |
| "rewards/margins": 5.339315414428711, |
| "rewards/rejected": -7.078637599945068, |
| "step": 575 |
| }, |
| { |
| "epoch": 1.2368000000000001, |
| "grad_norm": 5.8394670486450195, |
| "learning_rate": 3.3249211918630055e-06, |
| "logits/chosen": -1.6644290685653687, |
| "logits/rejected": -1.6048500537872314, |
| "logps/chosen": -228.67880249023438, |
| "logps/rejected": -353.0116271972656, |
| "loss": 0.1261, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.9887882471084595, |
| "rewards/margins": 5.781205177307129, |
| "rewards/rejected": -7.769992828369141, |
| "step": 580 |
| }, |
| { |
| "epoch": 1.2474666666666667, |
| "grad_norm": 4.61240816116333, |
| "learning_rate": 3.2976893231691424e-06, |
| "logits/chosen": -1.7240492105484009, |
| "logits/rejected": -1.7118297815322876, |
| "logps/chosen": -257.6966247558594, |
| "logps/rejected": -348.775146484375, |
| "loss": 0.1438, |
| "rewards/accuracies": 0.925000011920929, |
| "rewards/chosen": -1.5069431066513062, |
| "rewards/margins": 5.710946559906006, |
| "rewards/rejected": -7.217889308929443, |
| "step": 585 |
| }, |
| { |
| "epoch": 1.2581333333333333, |
| "grad_norm": 1.9813047647476196, |
| "learning_rate": 3.270351665601691e-06, |
| "logits/chosen": -1.636423110961914, |
| "logits/rejected": -1.626508355140686, |
| "logps/chosen": -241.73611450195312, |
| "logps/rejected": -322.9097595214844, |
| "loss": 0.1514, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.60867178440094, |
| "rewards/margins": 5.148347854614258, |
| "rewards/rejected": -6.757019996643066, |
| "step": 590 |
| }, |
| { |
| "epoch": 1.2688, |
| "grad_norm": 5.205898284912109, |
| "learning_rate": 3.2429118446573417e-06, |
| "logits/chosen": -1.7877315282821655, |
| "logits/rejected": -1.6699727773666382, |
| "logps/chosen": -235.1413116455078, |
| "logps/rejected": -327.6995544433594, |
| "loss": 0.2184, |
| "rewards/accuracies": 0.9125000238418579, |
| "rewards/chosen": -1.8307580947875977, |
| "rewards/margins": 4.723857879638672, |
| "rewards/rejected": -6.554616451263428, |
| "step": 595 |
| }, |
| { |
| "epoch": 1.2794666666666665, |
| "grad_norm": 1.4653977155685425, |
| "learning_rate": 3.215373499381602e-06, |
| "logits/chosen": -1.7254440784454346, |
| "logits/rejected": -1.571353554725647, |
| "logps/chosen": -249.1863250732422, |
| "logps/rejected": -340.4194641113281, |
| "loss": 0.1343, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.406814694404602, |
| "rewards/margins": 5.116804122924805, |
| "rewards/rejected": -6.523619174957275, |
| "step": 600 |
| }, |
| { |
| "epoch": 1.2901333333333334, |
| "grad_norm": 1.798576831817627, |
| "learning_rate": 3.1877402818861954e-06, |
| "logits/chosen": -1.7054555416107178, |
| "logits/rejected": -1.6197830438613892, |
| "logps/chosen": -259.9774475097656, |
| "logps/rejected": -341.87542724609375, |
| "loss": 0.0874, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -1.6282418966293335, |
| "rewards/margins": 5.709589958190918, |
| "rewards/rejected": -7.337832450866699, |
| "step": 605 |
| }, |
| { |
| "epoch": 1.3008, |
| "grad_norm": 5.687100887298584, |
| "learning_rate": 3.1600158568647145e-06, |
| "logits/chosen": -1.7209688425064087, |
| "logits/rejected": -1.6104001998901367, |
| "logps/chosen": -253.5648956298828, |
| "logps/rejected": -316.03363037109375, |
| "loss": 0.1435, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.4639036655426025, |
| "rewards/margins": 5.040863037109375, |
| "rewards/rejected": -6.504767417907715, |
| "step": 610 |
| }, |
| { |
| "epoch": 1.3114666666666666, |
| "grad_norm": 3.0057663917541504, |
| "learning_rate": 3.1322039011066172e-06, |
| "logits/chosen": -1.7133491039276123, |
| "logits/rejected": -1.6010112762451172, |
| "logps/chosen": -265.01580810546875, |
| "logps/rejected": -333.35260009765625, |
| "loss": 0.1287, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.5199475288391113, |
| "rewards/margins": 5.0299577713012695, |
| "rewards/rejected": -6.549904823303223, |
| "step": 615 |
| }, |
| { |
| "epoch": 1.3221333333333334, |
| "grad_norm": 2.881352663040161, |
| "learning_rate": 3.1043081030096125e-06, |
| "logits/chosen": -1.6966040134429932, |
| "logits/rejected": -1.6292721033096313, |
| "logps/chosen": -230.958740234375, |
| "logps/rejected": -332.935791015625, |
| "loss": 0.146, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.135877251625061, |
| "rewards/margins": 5.311655521392822, |
| "rewards/rejected": -6.447532653808594, |
| "step": 620 |
| }, |
| { |
| "epoch": 1.3328, |
| "grad_norm": 3.1088597774505615, |
| "learning_rate": 3.0763321620905072e-06, |
| "logits/chosen": -1.6871414184570312, |
| "logits/rejected": -1.6046022176742554, |
| "logps/chosen": -258.32257080078125, |
| "logps/rejected": -330.11175537109375, |
| "loss": 0.1706, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.5846526622772217, |
| "rewards/margins": 5.45190954208374, |
| "rewards/rejected": -7.036561489105225, |
| "step": 625 |
| }, |
| { |
| "epoch": 1.3434666666666666, |
| "grad_norm": 5.220170974731445, |
| "learning_rate": 3.048279788494583e-06, |
| "logits/chosen": -1.7173973321914673, |
| "logits/rejected": -1.6032285690307617, |
| "logps/chosen": -216.53848266601562, |
| "logps/rejected": -333.8974609375, |
| "loss": 0.1272, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.540093183517456, |
| "rewards/margins": 5.554872035980225, |
| "rewards/rejected": -7.09496545791626, |
| "step": 630 |
| }, |
| { |
| "epoch": 1.3541333333333334, |
| "grad_norm": 3.8425443172454834, |
| "learning_rate": 3.0201547025035553e-06, |
| "logits/chosen": -1.705578088760376, |
| "logits/rejected": -1.5608021020889282, |
| "logps/chosen": -242.1476593017578, |
| "logps/rejected": -323.027587890625, |
| "loss": 0.1314, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.3977354764938354, |
| "rewards/margins": 5.5933837890625, |
| "rewards/rejected": -6.991118431091309, |
| "step": 635 |
| }, |
| { |
| "epoch": 1.3648, |
| "grad_norm": 5.495195388793945, |
| "learning_rate": 2.9919606340421963e-06, |
| "logits/chosen": -1.7403780221939087, |
| "logits/rejected": -1.596381425857544, |
| "logps/chosen": -233.91488647460938, |
| "logps/rejected": -321.4733581542969, |
| "loss": 0.1461, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.7344648838043213, |
| "rewards/margins": 5.078824996948242, |
| "rewards/rejected": -6.813290596008301, |
| "step": 640 |
| }, |
| { |
| "epoch": 1.3754666666666666, |
| "grad_norm": 2.574321746826172, |
| "learning_rate": 2.9637013221836723e-06, |
| "logits/chosen": -1.8257449865341187, |
| "logits/rejected": -1.6127040386199951, |
| "logps/chosen": -254.885009765625, |
| "logps/rejected": -340.189453125, |
| "loss": 0.1662, |
| "rewards/accuracies": 0.8999999761581421, |
| "rewards/chosen": -1.7457427978515625, |
| "rewards/margins": 5.908649444580078, |
| "rewards/rejected": -7.654392242431641, |
| "step": 645 |
| }, |
| { |
| "epoch": 1.3861333333333334, |
| "grad_norm": 3.7627923488616943, |
| "learning_rate": 2.9353805146536762e-06, |
| "logits/chosen": -1.7805149555206299, |
| "logits/rejected": -1.5877693891525269, |
| "logps/chosen": -255.69650268554688, |
| "logps/rejected": -325.4459228515625, |
| "loss": 0.1219, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.5994096994400024, |
| "rewards/margins": 5.30047607421875, |
| "rewards/rejected": -6.899886131286621, |
| "step": 650 |
| }, |
| { |
| "epoch": 1.3968, |
| "grad_norm": 4.437411785125732, |
| "learning_rate": 2.907001967333401e-06, |
| "logits/chosen": -1.8212392330169678, |
| "logits/rejected": -1.7462043762207031, |
| "logps/chosen": -227.9088897705078, |
| "logps/rejected": -346.3284912109375, |
| "loss": 0.1388, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.6472021341323853, |
| "rewards/margins": 6.0712714195251465, |
| "rewards/rejected": -7.718472957611084, |
| "step": 655 |
| }, |
| { |
| "epoch": 1.4074666666666666, |
| "grad_norm": 3.5579206943511963, |
| "learning_rate": 2.878569443761442e-06, |
| "logits/chosen": -1.7327353954315186, |
| "logits/rejected": -1.5691357851028442, |
| "logps/chosen": -235.33078002929688, |
| "logps/rejected": -324.94476318359375, |
| "loss": 0.1518, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.870516061782837, |
| "rewards/margins": 5.5859222412109375, |
| "rewards/rejected": -7.4564385414123535, |
| "step": 660 |
| }, |
| { |
| "epoch": 1.4181333333333335, |
| "grad_norm": 3.057058095932007, |
| "learning_rate": 2.85008671463468e-06, |
| "logits/chosen": -1.7148669958114624, |
| "logits/rejected": -1.6496082544326782, |
| "logps/chosen": -233.3097381591797, |
| "logps/rejected": -328.40191650390625, |
| "loss": 0.1543, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.7619035243988037, |
| "rewards/margins": 5.8997483253479, |
| "rewards/rejected": -7.661650657653809, |
| "step": 665 |
| }, |
| { |
| "epoch": 1.4288, |
| "grad_norm": 3.640073299407959, |
| "learning_rate": 2.8215575573082127e-06, |
| "logits/chosen": -1.813275933265686, |
| "logits/rejected": -1.7277463674545288, |
| "logps/chosen": -225.76553344726562, |
| "logps/rejected": -361.9228820800781, |
| "loss": 0.1072, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.7821929454803467, |
| "rewards/margins": 6.15872859954834, |
| "rewards/rejected": -7.940920829772949, |
| "step": 670 |
| }, |
| { |
| "epoch": 1.4394666666666667, |
| "grad_norm": 1.9339507818222046, |
| "learning_rate": 2.792985755294404e-06, |
| "logits/chosen": -1.7263004779815674, |
| "logits/rejected": -1.6594091653823853, |
| "logps/chosen": -220.85653686523438, |
| "logps/rejected": -345.27886962890625, |
| "loss": 0.121, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.9338741302490234, |
| "rewards/margins": 5.8638129234313965, |
| "rewards/rejected": -7.797686576843262, |
| "step": 675 |
| }, |
| { |
| "epoch": 1.4501333333333333, |
| "grad_norm": 4.9970011711120605, |
| "learning_rate": 2.764375097761122e-06, |
| "logits/chosen": -1.6363716125488281, |
| "logits/rejected": -1.5276905298233032, |
| "logps/chosen": -232.1802215576172, |
| "logps/rejected": -340.8988952636719, |
| "loss": 0.1607, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.3411585092544556, |
| "rewards/margins": 5.480500221252441, |
| "rewards/rejected": -6.821658134460449, |
| "step": 680 |
| }, |
| { |
| "epoch": 1.4607999999999999, |
| "grad_norm": 2.177555799484253, |
| "learning_rate": 2.73572937902922e-06, |
| "logits/chosen": -1.7766189575195312, |
| "logits/rejected": -1.6084314584732056, |
| "logps/chosen": -240.9921417236328, |
| "logps/rejected": -330.74853515625, |
| "loss": 0.1677, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.3653053045272827, |
| "rewards/margins": 5.720976829528809, |
| "rewards/rejected": -7.086282253265381, |
| "step": 685 |
| }, |
| { |
| "epoch": 1.4714666666666667, |
| "grad_norm": 1.8586421012878418, |
| "learning_rate": 2.7070523980693362e-06, |
| "logits/chosen": -1.6043384075164795, |
| "logits/rejected": -1.5343188047409058, |
| "logps/chosen": -251.8693084716797, |
| "logps/rejected": -341.179931640625, |
| "loss": 0.1193, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.6754684448242188, |
| "rewards/margins": 5.657227516174316, |
| "rewards/rejected": -7.332695960998535, |
| "step": 690 |
| }, |
| { |
| "epoch": 1.4821333333333333, |
| "grad_norm": 2.7305967807769775, |
| "learning_rate": 2.678347957998081e-06, |
| "logits/chosen": -1.7863848209381104, |
| "logits/rejected": -1.7329514026641846, |
| "logps/chosen": -242.52517700195312, |
| "logps/rejected": -347.587646484375, |
| "loss": 0.1191, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.808240294456482, |
| "rewards/margins": 5.8397536277771, |
| "rewards/rejected": -7.647993564605713, |
| "step": 695 |
| }, |
| { |
| "epoch": 1.4928, |
| "grad_norm": 3.4223201274871826, |
| "learning_rate": 2.649619865573666e-06, |
| "logits/chosen": -1.7167861461639404, |
| "logits/rejected": -1.6005874872207642, |
| "logps/chosen": -232.20309448242188, |
| "logps/rejected": -352.6765441894531, |
| "loss": 0.1321, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.9672311544418335, |
| "rewards/margins": 5.907352447509766, |
| "rewards/rejected": -7.874582767486572, |
| "step": 700 |
| }, |
| { |
| "epoch": 1.5034666666666667, |
| "grad_norm": 4.760759353637695, |
| "learning_rate": 2.620871930691058e-06, |
| "logits/chosen": -1.6677753925323486, |
| "logits/rejected": -1.6088497638702393, |
| "logps/chosen": -229.5364990234375, |
| "logps/rejected": -340.87298583984375, |
| "loss": 0.1338, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.956604242324829, |
| "rewards/margins": 6.195125579833984, |
| "rewards/rejected": -8.15173053741455, |
| "step": 705 |
| }, |
| { |
| "epoch": 1.5141333333333333, |
| "grad_norm": 5.057645797729492, |
| "learning_rate": 2.5921079658767152e-06, |
| "logits/chosen": -1.7027199268341064, |
| "logits/rejected": -1.6212679147720337, |
| "logps/chosen": -242.75393676757812, |
| "logps/rejected": -342.66180419921875, |
| "loss": 0.1214, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.713690996170044, |
| "rewards/margins": 6.617132663726807, |
| "rewards/rejected": -8.33082389831543, |
| "step": 710 |
| }, |
| { |
| "epoch": 1.5248, |
| "grad_norm": 4.691673755645752, |
| "learning_rate": 2.56333178578297e-06, |
| "logits/chosen": -1.7766993045806885, |
| "logits/rejected": -1.6785423755645752, |
| "logps/chosen": -224.8117218017578, |
| "logps/rejected": -356.70135498046875, |
| "loss": 0.1368, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.9061094522476196, |
| "rewards/margins": 5.97117805480957, |
| "rewards/rejected": -7.877286434173584, |
| "step": 715 |
| }, |
| { |
| "epoch": 1.5354666666666668, |
| "grad_norm": 2.651674270629883, |
| "learning_rate": 2.534547206682136e-06, |
| "logits/chosen": -1.7267783880233765, |
| "logits/rejected": -1.591758131980896, |
| "logps/chosen": -245.8301544189453, |
| "logps/rejected": -348.2117919921875, |
| "loss": 0.1385, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -2.353121042251587, |
| "rewards/margins": 5.891472816467285, |
| "rewards/rejected": -8.244592666625977, |
| "step": 720 |
| }, |
| { |
| "epoch": 1.5461333333333334, |
| "grad_norm": 3.751432418823242, |
| "learning_rate": 2.5057580459603947e-06, |
| "logits/chosen": -1.7399965524673462, |
| "logits/rejected": -1.636944055557251, |
| "logps/chosen": -217.2939453125, |
| "logps/rejected": -329.7176513671875, |
| "loss": 0.1016, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.9745547771453857, |
| "rewards/margins": 6.236304759979248, |
| "rewards/rejected": -8.210859298706055, |
| "step": 725 |
| }, |
| { |
| "epoch": 1.5568, |
| "grad_norm": 3.8177692890167236, |
| "learning_rate": 2.4769681216115384e-06, |
| "logits/chosen": -1.6969635486602783, |
| "logits/rejected": -1.6023454666137695, |
| "logps/chosen": -243.0435028076172, |
| "logps/rejected": -339.3236389160156, |
| "loss": 0.1492, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.8978859186172485, |
| "rewards/margins": 5.849517345428467, |
| "rewards/rejected": -7.747402191162109, |
| "step": 730 |
| }, |
| { |
| "epoch": 1.5674666666666668, |
| "grad_norm": 2.7171826362609863, |
| "learning_rate": 2.448181251730632e-06, |
| "logits/chosen": -1.7444900274276733, |
| "logits/rejected": -1.5893446207046509, |
| "logps/chosen": -228.6908416748047, |
| "logps/rejected": -332.8255615234375, |
| "loss": 0.1146, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.6262632608413696, |
| "rewards/margins": 6.102635383605957, |
| "rewards/rejected": -7.7288994789123535, |
| "step": 735 |
| }, |
| { |
| "epoch": 1.5781333333333334, |
| "grad_norm": 1.1198697090148926, |
| "learning_rate": 2.419401254007658e-06, |
| "logits/chosen": -1.7157386541366577, |
| "logits/rejected": -1.6218187808990479, |
| "logps/chosen": -251.6882781982422, |
| "logps/rejected": -319.733154296875, |
| "loss": 0.1369, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.6356916427612305, |
| "rewards/margins": 5.369327545166016, |
| "rewards/rejected": -7.005019187927246, |
| "step": 740 |
| }, |
| { |
| "epoch": 1.5888, |
| "grad_norm": 2.529115915298462, |
| "learning_rate": 2.390631945221221e-06, |
| "logits/chosen": -1.7237428426742554, |
| "logits/rejected": -1.600476861000061, |
| "logps/chosen": -245.54052734375, |
| "logps/rejected": -338.63897705078125, |
| "loss": 0.128, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.8269141912460327, |
| "rewards/margins": 5.957518577575684, |
| "rewards/rejected": -7.784433841705322, |
| "step": 745 |
| }, |
| { |
| "epoch": 1.5994666666666668, |
| "grad_norm": 3.8999643325805664, |
| "learning_rate": 2.361877140732366e-06, |
| "logits/chosen": -1.6685173511505127, |
| "logits/rejected": -1.5947197675704956, |
| "logps/chosen": -249.70657348632812, |
| "logps/rejected": -360.75921630859375, |
| "loss": 0.133, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.994818091392517, |
| "rewards/margins": 6.414160251617432, |
| "rewards/rejected": -8.408979415893555, |
| "step": 750 |
| }, |
| { |
| "epoch": 1.6101333333333332, |
| "grad_norm": 1.3885916471481323, |
| "learning_rate": 2.333140653978592e-06, |
| "logits/chosen": -1.7160930633544922, |
| "logits/rejected": -1.6322555541992188, |
| "logps/chosen": -229.70022583007812, |
| "logps/rejected": -342.3292541503906, |
| "loss": 0.1199, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -1.6185846328735352, |
| "rewards/margins": 6.081858158111572, |
| "rewards/rejected": -7.700442314147949, |
| "step": 755 |
| }, |
| { |
| "epoch": 1.6208, |
| "grad_norm": 1.6971677541732788, |
| "learning_rate": 2.3044262959681156e-06, |
| "logits/chosen": -1.6721397638320923, |
| "logits/rejected": -1.5602915287017822, |
| "logps/chosen": -228.8137969970703, |
| "logps/rejected": -322.86041259765625, |
| "loss": 0.1142, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.6498756408691406, |
| "rewards/margins": 5.735910892486572, |
| "rewards/rejected": -7.385787010192871, |
| "step": 760 |
| }, |
| { |
| "epoch": 1.6314666666666666, |
| "grad_norm": 3.690915107727051, |
| "learning_rate": 2.2757378747744575e-06, |
| "logits/chosen": -1.7795826196670532, |
| "logits/rejected": -1.5900989770889282, |
| "logps/chosen": -273.39202880859375, |
| "logps/rejected": -340.42633056640625, |
| "loss": 0.1149, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.4165130853652954, |
| "rewards/margins": 6.7276177406311035, |
| "rewards/rejected": -8.14413070678711, |
| "step": 765 |
| }, |
| { |
| "epoch": 1.6421333333333332, |
| "grad_norm": 3.8674254417419434, |
| "learning_rate": 2.24707919503142e-06, |
| "logits/chosen": -1.6882655620574951, |
| "logits/rejected": -1.6840887069702148, |
| "logps/chosen": -229.9908905029297, |
| "logps/rejected": -366.3896484375, |
| "loss": 0.1271, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -2.206510066986084, |
| "rewards/margins": 6.366598606109619, |
| "rewards/rejected": -8.573108673095703, |
| "step": 770 |
| }, |
| { |
| "epoch": 1.6528, |
| "grad_norm": 2.245314598083496, |
| "learning_rate": 2.2184540574285213e-06, |
| "logits/chosen": -1.626455545425415, |
| "logits/rejected": -1.5366604328155518, |
| "logps/chosen": -252.74417114257812, |
| "logps/rejected": -329.74664306640625, |
| "loss": 0.1131, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.4865179061889648, |
| "rewards/margins": 6.190736293792725, |
| "rewards/rejected": -7.677253723144531, |
| "step": 775 |
| }, |
| { |
| "epoch": 1.6634666666666666, |
| "grad_norm": 4.231085777282715, |
| "learning_rate": 2.1898662582069486e-06, |
| "logits/chosen": -1.7769323587417603, |
| "logits/rejected": -1.6287139654159546, |
| "logps/chosen": -236.2292022705078, |
| "logps/rejected": -332.585205078125, |
| "loss": 0.1527, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.8004322052001953, |
| "rewards/margins": 5.81653356552124, |
| "rewards/rejected": -7.616966247558594, |
| "step": 780 |
| }, |
| { |
| "epoch": 1.6741333333333333, |
| "grad_norm": 1.3001052141189575, |
| "learning_rate": 2.1613195886561044e-06, |
| "logits/chosen": -1.7131919860839844, |
| "logits/rejected": -1.6334612369537354, |
| "logps/chosen": -216.60595703125, |
| "logps/rejected": -348.07049560546875, |
| "loss": 0.0907, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -1.4661469459533691, |
| "rewards/margins": 6.375014781951904, |
| "rewards/rejected": -7.841162204742432, |
| "step": 785 |
| }, |
| { |
| "epoch": 1.6848, |
| "grad_norm": 3.4341273307800293, |
| "learning_rate": 2.132817834610811e-06, |
| "logits/chosen": -1.7027778625488281, |
| "logits/rejected": -1.5170468091964722, |
| "logps/chosen": -239.1169891357422, |
| "logps/rejected": -320.68255615234375, |
| "loss": 0.1244, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.5550756454467773, |
| "rewards/margins": 5.473415851593018, |
| "rewards/rejected": -7.028491020202637, |
| "step": 790 |
| }, |
| { |
| "epoch": 1.6954666666666667, |
| "grad_norm": 2.6593456268310547, |
| "learning_rate": 2.1043647759492338e-06, |
| "logits/chosen": -1.7209646701812744, |
| "logits/rejected": -1.5738383531570435, |
| "logps/chosen": -260.90423583984375, |
| "logps/rejected": -323.51531982421875, |
| "loss": 0.1114, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1319111585617065, |
| "rewards/margins": 5.710280418395996, |
| "rewards/rejected": -6.84219217300415, |
| "step": 795 |
| }, |
| { |
| "epoch": 1.7061333333333333, |
| "grad_norm": 4.085716247558594, |
| "learning_rate": 2.075964186091599e-06, |
| "logits/chosen": -1.768932580947876, |
| "logits/rejected": -1.6225475072860718, |
| "logps/chosen": -286.230712890625, |
| "logps/rejected": -332.88458251953125, |
| "loss": 0.1649, |
| "rewards/accuracies": 0.9312499761581421, |
| "rewards/chosen": -1.7526746988296509, |
| "rewards/margins": 5.637965202331543, |
| "rewards/rejected": -7.390639305114746, |
| "step": 800 |
| }, |
| { |
| "epoch": 1.7168, |
| "grad_norm": 3.297569513320923, |
| "learning_rate": 2.047619831499765e-06, |
| "logits/chosen": -1.633087158203125, |
| "logits/rejected": -1.5306732654571533, |
| "logps/chosen": -253.0421600341797, |
| "logps/rejected": -369.7610778808594, |
| "loss": 0.1162, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.5865943431854248, |
| "rewards/margins": 6.107217311859131, |
| "rewards/rejected": -7.693811893463135, |
| "step": 805 |
| }, |
| { |
| "epoch": 1.7274666666666667, |
| "grad_norm": 5.6189866065979, |
| "learning_rate": 2.0193354711777155e-06, |
| "logits/chosen": -1.793038010597229, |
| "logits/rejected": -1.600799560546875, |
| "logps/chosen": -233.15902709960938, |
| "logps/rejected": -322.8293762207031, |
| "loss": 0.1484, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.6604406833648682, |
| "rewards/margins": 5.723924160003662, |
| "rewards/rejected": -7.384364128112793, |
| "step": 810 |
| }, |
| { |
| "epoch": 1.7381333333333333, |
| "grad_norm": 5.962138652801514, |
| "learning_rate": 1.991114856173048e-06, |
| "logits/chosen": -1.7385631799697876, |
| "logits/rejected": -1.6375945806503296, |
| "logps/chosen": -235.4108428955078, |
| "logps/rejected": -348.2346496582031, |
| "loss": 0.1628, |
| "rewards/accuracies": 0.918749988079071, |
| "rewards/chosen": -2.211745500564575, |
| "rewards/margins": 5.744502067565918, |
| "rewards/rejected": -7.956246852874756, |
| "step": 815 |
| }, |
| { |
| "epoch": 1.7488000000000001, |
| "grad_norm": 3.5429904460906982, |
| "learning_rate": 1.9629617290795115e-06, |
| "logits/chosen": -1.720656394958496, |
| "logits/rejected": -1.5785796642303467, |
| "logps/chosen": -229.98727416992188, |
| "logps/rejected": -335.8892822265625, |
| "loss": 0.1678, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.8988584280014038, |
| "rewards/margins": 6.257178783416748, |
| "rewards/rejected": -8.156037330627441, |
| "step": 820 |
| }, |
| { |
| "epoch": 1.7594666666666665, |
| "grad_norm": 4.259081840515137, |
| "learning_rate": 1.934879823540663e-06, |
| "logits/chosen": -1.7053463459014893, |
| "logits/rejected": -1.6394329071044922, |
| "logps/chosen": -247.9888458251953, |
| "logps/rejected": -358.6705017089844, |
| "loss": 0.0996, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -1.9432106018066406, |
| "rewards/margins": 6.078979969024658, |
| "rewards/rejected": -8.022192001342773, |
| "step": 825 |
| }, |
| { |
| "epoch": 1.7701333333333333, |
| "grad_norm": 1.0865824222564697, |
| "learning_rate": 1.9068728637547195e-06, |
| "logits/chosen": -1.727698564529419, |
| "logits/rejected": -1.6279067993164062, |
| "logps/chosen": -244.95751953125, |
| "logps/rejected": -347.68206787109375, |
| "loss": 0.1208, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.9562523365020752, |
| "rewards/margins": 5.863669395446777, |
| "rewards/rejected": -7.81992244720459, |
| "step": 830 |
| }, |
| { |
| "epoch": 1.7808000000000002, |
| "grad_norm": 3.200122356414795, |
| "learning_rate": 1.878944563980658e-06, |
| "logits/chosen": -1.685605764389038, |
| "logits/rejected": -1.5968866348266602, |
| "logps/chosen": -233.25741577148438, |
| "logps/rejected": -340.8912658691406, |
| "loss": 0.1096, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.931910753250122, |
| "rewards/margins": 6.27100944519043, |
| "rewards/rejected": -8.202919006347656, |
| "step": 835 |
| }, |
| { |
| "epoch": 1.7914666666666665, |
| "grad_norm": 2.352064609527588, |
| "learning_rate": 1.85109862804563e-06, |
| "logits/chosen": -1.71112859249115, |
| "logits/rejected": -1.65338933467865, |
| "logps/chosen": -251.2657012939453, |
| "logps/rejected": -346.6195373535156, |
| "loss": 0.1354, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.7670342922210693, |
| "rewards/margins": 6.0953803062438965, |
| "rewards/rejected": -7.862415313720703, |
| "step": 840 |
| }, |
| { |
| "epoch": 1.8021333333333334, |
| "grad_norm": 3.149423599243164, |
| "learning_rate": 1.823338748853766e-06, |
| "logits/chosen": -1.730526328086853, |
| "logits/rejected": -1.6701030731201172, |
| "logps/chosen": -244.6688995361328, |
| "logps/rejected": -354.5580139160156, |
| "loss": 0.1045, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.046992063522339, |
| "rewards/margins": 6.459168910980225, |
| "rewards/rejected": -8.506160736083984, |
| "step": 845 |
| }, |
| { |
| "epoch": 1.8128, |
| "grad_norm": 3.0293586254119873, |
| "learning_rate": 1.7956686078964257e-06, |
| "logits/chosen": -1.8239294290542603, |
| "logits/rejected": -1.7257106304168701, |
| "logps/chosen": -219.5343475341797, |
| "logps/rejected": -334.0470275878906, |
| "loss": 0.0864, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.1547927856445312, |
| "rewards/margins": 6.76791524887085, |
| "rewards/rejected": -7.922706604003906, |
| "step": 850 |
| }, |
| { |
| "epoch": 1.8234666666666666, |
| "grad_norm": 1.0889246463775635, |
| "learning_rate": 1.7680918747639608e-06, |
| "logits/chosen": -1.7029361724853516, |
| "logits/rejected": -1.629794716835022, |
| "logps/chosen": -241.243408203125, |
| "logps/rejected": -357.905029296875, |
| "loss": 0.0971, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.594839334487915, |
| "rewards/margins": 6.8463897705078125, |
| "rewards/rejected": -8.441228866577148, |
| "step": 855 |
| }, |
| { |
| "epoch": 1.8341333333333334, |
| "grad_norm": 3.799083948135376, |
| "learning_rate": 1.740612206659057e-06, |
| "logits/chosen": -1.6510183811187744, |
| "logits/rejected": -1.6240144968032837, |
| "logps/chosen": -243.8262481689453, |
| "logps/rejected": -349.89569091796875, |
| "loss": 0.1311, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.8060070276260376, |
| "rewards/margins": 5.853053092956543, |
| "rewards/rejected": -7.659060001373291, |
| "step": 860 |
| }, |
| { |
| "epoch": 1.8448, |
| "grad_norm": 2.7306034564971924, |
| "learning_rate": 1.7132332479117204e-06, |
| "logits/chosen": -1.7491909265518188, |
| "logits/rejected": -1.6059234142303467, |
| "logps/chosen": -239.9043426513672, |
| "logps/rejected": -348.3580322265625, |
| "loss": 0.1165, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.591201901435852, |
| "rewards/margins": 6.327251434326172, |
| "rewards/rejected": -7.918453216552734, |
| "step": 865 |
| }, |
| { |
| "epoch": 1.8554666666666666, |
| "grad_norm": 2.9349820613861084, |
| "learning_rate": 1.6859586294959666e-06, |
| "logits/chosen": -1.7439231872558594, |
| "logits/rejected": -1.6497364044189453, |
| "logps/chosen": -235.9754638671875, |
| "logps/rejected": -345.97601318359375, |
| "loss": 0.1038, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.5834319591522217, |
| "rewards/margins": 6.660479545593262, |
| "rewards/rejected": -8.243912696838379, |
| "step": 870 |
| }, |
| { |
| "epoch": 1.8661333333333334, |
| "grad_norm": 0.882309079170227, |
| "learning_rate": 1.6587919685482884e-06, |
| "logits/chosen": -1.7273489236831665, |
| "logits/rejected": -1.6305004358291626, |
| "logps/chosen": -225.79464721679688, |
| "logps/rejected": -357.1151428222656, |
| "loss": 0.0636, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -1.861360788345337, |
| "rewards/margins": 6.6892828941345215, |
| "rewards/rejected": -8.550643920898438, |
| "step": 875 |
| }, |
| { |
| "epoch": 1.8768, |
| "grad_norm": 2.656388759613037, |
| "learning_rate": 1.6317368678879497e-06, |
| "logits/chosen": -1.6046922206878662, |
| "logits/rejected": -1.5097949504852295, |
| "logps/chosen": -230.23007202148438, |
| "logps/rejected": -341.9237365722656, |
| "loss": 0.1282, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.8171331882476807, |
| "rewards/margins": 6.508596897125244, |
| "rewards/rejected": -8.32573127746582, |
| "step": 880 |
| }, |
| { |
| "epoch": 1.8874666666666666, |
| "grad_norm": 1.9275213479995728, |
| "learning_rate": 1.6047969155391856e-06, |
| "logits/chosen": -1.6167736053466797, |
| "logits/rejected": -1.5170481204986572, |
| "logps/chosen": -273.21685791015625, |
| "logps/rejected": -357.8608093261719, |
| "loss": 0.0853, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -1.7052383422851562, |
| "rewards/margins": 6.317803859710693, |
| "rewards/rejected": -8.023042678833008, |
| "step": 885 |
| }, |
| { |
| "epoch": 1.8981333333333335, |
| "grad_norm": 4.2159342765808105, |
| "learning_rate": 1.5779756842553617e-06, |
| "logits/chosen": -1.6312297582626343, |
| "logits/rejected": -1.6414638757705688, |
| "logps/chosen": -265.60101318359375, |
| "logps/rejected": -363.58709716796875, |
| "loss": 0.113, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.922377586364746, |
| "rewards/margins": 6.6036553382873535, |
| "rewards/rejected": -8.526033401489258, |
| "step": 890 |
| }, |
| { |
| "epoch": 1.9088, |
| "grad_norm": 5.185214519500732, |
| "learning_rate": 1.551276731045157e-06, |
| "logits/chosen": -1.6335700750350952, |
| "logits/rejected": -1.6021028757095337, |
| "logps/chosen": -260.4001770019531, |
| "logps/rejected": -361.32861328125, |
| "loss": 0.1311, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.589746356010437, |
| "rewards/margins": 6.472695827484131, |
| "rewards/rejected": -8.062442779541016, |
| "step": 895 |
| }, |
| { |
| "epoch": 1.9194666666666667, |
| "grad_norm": 4.318290710449219, |
| "learning_rate": 1.5247035967008369e-06, |
| "logits/chosen": -1.7304105758666992, |
| "logits/rejected": -1.580280065536499, |
| "logps/chosen": -233.4041748046875, |
| "logps/rejected": -334.1758728027344, |
| "loss": 0.1128, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.5012531280517578, |
| "rewards/margins": 6.586421966552734, |
| "rewards/rejected": -8.087675094604492, |
| "step": 900 |
| }, |
| { |
| "epoch": 1.9301333333333335, |
| "grad_norm": 3.5980780124664307, |
| "learning_rate": 1.498259805328678e-06, |
| "logits/chosen": -1.761781930923462, |
| "logits/rejected": -1.5910484790802002, |
| "logps/chosen": -235.13815307617188, |
| "logps/rejected": -339.7412109375, |
| "loss": 0.1096, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -1.9504568576812744, |
| "rewards/margins": 6.570442199707031, |
| "rewards/rejected": -8.520899772644043, |
| "step": 905 |
| }, |
| { |
| "epoch": 1.9407999999999999, |
| "grad_norm": 5.879115104675293, |
| "learning_rate": 1.471948863881601e-06, |
| "logits/chosen": -1.723870038986206, |
| "logits/rejected": -1.5144500732421875, |
| "logps/chosen": -255.23355102539062, |
| "logps/rejected": -321.55352783203125, |
| "loss": 0.1381, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.5773289203643799, |
| "rewards/margins": 5.967000484466553, |
| "rewards/rejected": -7.544329643249512, |
| "step": 910 |
| }, |
| { |
| "epoch": 1.9514666666666667, |
| "grad_norm": 3.6410043239593506, |
| "learning_rate": 1.4457742616940864e-06, |
| "logits/chosen": -1.6668065786361694, |
| "logits/rejected": -1.5639894008636475, |
| "logps/chosen": -259.15362548828125, |
| "logps/rejected": -351.24542236328125, |
| "loss": 0.1139, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.8555443286895752, |
| "rewards/margins": 6.0399627685546875, |
| "rewards/rejected": -7.895506858825684, |
| "step": 915 |
| }, |
| { |
| "epoch": 1.9621333333333333, |
| "grad_norm": 3.0186233520507812, |
| "learning_rate": 1.4197394700194165e-06, |
| "logits/chosen": -1.755916953086853, |
| "logits/rejected": -1.6539405584335327, |
| "logps/chosen": -265.0111389160156, |
| "logps/rejected": -341.22918701171875, |
| "loss": 0.0748, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.075282573699951, |
| "rewards/margins": 6.287680149078369, |
| "rewards/rejected": -8.36296272277832, |
| "step": 920 |
| }, |
| { |
| "epoch": 1.9727999999999999, |
| "grad_norm": 2.5974314212799072, |
| "learning_rate": 1.3938479415693244e-06, |
| "logits/chosen": -1.7911674976348877, |
| "logits/rejected": -1.738271951675415, |
| "logps/chosen": -243.0534210205078, |
| "logps/rejected": -364.8657531738281, |
| "loss": 0.0822, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -1.5279881954193115, |
| "rewards/margins": 6.4855852127075195, |
| "rewards/rejected": -8.013572692871094, |
| "step": 925 |
| }, |
| { |
| "epoch": 1.9834666666666667, |
| "grad_norm": 4.313776016235352, |
| "learning_rate": 1.368103110056096e-06, |
| "logits/chosen": -1.6735479831695557, |
| "logits/rejected": -1.6453258991241455, |
| "logps/chosen": -229.39956665039062, |
| "logps/rejected": -348.58221435546875, |
| "loss": 0.0763, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -1.9789535999298096, |
| "rewards/margins": 7.005384922027588, |
| "rewards/rejected": -8.984338760375977, |
| "step": 930 |
| }, |
| { |
| "epoch": 1.9941333333333333, |
| "grad_norm": 2.7341182231903076, |
| "learning_rate": 1.3425083897371983e-06, |
| "logits/chosen": -1.6279489994049072, |
| "logits/rejected": -1.5178601741790771, |
| "logps/chosen": -255.06033325195312, |
| "logps/rejected": -350.7305603027344, |
| "loss": 0.1321, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -1.8026683330535889, |
| "rewards/margins": 6.296760559082031, |
| "rewards/rejected": -8.099428176879883, |
| "step": 935 |
| }, |
| { |
| "epoch": 2.0042666666666666, |
| "grad_norm": 1.8301348686218262, |
| "learning_rate": 1.317067174962478e-06, |
| "logits/chosen": -1.5837254524230957, |
| "logits/rejected": -1.5579044818878174, |
| "logps/chosen": -235.1605987548828, |
| "logps/rejected": -323.0276794433594, |
| "loss": 0.1251, |
| "rewards/accuracies": 0.9539473652839661, |
| "rewards/chosen": -2.1340339183807373, |
| "rewards/margins": 5.8628339767456055, |
| "rewards/rejected": -7.996868133544922, |
| "step": 940 |
| }, |
| { |
| "epoch": 2.0149333333333335, |
| "grad_norm": 1.6975414752960205, |
| "learning_rate": 1.291782839724009e-06, |
| "logits/chosen": -1.6794497966766357, |
| "logits/rejected": -1.6130616664886475, |
| "logps/chosen": -244.8574676513672, |
| "logps/rejected": -361.6617431640625, |
| "loss": 0.0732, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -1.9459739923477173, |
| "rewards/margins": 7.019693851470947, |
| "rewards/rejected": -8.965668678283691, |
| "step": 945 |
| }, |
| { |
| "epoch": 2.0256, |
| "grad_norm": 0.8381558656692505, |
| "learning_rate": 1.2666587372086352e-06, |
| "logits/chosen": -1.6198406219482422, |
| "logits/rejected": -1.5704479217529297, |
| "logps/chosen": -228.9700927734375, |
| "logps/rejected": -338.58782958984375, |
| "loss": 0.1104, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -1.4630517959594727, |
| "rewards/margins": 6.167015552520752, |
| "rewards/rejected": -7.630067348480225, |
| "step": 950 |
| }, |
| { |
| "epoch": 2.0362666666666667, |
| "grad_norm": 2.7961416244506836, |
| "learning_rate": 1.241698199353275e-06, |
| "logits/chosen": -1.7360506057739258, |
| "logits/rejected": -1.616646409034729, |
| "logps/chosen": -256.83477783203125, |
| "logps/rejected": -338.1868896484375, |
| "loss": 0.1565, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -2.1831955909729004, |
| "rewards/margins": 6.078934192657471, |
| "rewards/rejected": -8.262128829956055, |
| "step": 955 |
| }, |
| { |
| "epoch": 2.0469333333333335, |
| "grad_norm": 4.844041347503662, |
| "learning_rate": 1.2169045364030404e-06, |
| "logits/chosen": -1.7123730182647705, |
| "logits/rejected": -1.5840885639190674, |
| "logps/chosen": -279.18988037109375, |
| "logps/rejected": -340.1874694824219, |
| "loss": 0.116, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -1.7493995428085327, |
| "rewards/margins": 6.935929298400879, |
| "rewards/rejected": -8.68532943725586, |
| "step": 960 |
| }, |
| { |
| "epoch": 2.0576, |
| "grad_norm": 1.0539467334747314, |
| "learning_rate": 1.1922810364722402e-06, |
| "logits/chosen": -1.7058155536651611, |
| "logits/rejected": -1.5903913974761963, |
| "logps/chosen": -246.2003631591797, |
| "logps/rejected": -343.7231750488281, |
| "loss": 0.1024, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -1.8906090259552002, |
| "rewards/margins": 6.115625381469727, |
| "rewards/rejected": -8.006235122680664, |
| "step": 965 |
| }, |
| { |
| "epoch": 2.0682666666666667, |
| "grad_norm": 1.198693871498108, |
| "learning_rate": 1.1678309651083055e-06, |
| "logits/chosen": -1.7139508724212646, |
| "logits/rejected": -1.6323169469833374, |
| "logps/chosen": -233.64096069335938, |
| "logps/rejected": -358.6378479003906, |
| "loss": 0.0696, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.2217655181884766, |
| "rewards/margins": 6.9201250076293945, |
| "rewards/rejected": -9.141891479492188, |
| "step": 970 |
| }, |
| { |
| "epoch": 2.0789333333333335, |
| "grad_norm": 1.8502142429351807, |
| "learning_rate": 1.1435575648587224e-06, |
| "logits/chosen": -1.6855157613754272, |
| "logits/rejected": -1.6282001733779907, |
| "logps/chosen": -232.8811492919922, |
| "logps/rejected": -367.14984130859375, |
| "loss": 0.066, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -1.8243892192840576, |
| "rewards/margins": 7.305993556976318, |
| "rewards/rejected": -9.130383491516113, |
| "step": 975 |
| }, |
| { |
| "epoch": 2.0896, |
| "grad_norm": 6.383020877838135, |
| "learning_rate": 1.1194640548410039e-06, |
| "logits/chosen": -1.5834999084472656, |
| "logits/rejected": -1.4955960512161255, |
| "logps/chosen": -263.1512451171875, |
| "logps/rejected": -324.0372314453125, |
| "loss": 0.1725, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -2.092085123062134, |
| "rewards/margins": 5.891361236572266, |
| "rewards/rejected": -7.9834465980529785, |
| "step": 980 |
| }, |
| { |
| "epoch": 2.1002666666666667, |
| "grad_norm": 4.000116348266602, |
| "learning_rate": 1.095553630315774e-06, |
| "logits/chosen": -1.6087535619735718, |
| "logits/rejected": -1.558837890625, |
| "logps/chosen": -256.40777587890625, |
| "logps/rejected": -364.42132568359375, |
| "loss": 0.0999, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.2485008239746094, |
| "rewards/margins": 6.468873500823975, |
| "rewards/rejected": -8.717373847961426, |
| "step": 985 |
| }, |
| { |
| "epoch": 2.1109333333333336, |
| "grad_norm": 4.535819053649902, |
| "learning_rate": 1.0718294622630188e-06, |
| "logits/chosen": -1.7264890670776367, |
| "logits/rejected": -1.6320692300796509, |
| "logps/chosen": -239.48046875, |
| "logps/rejected": -359.3309020996094, |
| "loss": 0.1133, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -1.8677961826324463, |
| "rewards/margins": 6.378199577331543, |
| "rewards/rejected": -8.245996475219727, |
| "step": 990 |
| }, |
| { |
| "epoch": 2.1216, |
| "grad_norm": 2.3757009506225586, |
| "learning_rate": 1.0482946969615495e-06, |
| "logits/chosen": -1.726922631263733, |
| "logits/rejected": -1.6309759616851807, |
| "logps/chosen": -239.71511840820312, |
| "logps/rejected": -373.7091979980469, |
| "loss": 0.0649, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -2.3566739559173584, |
| "rewards/margins": 7.294585227966309, |
| "rewards/rejected": -9.651260375976562, |
| "step": 995 |
| }, |
| { |
| "epoch": 2.1322666666666668, |
| "grad_norm": 1.9476866722106934, |
| "learning_rate": 1.0249524555717461e-06, |
| "logits/chosen": -1.7482807636260986, |
| "logits/rejected": -1.6069862842559814, |
| "logps/chosen": -240.14382934570312, |
| "logps/rejected": -361.3963623046875, |
| "loss": 0.1027, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -1.955012559890747, |
| "rewards/margins": 7.11974573135376, |
| "rewards/rejected": -9.074758529663086, |
| "step": 1000 |
| }, |
| { |
| "epoch": 2.142933333333333, |
| "grad_norm": 2.6776695251464844, |
| "learning_rate": 1.0018058337216327e-06, |
| "logits/chosen": -1.621201753616333, |
| "logits/rejected": -1.5100598335266113, |
| "logps/chosen": -238.4880828857422, |
| "logps/rejected": -348.99945068359375, |
| "loss": 0.0706, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.1084201335906982, |
| "rewards/margins": 7.692069053649902, |
| "rewards/rejected": -9.80048942565918, |
| "step": 1005 |
| }, |
| { |
| "epoch": 2.1536, |
| "grad_norm": 2.4083611965179443, |
| "learning_rate": 9.788579010963387e-07, |
| "logits/chosen": -1.6992480754852295, |
| "logits/rejected": -1.6273291110992432, |
| "logps/chosen": -256.5484924316406, |
| "logps/rejected": -333.29315185546875, |
| "loss": 0.0934, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -1.9675493240356445, |
| "rewards/margins": 6.3934197425842285, |
| "rewards/rejected": -8.360968589782715, |
| "step": 1010 |
| }, |
| { |
| "epoch": 2.164266666666667, |
| "grad_norm": 4.460314750671387, |
| "learning_rate": 9.56111701030998e-07, |
| "logits/chosen": -1.7661964893341064, |
| "logits/rejected": -1.620145559310913, |
| "logps/chosen": -249.1160125732422, |
| "logps/rejected": -336.40936279296875, |
| "loss": 0.1163, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.3201756477355957, |
| "rewards/margins": 6.340085506439209, |
| "rewards/rejected": -8.660261154174805, |
| "step": 1015 |
| }, |
| { |
| "epoch": 2.174933333333333, |
| "grad_norm": 6.481249809265137, |
| "learning_rate": 9.335702501071459e-07, |
| "logits/chosen": -1.6359132528305054, |
| "logits/rejected": -1.5414669513702393, |
| "logps/chosen": -249.36630249023438, |
| "logps/rejected": -333.96807861328125, |
| "loss": 0.137, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -1.8785040378570557, |
| "rewards/margins": 5.66298770904541, |
| "rewards/rejected": -7.541491508483887, |
| "step": 1020 |
| }, |
| { |
| "epoch": 2.1856, |
| "grad_norm": 1.7318475246429443, |
| "learning_rate": 9.112365377526636e-07, |
| "logits/chosen": -1.7494544982910156, |
| "logits/rejected": -1.6931339502334595, |
| "logps/chosen": -219.39083862304688, |
| "logps/rejected": -379.05084228515625, |
| "loss": 0.06, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.0614876747131348, |
| "rewards/margins": 7.624110221862793, |
| "rewards/rejected": -9.685598373413086, |
| "step": 1025 |
| }, |
| { |
| "epoch": 2.196266666666667, |
| "grad_norm": 2.786151885986328, |
| "learning_rate": 8.891135258453195e-07, |
| "logits/chosen": -1.6171119213104248, |
| "logits/rejected": -1.507101058959961, |
| "logps/chosen": -243.9236297607422, |
| "logps/rejected": -336.0796813964844, |
| "loss": 0.0979, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.2703404426574707, |
| "rewards/margins": 6.305020809173584, |
| "rewards/rejected": -8.575361251831055, |
| "step": 1030 |
| }, |
| { |
| "epoch": 2.206933333333333, |
| "grad_norm": 2.109221935272217, |
| "learning_rate": 8.672041483199709e-07, |
| "logits/chosen": -1.6782341003417969, |
| "logits/rejected": -1.5760619640350342, |
| "logps/chosen": -240.6625518798828, |
| "logps/rejected": -367.6160888671875, |
| "loss": 0.0783, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.0386109352111816, |
| "rewards/margins": 7.059946537017822, |
| "rewards/rejected": -9.09855842590332, |
| "step": 1035 |
| }, |
| { |
| "epoch": 2.2176, |
| "grad_norm": 2.13687801361084, |
| "learning_rate": 8.455113107794652e-07, |
| "logits/chosen": -1.6757415533065796, |
| "logits/rejected": -1.6712865829467773, |
| "logps/chosen": -230.51266479492188, |
| "logps/rejected": -366.68133544921875, |
| "loss": 0.0587, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -2.3439688682556152, |
| "rewards/margins": 7.779986381530762, |
| "rewards/rejected": -10.123955726623535, |
| "step": 1040 |
| }, |
| { |
| "epoch": 2.228266666666667, |
| "grad_norm": 2.922100067138672, |
| "learning_rate": 8.240378901093035e-07, |
| "logits/chosen": -1.704439401626587, |
| "logits/rejected": -1.50922429561615, |
| "logps/chosen": -253.18704223632812, |
| "logps/rejected": -359.7574768066406, |
| "loss": 0.1019, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.250194549560547, |
| "rewards/margins": 7.344038486480713, |
| "rewards/rejected": -9.594232559204102, |
| "step": 1045 |
| }, |
| { |
| "epoch": 2.238933333333333, |
| "grad_norm": 2.7870895862579346, |
| "learning_rate": 8.027867340961113e-07, |
| "logits/chosen": -1.6217025518417358, |
| "logits/rejected": -1.53660249710083, |
| "logps/chosen": -238.7826690673828, |
| "logps/rejected": -347.23980712890625, |
| "loss": 0.0898, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.125499725341797, |
| "rewards/margins": 6.903764247894287, |
| "rewards/rejected": -9.029264450073242, |
| "step": 1050 |
| }, |
| { |
| "epoch": 2.2496, |
| "grad_norm": 2.5745162963867188, |
| "learning_rate": 7.817606610499639e-07, |
| "logits/chosen": -1.671149492263794, |
| "logits/rejected": -1.5366013050079346, |
| "logps/chosen": -253.76779174804688, |
| "logps/rejected": -356.57720947265625, |
| "loss": 0.0588, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -2.3465628623962402, |
| "rewards/margins": 7.49077844619751, |
| "rewards/rejected": -9.837340354919434, |
| "step": 1055 |
| }, |
| { |
| "epoch": 2.2602666666666664, |
| "grad_norm": 5.530954837799072, |
| "learning_rate": 7.609624594306278e-07, |
| "logits/chosen": -1.709287405014038, |
| "logits/rejected": -1.6433643102645874, |
| "logps/chosen": -254.7502899169922, |
| "logps/rejected": -364.1354675292969, |
| "loss": 0.0799, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.653001070022583, |
| "rewards/margins": 6.731104850769043, |
| "rewards/rejected": -9.384105682373047, |
| "step": 1060 |
| }, |
| { |
| "epoch": 2.2709333333333332, |
| "grad_norm": 2.934730291366577, |
| "learning_rate": 7.403948874777556e-07, |
| "logits/chosen": -1.6000057458877563, |
| "logits/rejected": -1.5496574640274048, |
| "logps/chosen": -253.9340057373047, |
| "logps/rejected": -345.54180908203125, |
| "loss": 0.0853, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.3913419246673584, |
| "rewards/margins": 6.566730499267578, |
| "rewards/rejected": -8.958072662353516, |
| "step": 1065 |
| }, |
| { |
| "epoch": 2.2816, |
| "grad_norm": 4.504528999328613, |
| "learning_rate": 7.20060672845091e-07, |
| "logits/chosen": -1.6582725048065186, |
| "logits/rejected": -1.6277124881744385, |
| "logps/chosen": -243.41665649414062, |
| "logps/rejected": -363.15008544921875, |
| "loss": 0.1113, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.7339158058166504, |
| "rewards/margins": 7.162438869476318, |
| "rewards/rejected": -9.896354675292969, |
| "step": 1070 |
| }, |
| { |
| "epoch": 2.2922666666666665, |
| "grad_norm": 3.5266401767730713, |
| "learning_rate": 6.99962512238733e-07, |
| "logits/chosen": -1.569681167602539, |
| "logits/rejected": -1.5099362134933472, |
| "logps/chosen": -243.1770782470703, |
| "logps/rejected": -348.07501220703125, |
| "loss": 0.1, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.601651906967163, |
| "rewards/margins": 6.440603733062744, |
| "rewards/rejected": -9.042256355285645, |
| "step": 1075 |
| }, |
| { |
| "epoch": 2.3029333333333333, |
| "grad_norm": 4.0208740234375, |
| "learning_rate": 6.801030710594949e-07, |
| "logits/chosen": -1.6699787378311157, |
| "logits/rejected": -1.5480045080184937, |
| "logps/chosen": -257.97100830078125, |
| "logps/rejected": -384.45721435546875, |
| "loss": 0.0927, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.272108793258667, |
| "rewards/margins": 7.204779624938965, |
| "rewards/rejected": -9.476888656616211, |
| "step": 1080 |
| }, |
| { |
| "epoch": 2.3136, |
| "grad_norm": 4.031130790710449, |
| "learning_rate": 6.604849830494264e-07, |
| "logits/chosen": -1.6949599981307983, |
| "logits/rejected": -1.6505162715911865, |
| "logps/chosen": -245.27963256835938, |
| "logps/rejected": -368.87347412109375, |
| "loss": 0.0983, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.8141303062438965, |
| "rewards/margins": 7.234206199645996, |
| "rewards/rejected": -10.04833698272705, |
| "step": 1085 |
| }, |
| { |
| "epoch": 2.3242666666666665, |
| "grad_norm": 3.7834320068359375, |
| "learning_rate": 6.411108499425253e-07, |
| "logits/chosen": -1.7620437145233154, |
| "logits/rejected": -1.5727145671844482, |
| "logps/chosen": -248.0930633544922, |
| "logps/rejected": -334.4790954589844, |
| "loss": 0.0838, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.3493759632110596, |
| "rewards/margins": 6.401058197021484, |
| "rewards/rejected": -8.750434875488281, |
| "step": 1090 |
| }, |
| { |
| "epoch": 2.3349333333333333, |
| "grad_norm": 3.1130168437957764, |
| "learning_rate": 6.219832411197013e-07, |
| "logits/chosen": -1.5150299072265625, |
| "logits/rejected": -1.459728479385376, |
| "logps/chosen": -247.00454711914062, |
| "logps/rejected": -355.4990539550781, |
| "loss": 0.114, |
| "rewards/accuracies": 0.956250011920929, |
| "rewards/chosen": -2.584754228591919, |
| "rewards/margins": 6.42316198348999, |
| "rewards/rejected": -9.007917404174805, |
| "step": 1095 |
| }, |
| { |
| "epoch": 2.3456, |
| "grad_norm": 1.063434362411499, |
| "learning_rate": 6.031046932680229e-07, |
| "logits/chosen": -1.684495210647583, |
| "logits/rejected": -1.6073039770126343, |
| "logps/chosen": -249.10623168945312, |
| "logps/rejected": -368.306884765625, |
| "loss": 0.0974, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.6043505668640137, |
| "rewards/margins": 7.219007968902588, |
| "rewards/rejected": -9.823358535766602, |
| "step": 1100 |
| }, |
| { |
| "epoch": 2.3562666666666665, |
| "grad_norm": 1.9829832315444946, |
| "learning_rate": 5.844777100443091e-07, |
| "logits/chosen": -1.6180429458618164, |
| "logits/rejected": -1.6057523488998413, |
| "logps/chosen": -250.98989868164062, |
| "logps/rejected": -369.36737060546875, |
| "loss": 0.0887, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.9293911457061768, |
| "rewards/margins": 6.682594299316406, |
| "rewards/rejected": -9.61198616027832, |
| "step": 1105 |
| }, |
| { |
| "epoch": 2.3669333333333333, |
| "grad_norm": 2.474097728729248, |
| "learning_rate": 5.66104761743092e-07, |
| "logits/chosen": -1.7592651844024658, |
| "logits/rejected": -1.7193400859832764, |
| "logps/chosen": -249.07107543945312, |
| "logps/rejected": -377.8904113769531, |
| "loss": 0.081, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -3.1483211517333984, |
| "rewards/margins": 7.270550727844238, |
| "rewards/rejected": -10.41887092590332, |
| "step": 1110 |
| }, |
| { |
| "epoch": 2.3776, |
| "grad_norm": 2.6713147163391113, |
| "learning_rate": 5.479882849690102e-07, |
| "logits/chosen": -1.6553032398223877, |
| "logits/rejected": -1.5625696182250977, |
| "logps/chosen": -241.57406616210938, |
| "logps/rejected": -369.3149108886719, |
| "loss": 0.0719, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.5870490074157715, |
| "rewards/margins": 7.614378452301025, |
| "rewards/rejected": -10.201428413391113, |
| "step": 1115 |
| }, |
| { |
| "epoch": 2.3882666666666665, |
| "grad_norm": 1.6205579042434692, |
| "learning_rate": 5.301306823136687e-07, |
| "logits/chosen": -1.5774857997894287, |
| "logits/rejected": -1.534125566482544, |
| "logps/chosen": -224.51943969726562, |
| "logps/rejected": -351.11956787109375, |
| "loss": 0.0673, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.335993528366089, |
| "rewards/margins": 7.358288764953613, |
| "rewards/rejected": -9.694281578063965, |
| "step": 1120 |
| }, |
| { |
| "epoch": 2.3989333333333334, |
| "grad_norm": 3.4620277881622314, |
| "learning_rate": 5.125343220370088e-07, |
| "logits/chosen": -1.7100372314453125, |
| "logits/rejected": -1.6061903238296509, |
| "logps/chosen": -233.5575714111328, |
| "logps/rejected": -362.2322082519531, |
| "loss": 0.0903, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.3436803817749023, |
| "rewards/margins": 7.102326393127441, |
| "rewards/rejected": -9.446006774902344, |
| "step": 1125 |
| }, |
| { |
| "epoch": 2.4096, |
| "grad_norm": 2.0486199855804443, |
| "learning_rate": 4.952015377532329e-07, |
| "logits/chosen": -1.6620476245880127, |
| "logits/rejected": -1.5756416320800781, |
| "logps/chosen": -240.8001708984375, |
| "logps/rejected": -382.3775329589844, |
| "loss": 0.0946, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.7751338481903076, |
| "rewards/margins": 7.866480827331543, |
| "rewards/rejected": -10.641613960266113, |
| "step": 1130 |
| }, |
| { |
| "epoch": 2.4202666666666666, |
| "grad_norm": 5.349671840667725, |
| "learning_rate": 4.781346281213204e-07, |
| "logits/chosen": -1.8110148906707764, |
| "logits/rejected": -1.6730849742889404, |
| "logps/chosen": -255.79312133789062, |
| "logps/rejected": -365.10797119140625, |
| "loss": 0.0957, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.783545970916748, |
| "rewards/margins": 7.280545711517334, |
| "rewards/rejected": -10.064092636108398, |
| "step": 1135 |
| }, |
| { |
| "epoch": 2.4309333333333334, |
| "grad_norm": 2.677881956100464, |
| "learning_rate": 4.613358565401818e-07, |
| "logits/chosen": -1.603020429611206, |
| "logits/rejected": -1.5301001071929932, |
| "logps/chosen": -259.9833068847656, |
| "logps/rejected": -365.3714599609375, |
| "loss": 0.0842, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.8131003379821777, |
| "rewards/margins": 6.671302795410156, |
| "rewards/rejected": -9.484403610229492, |
| "step": 1140 |
| }, |
| { |
| "epoch": 2.4416, |
| "grad_norm": 1.704085350036621, |
| "learning_rate": 4.448074508484906e-07, |
| "logits/chosen": -1.7008240222930908, |
| "logits/rejected": -1.6075012683868408, |
| "logps/chosen": -255.2787322998047, |
| "logps/rejected": -349.6066589355469, |
| "loss": 0.1014, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.5429394245147705, |
| "rewards/margins": 6.982499599456787, |
| "rewards/rejected": -9.525439262390137, |
| "step": 1145 |
| }, |
| { |
| "epoch": 2.4522666666666666, |
| "grad_norm": 3.8598265647888184, |
| "learning_rate": 4.2855160302922777e-07, |
| "logits/chosen": -1.7050069570541382, |
| "logits/rejected": -1.6236718893051147, |
| "logps/chosen": -230.6517333984375, |
| "logps/rejected": -369.06280517578125, |
| "loss": 0.0873, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.6450648307800293, |
| "rewards/margins": 7.3477582931518555, |
| "rewards/rejected": -9.992822647094727, |
| "step": 1150 |
| }, |
| { |
| "epoch": 2.4629333333333334, |
| "grad_norm": 1.6293611526489258, |
| "learning_rate": 4.125704689189819e-07, |
| "logits/chosen": -1.6456454992294312, |
| "logits/rejected": -1.4503004550933838, |
| "logps/chosen": -253.2928009033203, |
| "logps/rejected": -352.13055419921875, |
| "loss": 0.088, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.3692188262939453, |
| "rewards/margins": 6.8818864822387695, |
| "rewards/rejected": -9.251104354858398, |
| "step": 1155 |
| }, |
| { |
| "epoch": 2.4736000000000002, |
| "grad_norm": 1.0771186351776123, |
| "learning_rate": 3.9686616792204677e-07, |
| "logits/chosen": -1.6047093868255615, |
| "logits/rejected": -1.5732990503311157, |
| "logps/chosen": -258.16510009765625, |
| "logps/rejected": -393.47235107421875, |
| "loss": 0.0635, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.996021270751953, |
| "rewards/margins": 7.686588287353516, |
| "rewards/rejected": -10.682609558105469, |
| "step": 1160 |
| }, |
| { |
| "epoch": 2.4842666666666666, |
| "grad_norm": 1.822938323020935, |
| "learning_rate": 3.8144078272934676e-07, |
| "logits/chosen": -1.6440626382827759, |
| "logits/rejected": -1.5163719654083252, |
| "logps/chosen": -268.2124328613281, |
| "logps/rejected": -355.87786865234375, |
| "loss": 0.0713, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.7518844604492188, |
| "rewards/margins": 7.2615203857421875, |
| "rewards/rejected": -10.013406753540039, |
| "step": 1165 |
| }, |
| { |
| "epoch": 2.4949333333333334, |
| "grad_norm": 3.44150447845459, |
| "learning_rate": 3.6629635904222943e-07, |
| "logits/chosen": -1.6613085269927979, |
| "logits/rejected": -1.5570452213287354, |
| "logps/chosen": -239.2145233154297, |
| "logps/rejected": -363.83465576171875, |
| "loss": 0.0576, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -2.467777729034424, |
| "rewards/margins": 7.381143093109131, |
| "rewards/rejected": -9.848920822143555, |
| "step": 1170 |
| }, |
| { |
| "epoch": 2.5056000000000003, |
| "grad_norm": 1.4450147151947021, |
| "learning_rate": 3.514349053011698e-07, |
| "logits/chosen": -1.5924997329711914, |
| "logits/rejected": -1.510423183441162, |
| "logps/chosen": -245.48831176757812, |
| "logps/rejected": -368.66912841796875, |
| "loss": 0.0768, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.4841134548187256, |
| "rewards/margins": 7.180917263031006, |
| "rewards/rejected": -9.665030479431152, |
| "step": 1175 |
| }, |
| { |
| "epoch": 2.5162666666666667, |
| "grad_norm": 3.4693479537963867, |
| "learning_rate": 3.3685839241941133e-07, |
| "logits/chosen": -1.6568119525909424, |
| "logits/rejected": -1.596027135848999, |
| "logps/chosen": -246.05126953125, |
| "logps/rejected": -328.98858642578125, |
| "loss": 0.0848, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.629823684692383, |
| "rewards/margins": 6.541849613189697, |
| "rewards/rejected": -9.171671867370605, |
| "step": 1180 |
| }, |
| { |
| "epoch": 2.5269333333333335, |
| "grad_norm": 1.5991712808609009, |
| "learning_rate": 3.2256875352158435e-07, |
| "logits/chosen": -1.6681095361709595, |
| "logits/rejected": -1.5474411249160767, |
| "logps/chosen": -246.9559783935547, |
| "logps/rejected": -365.996337890625, |
| "loss": 0.0796, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.6049845218658447, |
| "rewards/margins": 7.331271171569824, |
| "rewards/rejected": -9.93625545501709, |
| "step": 1185 |
| }, |
| { |
| "epoch": 2.5376, |
| "grad_norm": 1.4303497076034546, |
| "learning_rate": 3.085678836873399e-07, |
| "logits/chosen": -1.7470356225967407, |
| "logits/rejected": -1.6877567768096924, |
| "logps/chosen": -237.14285278320312, |
| "logps/rejected": -378.49774169921875, |
| "loss": 0.0815, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.6633784770965576, |
| "rewards/margins": 7.212291717529297, |
| "rewards/rejected": -9.875669479370117, |
| "step": 1190 |
| }, |
| { |
| "epoch": 2.5482666666666667, |
| "grad_norm": 1.2953486442565918, |
| "learning_rate": 2.9485763970002106e-07, |
| "logits/chosen": -1.7139877080917358, |
| "logits/rejected": -1.6032788753509521, |
| "logps/chosen": -254.2081756591797, |
| "logps/rejected": -370.1174621582031, |
| "loss": 0.0741, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.655818223953247, |
| "rewards/margins": 6.725550651550293, |
| "rewards/rejected": -9.381368637084961, |
| "step": 1195 |
| }, |
| { |
| "epoch": 2.558933333333333, |
| "grad_norm": 2.0400030612945557, |
| "learning_rate": 2.8143983980042174e-07, |
| "logits/chosen": -1.7738536596298218, |
| "logits/rejected": -1.5883662700653076, |
| "logps/chosen": -240.4284210205078, |
| "logps/rejected": -364.11614990234375, |
| "loss": 0.0749, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.759413480758667, |
| "rewards/margins": 7.233453273773193, |
| "rewards/rejected": -9.992866516113281, |
| "step": 1200 |
| }, |
| { |
| "epoch": 2.5696, |
| "grad_norm": 2.236142635345459, |
| "learning_rate": 2.683162634456515e-07, |
| "logits/chosen": -1.6081100702285767, |
| "logits/rejected": -1.5720396041870117, |
| "logps/chosen": -266.195556640625, |
| "logps/rejected": -372.3501281738281, |
| "loss": 0.0692, |
| "rewards/accuracies": 1.0, |
| "rewards/chosen": -2.3031420707702637, |
| "rewards/margins": 7.687568664550781, |
| "rewards/rejected": -9.990710258483887, |
| "step": 1205 |
| }, |
| { |
| "epoch": 2.5802666666666667, |
| "grad_norm": 1.1345685720443726, |
| "learning_rate": 2.5548865107314606e-07, |
| "logits/chosen": -1.586785912513733, |
| "logits/rejected": -1.547318696975708, |
| "logps/chosen": -271.5514831542969, |
| "logps/rejected": -361.0550842285156, |
| "loss": 0.1134, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -2.704498767852783, |
| "rewards/margins": 7.07074499130249, |
| "rewards/rejected": -9.775243759155273, |
| "step": 1210 |
| }, |
| { |
| "epoch": 2.590933333333333, |
| "grad_norm": 2.8865694999694824, |
| "learning_rate": 2.429587038698511e-07, |
| "logits/chosen": -1.6625696420669556, |
| "logits/rejected": -1.5965254306793213, |
| "logps/chosen": -244.66842651367188, |
| "logps/rejected": -362.2137145996094, |
| "loss": 0.0739, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.790815830230713, |
| "rewards/margins": 7.540101528167725, |
| "rewards/rejected": -10.330916404724121, |
| "step": 1215 |
| }, |
| { |
| "epoch": 2.6016, |
| "grad_norm": 1.5729948282241821, |
| "learning_rate": 2.307280835466144e-07, |
| "logits/chosen": -1.6541128158569336, |
| "logits/rejected": -1.5849196910858154, |
| "logps/chosen": -237.2561492919922, |
| "logps/rejected": -365.4972839355469, |
| "loss": 0.0791, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.8733794689178467, |
| "rewards/margins": 7.663126468658447, |
| "rewards/rejected": -10.536505699157715, |
| "step": 1220 |
| }, |
| { |
| "epoch": 2.6122666666666667, |
| "grad_norm": 1.4792896509170532, |
| "learning_rate": 2.1879841211780856e-07, |
| "logits/chosen": -1.6012680530548096, |
| "logits/rejected": -1.5582574605941772, |
| "logps/chosen": -234.076416015625, |
| "logps/rejected": -345.57452392578125, |
| "loss": 0.065, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -2.5481960773468018, |
| "rewards/margins": 6.975489616394043, |
| "rewards/rejected": -9.523685455322266, |
| "step": 1225 |
| }, |
| { |
| "epoch": 2.622933333333333, |
| "grad_norm": 1.1267932653427124, |
| "learning_rate": 2.0717127168622202e-07, |
| "logits/chosen": -1.6668701171875, |
| "logits/rejected": -1.5733473300933838, |
| "logps/chosen": -257.2702331542969, |
| "logps/rejected": -378.37188720703125, |
| "loss": 0.1129, |
| "rewards/accuracies": 0.9624999761581421, |
| "rewards/chosen": -2.888535261154175, |
| "rewards/margins": 7.025588035583496, |
| "rewards/rejected": -9.91412353515625, |
| "step": 1230 |
| }, |
| { |
| "epoch": 2.6336, |
| "grad_norm": 1.5751965045928955, |
| "learning_rate": 1.958482042332424e-07, |
| "logits/chosen": -1.667828917503357, |
| "logits/rejected": -1.5087411403656006, |
| "logps/chosen": -256.654296875, |
| "logps/rejected": -373.72021484375, |
| "loss": 0.0712, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.67441463470459, |
| "rewards/margins": 7.628767490386963, |
| "rewards/rejected": -10.303182601928711, |
| "step": 1235 |
| }, |
| { |
| "epoch": 2.6442666666666668, |
| "grad_norm": 1.7919318675994873, |
| "learning_rate": 1.8483071141435937e-07, |
| "logits/chosen": -1.6052993535995483, |
| "logits/rejected": -1.5203253030776978, |
| "logps/chosen": -277.48358154296875, |
| "logps/rejected": -362.0965881347656, |
| "loss": 0.0974, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.6504619121551514, |
| "rewards/margins": 6.992439270019531, |
| "rewards/rejected": -9.642900466918945, |
| "step": 1240 |
| }, |
| { |
| "epoch": 2.654933333333333, |
| "grad_norm": 2.0555083751678467, |
| "learning_rate": 1.741202543600176e-07, |
| "logits/chosen": -1.6291580200195312, |
| "logits/rejected": -1.556952714920044, |
| "logps/chosen": -239.12606811523438, |
| "logps/rejected": -355.2477111816406, |
| "loss": 0.1169, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -3.143803358078003, |
| "rewards/margins": 7.427267551422119, |
| "rewards/rejected": -10.571070671081543, |
| "step": 1245 |
| }, |
| { |
| "epoch": 2.6656, |
| "grad_norm": 2.651470184326172, |
| "learning_rate": 1.6371825348184155e-07, |
| "logits/chosen": -1.6337629556655884, |
| "logits/rejected": -1.583034873008728, |
| "logps/chosen": -264.42333984375, |
| "logps/rejected": -360.38616943359375, |
| "loss": 0.0886, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.7354702949523926, |
| "rewards/margins": 7.126223087310791, |
| "rewards/rejected": -9.861692428588867, |
| "step": 1250 |
| }, |
| { |
| "epoch": 2.676266666666667, |
| "grad_norm": 2.948840618133545, |
| "learning_rate": 1.5362608828426246e-07, |
| "logits/chosen": -1.6709390878677368, |
| "logits/rejected": -1.5679157972335815, |
| "logps/chosen": -299.9473571777344, |
| "logps/rejected": -367.95245361328125, |
| "loss": 0.077, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.1444687843322754, |
| "rewards/margins": 7.233567714691162, |
| "rewards/rejected": -9.378034591674805, |
| "step": 1255 |
| }, |
| { |
| "epoch": 2.686933333333333, |
| "grad_norm": 2.0014069080352783, |
| "learning_rate": 1.438450971815686e-07, |
| "logits/chosen": -1.67729914188385, |
| "logits/rejected": -1.5980799198150635, |
| "logps/chosen": -247.76663208007812, |
| "logps/rejected": -371.108154296875, |
| "loss": 0.0757, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -3.236902952194214, |
| "rewards/margins": 7.365381717681885, |
| "rewards/rejected": -10.602285385131836, |
| "step": 1260 |
| }, |
| { |
| "epoch": 2.6976, |
| "grad_norm": 1.5738085508346558, |
| "learning_rate": 1.3437657732040783e-07, |
| "logits/chosen": -1.6811107397079468, |
| "logits/rejected": -1.5563424825668335, |
| "logps/chosen": -246.9635772705078, |
| "logps/rejected": -372.36859130859375, |
| "loss": 0.0737, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.4004998207092285, |
| "rewards/margins": 7.776610374450684, |
| "rewards/rejected": -10.177111625671387, |
| "step": 1265 |
| }, |
| { |
| "epoch": 2.708266666666667, |
| "grad_norm": 3.890855550765991, |
| "learning_rate": 1.2522178440776022e-07, |
| "logits/chosen": -1.6264982223510742, |
| "logits/rejected": -1.5898141860961914, |
| "logps/chosen": -266.61248779296875, |
| "logps/rejected": -345.47845458984375, |
| "loss": 0.0846, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -2.7661638259887695, |
| "rewards/margins": 6.510016441345215, |
| "rewards/rejected": -9.2761812210083, |
| "step": 1270 |
| }, |
| { |
| "epoch": 2.718933333333333, |
| "grad_norm": 1.8012325763702393, |
| "learning_rate": 1.1638193254440655e-07, |
| "logits/chosen": -1.6986353397369385, |
| "logits/rejected": -1.5819010734558105, |
| "logps/chosen": -250.3184051513672, |
| "logps/rejected": -348.66973876953125, |
| "loss": 0.0675, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -2.715459108352661, |
| "rewards/margins": 6.96172571182251, |
| "rewards/rejected": -9.67718505859375, |
| "step": 1275 |
| }, |
| { |
| "epoch": 2.7296, |
| "grad_norm": 2.982224464416504, |
| "learning_rate": 1.0785819406391562e-07, |
| "logits/chosen": -1.5906603336334229, |
| "logits/rejected": -1.5850645303726196, |
| "logps/chosen": -247.61328125, |
| "logps/rejected": -395.35711669921875, |
| "loss": 0.1005, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -2.7781596183776855, |
| "rewards/margins": 7.818112850189209, |
| "rewards/rejected": -10.596271514892578, |
| "step": 1280 |
| }, |
| { |
| "epoch": 2.740266666666667, |
| "grad_norm": 3.6533961296081543, |
| "learning_rate": 9.965169937717095e-08, |
| "logits/chosen": -1.6935789585113525, |
| "logits/rejected": -1.602374792098999, |
| "logps/chosen": -239.06887817382812, |
| "logps/rejected": -341.3331604003906, |
| "loss": 0.0929, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.9050512313842773, |
| "rewards/margins": 6.8136305809021, |
| "rewards/rejected": -9.718682289123535, |
| "step": 1285 |
| }, |
| { |
| "epoch": 2.7509333333333332, |
| "grad_norm": 2.6796557903289795, |
| "learning_rate": 9.176353682245676e-08, |
| "logits/chosen": -1.6797822713851929, |
| "logits/rejected": -1.6272222995758057, |
| "logps/chosen": -241.6938934326172, |
| "logps/rejected": -374.75958251953125, |
| "loss": 0.0998, |
| "rewards/accuracies": 0.9437500238418579, |
| "rewards/chosen": -2.4390995502471924, |
| "rewards/margins": 7.6869096755981445, |
| "rewards/rejected": -10.126008987426758, |
| "step": 1290 |
| }, |
| { |
| "epoch": 2.7616, |
| "grad_norm": 1.120842456817627, |
| "learning_rate": 8.41947525211223e-08, |
| "logits/chosen": -1.5870236158370972, |
| "logits/rejected": -1.5749309062957764, |
| "logps/chosen": -243.9224090576172, |
| "logps/rejected": -361.07000732421875, |
| "loss": 0.0925, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.4479503631591797, |
| "rewards/margins": 7.581417083740234, |
| "rewards/rejected": -10.029366493225098, |
| "step": 1295 |
| }, |
| { |
| "epoch": 2.772266666666667, |
| "grad_norm": 1.1996480226516724, |
| "learning_rate": 7.694635023884789e-08, |
| "logits/chosen": -1.5754233598709106, |
| "logits/rejected": -1.5047509670257568, |
| "logps/chosen": -256.6371765136719, |
| "logps/rejected": -357.0195007324219, |
| "loss": 0.0703, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -2.477647542953491, |
| "rewards/margins": 7.629544734954834, |
| "rewards/rejected": -10.107192039489746, |
| "step": 1300 |
| }, |
| { |
| "epoch": 2.7829333333333333, |
| "grad_norm": 0.44732823967933655, |
| "learning_rate": 7.00192912525241e-08, |
| "logits/chosen": -1.6828523874282837, |
| "logits/rejected": -1.6784083843231201, |
| "logps/chosen": -259.8722229003906, |
| "logps/rejected": -372.0727844238281, |
| "loss": 0.0545, |
| "rewards/accuracies": 0.9937499761581421, |
| "rewards/chosen": -2.719404697418213, |
| "rewards/margins": 7.491476535797119, |
| "rewards/rejected": -10.210882186889648, |
| "step": 1305 |
| }, |
| { |
| "epoch": 2.7936, |
| "grad_norm": 5.6885199546813965, |
| "learning_rate": 6.341449422277047e-08, |
| "logits/chosen": -1.6398189067840576, |
| "logits/rejected": -1.587944746017456, |
| "logps/chosen": -265.2816162109375, |
| "logps/rejected": -378.03228759765625, |
| "loss": 0.0913, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.6671059131622314, |
| "rewards/margins": 7.217464447021484, |
| "rewards/rejected": -9.884570121765137, |
| "step": 1310 |
| }, |
| { |
| "epoch": 2.804266666666667, |
| "grad_norm": 5.603551864624023, |
| "learning_rate": 5.7132835072101486e-08, |
| "logits/chosen": -1.769521713256836, |
| "logits/rejected": -1.6387608051300049, |
| "logps/chosen": -257.2008056640625, |
| "logps/rejected": -384.9828186035156, |
| "loss": 0.0941, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.8595147132873535, |
| "rewards/margins": 7.990152835845947, |
| "rewards/rejected": -10.8496675491333, |
| "step": 1315 |
| }, |
| { |
| "epoch": 2.8149333333333333, |
| "grad_norm": 2.971264362335205, |
| "learning_rate": 5.117514686876379e-08, |
| "logits/chosen": -1.5761768817901611, |
| "logits/rejected": -1.5605074167251587, |
| "logps/chosen": -253.6317138671875, |
| "logps/rejected": -371.7652587890625, |
| "loss": 0.1064, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.9549922943115234, |
| "rewards/margins": 7.519838809967041, |
| "rewards/rejected": -10.474831581115723, |
| "step": 1320 |
| }, |
| { |
| "epoch": 2.8256, |
| "grad_norm": 4.489383697509766, |
| "learning_rate": 4.5542219716255274e-08, |
| "logits/chosen": -1.6371389627456665, |
| "logits/rejected": -1.5999717712402344, |
| "logps/chosen": -253.1427459716797, |
| "logps/rejected": -372.0256042480469, |
| "loss": 0.0978, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.740422487258911, |
| "rewards/margins": 6.826735496520996, |
| "rewards/rejected": -9.567158699035645, |
| "step": 1325 |
| }, |
| { |
| "epoch": 2.836266666666667, |
| "grad_norm": 2.8886446952819824, |
| "learning_rate": 4.023480064854174e-08, |
| "logits/chosen": -1.7825405597686768, |
| "logits/rejected": -1.6278918981552124, |
| "logps/chosen": -247.08450317382812, |
| "logps/rejected": -344.13739013671875, |
| "loss": 0.0729, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.63508677482605, |
| "rewards/margins": 7.258586883544922, |
| "rewards/rejected": -9.893674850463867, |
| "step": 1330 |
| }, |
| { |
| "epoch": 2.8469333333333333, |
| "grad_norm": 2.435647964477539, |
| "learning_rate": 3.525359353098584e-08, |
| "logits/chosen": -1.6750576496124268, |
| "logits/rejected": -1.5368554592132568, |
| "logps/chosen": -243.14306640625, |
| "logps/rejected": -371.97296142578125, |
| "loss": 0.0935, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.809309482574463, |
| "rewards/margins": 7.868110656738281, |
| "rewards/rejected": -10.677419662475586, |
| "step": 1335 |
| }, |
| { |
| "epoch": 2.8576, |
| "grad_norm": 1.5942474603652954, |
| "learning_rate": 3.059925896700094e-08, |
| "logits/chosen": -1.5710580348968506, |
| "logits/rejected": -1.435702919960022, |
| "logps/chosen": -283.32379150390625, |
| "logps/rejected": -348.6867980957031, |
| "loss": 0.0785, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.5561439990997314, |
| "rewards/margins": 6.856298923492432, |
| "rewards/rejected": -9.412443161010742, |
| "step": 1340 |
| }, |
| { |
| "epoch": 2.8682666666666665, |
| "grad_norm": 2.419877529144287, |
| "learning_rate": 2.6272414210443964e-08, |
| "logits/chosen": -1.555304765701294, |
| "logits/rejected": -1.5223093032836914, |
| "logps/chosen": -255.93295288085938, |
| "logps/rejected": -367.8576354980469, |
| "loss": 0.1082, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.8896031379699707, |
| "rewards/margins": 7.200318813323975, |
| "rewards/rejected": -10.089921951293945, |
| "step": 1345 |
| }, |
| { |
| "epoch": 2.8789333333333333, |
| "grad_norm": 3.1487884521484375, |
| "learning_rate": 2.2273633083753632e-08, |
| "logits/chosen": -1.6507923603057861, |
| "logits/rejected": -1.558606505393982, |
| "logps/chosen": -251.55078125, |
| "logps/rejected": -367.0152893066406, |
| "loss": 0.0617, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -3.1786768436431885, |
| "rewards/margins": 7.703708648681641, |
| "rewards/rejected": -10.882387161254883, |
| "step": 1350 |
| }, |
| { |
| "epoch": 2.8895999999999997, |
| "grad_norm": 3.5375068187713623, |
| "learning_rate": 1.8603445901851647e-08, |
| "logits/chosen": -1.5756328105926514, |
| "logits/rejected": -1.562476396560669, |
| "logps/chosen": -259.0333557128906, |
| "logps/rejected": -377.33734130859375, |
| "loss": 0.1057, |
| "rewards/accuracies": 0.96875, |
| "rewards/chosen": -2.788533926010132, |
| "rewards/margins": 7.0015106201171875, |
| "rewards/rejected": -9.790044784545898, |
| "step": 1355 |
| }, |
| { |
| "epoch": 2.9002666666666665, |
| "grad_norm": 1.4896981716156006, |
| "learning_rate": 1.5262339401813375e-08, |
| "logits/chosen": -1.6661159992218018, |
| "logits/rejected": -1.6331936120986938, |
| "logps/chosen": -235.8148193359375, |
| "logps/rejected": -373.57183837890625, |
| "loss": 0.1032, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.9084439277648926, |
| "rewards/margins": 7.1836676597595215, |
| "rewards/rejected": -10.092111587524414, |
| "step": 1360 |
| }, |
| { |
| "epoch": 2.9109333333333334, |
| "grad_norm": 3.489358901977539, |
| "learning_rate": 1.2250756678315601e-08, |
| "logits/chosen": -1.667912483215332, |
| "logits/rejected": -1.5959059000015259, |
| "logps/chosen": -257.36187744140625, |
| "logps/rejected": -358.2672424316406, |
| "loss": 0.0731, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.9006903171539307, |
| "rewards/margins": 7.321658134460449, |
| "rewards/rejected": -10.222349166870117, |
| "step": 1365 |
| }, |
| { |
| "epoch": 2.9215999999999998, |
| "grad_norm": 2.710655450820923, |
| "learning_rate": 9.569097124875193e-09, |
| "logits/chosen": -1.6697155237197876, |
| "logits/rejected": -1.580635666847229, |
| "logps/chosen": -233.53555297851562, |
| "logps/rejected": -360.79888916015625, |
| "loss": 0.1151, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -2.5065906047821045, |
| "rewards/margins": 7.7552170753479, |
| "rewards/rejected": -10.261807441711426, |
| "step": 1370 |
| }, |
| { |
| "epoch": 2.9322666666666666, |
| "grad_norm": 3.9822990894317627, |
| "learning_rate": 7.217716380881479e-09, |
| "logits/chosen": -1.6743650436401367, |
| "logits/rejected": -1.585886001586914, |
| "logps/chosen": -264.65167236328125, |
| "logps/rejected": -376.71588134765625, |
| "loss": 0.0825, |
| "rewards/accuracies": 0.987500011920929, |
| "rewards/chosen": -3.1264894008636475, |
| "rewards/margins": 7.663410186767578, |
| "rewards/rejected": -10.789899826049805, |
| "step": 1375 |
| }, |
| { |
| "epoch": 2.9429333333333334, |
| "grad_norm": 2.6295809745788574, |
| "learning_rate": 5.196926284430359e-09, |
| "logits/chosen": -1.6660277843475342, |
| "logits/rejected": -1.558496356010437, |
| "logps/chosen": -252.54244995117188, |
| "logps/rejected": -352.0909118652344, |
| "loss": 0.1049, |
| "rewards/accuracies": 0.9750000238418579, |
| "rewards/chosen": -2.452617883682251, |
| "rewards/margins": 7.0967607498168945, |
| "rewards/rejected": -9.549378395080566, |
| "step": 1380 |
| }, |
| { |
| "epoch": 2.9536, |
| "grad_norm": 1.3974653482437134, |
| "learning_rate": 3.5069948309712775e-09, |
| "logits/chosen": -1.5303410291671753, |
| "logits/rejected": -1.5407088994979858, |
| "logps/chosen": -245.9228515625, |
| "logps/rejected": -395.0731201171875, |
| "loss": 0.0716, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.417079448699951, |
| "rewards/margins": 7.852843284606934, |
| "rewards/rejected": -10.269923210144043, |
| "step": 1385 |
| }, |
| { |
| "epoch": 2.9642666666666666, |
| "grad_norm": 2.3230302333831787, |
| "learning_rate": 2.1481461377634294e-09, |
| "logits/chosen": -1.5849785804748535, |
| "logits/rejected": -1.5197336673736572, |
| "logps/chosen": -256.17095947265625, |
| "logps/rejected": -375.9013671875, |
| "loss": 0.0673, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.8717899322509766, |
| "rewards/margins": 7.789064884185791, |
| "rewards/rejected": -10.660855293273926, |
| "step": 1390 |
| }, |
| { |
| "epoch": 2.9749333333333334, |
| "grad_norm": 4.476577281951904, |
| "learning_rate": 1.120560414154259e-09, |
| "logits/chosen": -1.595900297164917, |
| "logits/rejected": -1.4698235988616943, |
| "logps/chosen": -250.3801727294922, |
| "logps/rejected": -344.8222351074219, |
| "loss": 0.116, |
| "rewards/accuracies": 0.949999988079071, |
| "rewards/chosen": -2.773193836212158, |
| "rewards/margins": 7.142364501953125, |
| "rewards/rejected": -9.915558815002441, |
| "step": 1395 |
| }, |
| { |
| "epoch": 2.9856, |
| "grad_norm": 2.3721888065338135, |
| "learning_rate": 4.2437393768079983e-10, |
| "logits/chosen": -1.6179622411727905, |
| "logits/rejected": -1.557403326034546, |
| "logps/chosen": -276.74114990234375, |
| "logps/rejected": -369.22088623046875, |
| "loss": 0.1017, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.8413569927215576, |
| "rewards/margins": 7.1871490478515625, |
| "rewards/rejected": -10.028505325317383, |
| "step": 1400 |
| }, |
| { |
| "epoch": 2.9962666666666666, |
| "grad_norm": 2.667341470718384, |
| "learning_rate": 5.967903599635171e-11, |
| "logits/chosen": -1.5483349561691284, |
| "logits/rejected": -1.463370680809021, |
| "logps/chosen": -256.5998229980469, |
| "logps/rejected": -369.9665832519531, |
| "loss": 0.0884, |
| "rewards/accuracies": 0.981249988079071, |
| "rewards/chosen": -2.8231709003448486, |
| "rewards/margins": 7.097599983215332, |
| "rewards/rejected": -9.920769691467285, |
| "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 |
| } |
|
|
