| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1274, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007849293563579278, | |
| "grad_norm": 11.458238972118851, | |
| "learning_rate": 2.34375e-09, | |
| "logits/chosen": 5914.86328125, | |
| "logits/rejected": 2785.05078125, | |
| "logps/chosen": -212.39956665039062, | |
| "logps/rejected": -98.58319854736328, | |
| "loss": 1.6265, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007849293563579277, | |
| "grad_norm": 10.944765962770576, | |
| "learning_rate": 2.34375e-08, | |
| "logits/chosen": 4973.87255859375, | |
| "logits/rejected": 4328.1787109375, | |
| "logps/chosen": -204.2371826171875, | |
| "logps/rejected": -179.72877502441406, | |
| "loss": 1.6266, | |
| "rewards/accuracies": 0.48148149251937866, | |
| "rewards/chosen": -0.029761962592601776, | |
| "rewards/margins": 0.007892987690865993, | |
| "rewards/rejected": -0.03765495494008064, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.015698587127158554, | |
| "grad_norm": 11.499267080001312, | |
| "learning_rate": 4.6875e-08, | |
| "logits/chosen": 6084.16796875, | |
| "logits/rejected": 4834.16455078125, | |
| "logps/chosen": -217.2034149169922, | |
| "logps/rejected": -196.7104034423828, | |
| "loss": 1.6266, | |
| "rewards/accuracies": 0.46666669845581055, | |
| "rewards/chosen": -0.013780605979263783, | |
| "rewards/margins": -0.03623500093817711, | |
| "rewards/rejected": 0.022454392164945602, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.023547880690737835, | |
| "grad_norm": 9.932951594192556, | |
| "learning_rate": 7.03125e-08, | |
| "logits/chosen": 6083.6298828125, | |
| "logits/rejected": 5104.6298828125, | |
| "logps/chosen": -250.5565185546875, | |
| "logps/rejected": -209.35964965820312, | |
| "loss": 1.6264, | |
| "rewards/accuracies": 0.5249999761581421, | |
| "rewards/chosen": 0.02980937995016575, | |
| "rewards/margins": 0.05544011667370796, | |
| "rewards/rejected": -0.025630736723542213, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03139717425431711, | |
| "grad_norm": 10.560583055604155, | |
| "learning_rate": 9.375e-08, | |
| "logits/chosen": 5310.8251953125, | |
| "logits/rejected": 4346.38134765625, | |
| "logps/chosen": -212.0478973388672, | |
| "logps/rejected": -181.7023162841797, | |
| "loss": 1.6262, | |
| "rewards/accuracies": 0.5166667103767395, | |
| "rewards/chosen": 0.04025173932313919, | |
| "rewards/margins": 0.06556080281734467, | |
| "rewards/rejected": -0.025309067219495773, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03924646781789639, | |
| "grad_norm": 10.427628853427613, | |
| "learning_rate": 1.1718749999999999e-07, | |
| "logits/chosen": 6422.7939453125, | |
| "logits/rejected": 5040.93017578125, | |
| "logps/chosen": -265.3838195800781, | |
| "logps/rejected": -206.63735961914062, | |
| "loss": 1.6255, | |
| "rewards/accuracies": 0.5833333730697632, | |
| "rewards/chosen": 0.23301315307617188, | |
| "rewards/margins": 0.15109717845916748, | |
| "rewards/rejected": 0.0819159746170044, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.04709576138147567, | |
| "grad_norm": 9.622689233248208, | |
| "learning_rate": 1.40625e-07, | |
| "logits/chosen": 5486.65087890625, | |
| "logits/rejected": 4560.60400390625, | |
| "logps/chosen": -213.68582153320312, | |
| "logps/rejected": -208.76248168945312, | |
| "loss": 1.6246, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.2914460599422455, | |
| "rewards/margins": 0.19062557816505432, | |
| "rewards/rejected": 0.10082048177719116, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.054945054945054944, | |
| "grad_norm": 9.587956347814307, | |
| "learning_rate": 1.640625e-07, | |
| "logits/chosen": 5199.4384765625, | |
| "logits/rejected": 4924.6708984375, | |
| "logps/chosen": -177.8567657470703, | |
| "logps/rejected": -176.4372100830078, | |
| "loss": 1.6231, | |
| "rewards/accuracies": 0.6250000596046448, | |
| "rewards/chosen": 0.6440274715423584, | |
| "rewards/margins": 0.3180614411830902, | |
| "rewards/rejected": 0.3259660303592682, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.06279434850863422, | |
| "grad_norm": 8.592067444398902, | |
| "learning_rate": 1.875e-07, | |
| "logits/chosen": 5778.53955078125, | |
| "logits/rejected": 5285.609375, | |
| "logps/chosen": -195.7733917236328, | |
| "logps/rejected": -181.14984130859375, | |
| "loss": 1.6212, | |
| "rewards/accuracies": 0.5666666626930237, | |
| "rewards/chosen": 0.8041629791259766, | |
| "rewards/margins": 0.39861544966697693, | |
| "rewards/rejected": 0.40554744005203247, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0706436420722135, | |
| "grad_norm": 10.124908249937006, | |
| "learning_rate": 2.109375e-07, | |
| "logits/chosen": 6056.23486328125, | |
| "logits/rejected": 5209.0205078125, | |
| "logps/chosen": -217.357421875, | |
| "logps/rejected": -184.17950439453125, | |
| "loss": 1.6184, | |
| "rewards/accuracies": 0.7000000476837158, | |
| "rewards/chosen": 1.5040231943130493, | |
| "rewards/margins": 1.2358932495117188, | |
| "rewards/rejected": 0.2681299149990082, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.07849293563579278, | |
| "grad_norm": 9.637444838737146, | |
| "learning_rate": 2.3437499999999998e-07, | |
| "logits/chosen": 5975.40087890625, | |
| "logits/rejected": 5763.3486328125, | |
| "logps/chosen": -208.479248046875, | |
| "logps/rejected": -198.80589294433594, | |
| "loss": 1.6183, | |
| "rewards/accuracies": 0.5916666388511658, | |
| "rewards/chosen": 1.752227544784546, | |
| "rewards/margins": 1.1407288312911987, | |
| "rewards/rejected": 0.6114988327026367, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08634222919937205, | |
| "grad_norm": 9.016250169124682, | |
| "learning_rate": 2.5781249999999997e-07, | |
| "logits/chosen": 6438.27734375, | |
| "logits/rejected": 5277.41748046875, | |
| "logps/chosen": -180.60360717773438, | |
| "logps/rejected": -178.02499389648438, | |
| "loss": 1.6146, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": 2.4966413974761963, | |
| "rewards/margins": 1.597015619277954, | |
| "rewards/rejected": 0.8996255993843079, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.09419152276295134, | |
| "grad_norm": 10.2875643972245, | |
| "learning_rate": 2.8125e-07, | |
| "logits/chosen": 6324.859375, | |
| "logits/rejected": 5255.02197265625, | |
| "logps/chosen": -200.77735900878906, | |
| "logps/rejected": -197.18515014648438, | |
| "loss": 1.6142, | |
| "rewards/accuracies": 0.5916667580604553, | |
| "rewards/chosen": 2.0967907905578613, | |
| "rewards/margins": 1.1320407390594482, | |
| "rewards/rejected": 0.9647501111030579, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.10204081632653061, | |
| "grad_norm": 8.720197352663515, | |
| "learning_rate": 2.999977454977299e-07, | |
| "logits/chosen": 6224.30419921875, | |
| "logits/rejected": 5787.24609375, | |
| "logps/chosen": -206.4401397705078, | |
| "logps/rejected": -196.00332641601562, | |
| "loss": 1.6121, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": 3.3742451667785645, | |
| "rewards/margins": 1.7132713794708252, | |
| "rewards/rejected": 1.6609739065170288, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.10989010989010989, | |
| "grad_norm": 9.129566422771084, | |
| "learning_rate": 2.999188450339425e-07, | |
| "logits/chosen": 6075.42626953125, | |
| "logits/rejected": 5363.8330078125, | |
| "logps/chosen": -182.21517944335938, | |
| "logps/rejected": -175.36424255371094, | |
| "loss": 1.6087, | |
| "rewards/accuracies": 0.6250000596046448, | |
| "rewards/chosen": 3.4619534015655518, | |
| "rewards/margins": 1.249176263809204, | |
| "rewards/rejected": 2.2127773761749268, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.11773940345368916, | |
| "grad_norm": 11.857486318606039, | |
| "learning_rate": 2.997272872179022e-07, | |
| "logits/chosen": 6205.095703125, | |
| "logits/rejected": 5208.36376953125, | |
| "logps/chosen": -198.15518188476562, | |
| "logps/rejected": -171.08432006835938, | |
| "loss": 1.6098, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 3.4637317657470703, | |
| "rewards/margins": 2.2425143718719482, | |
| "rewards/rejected": 1.2212175130844116, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12558869701726844, | |
| "grad_norm": 8.968030441205661, | |
| "learning_rate": 2.994232159967983e-07, | |
| "logits/chosen": 6122.28564453125, | |
| "logits/rejected": 5471.36572265625, | |
| "logps/chosen": -194.1709747314453, | |
| "logps/rejected": -195.9803009033203, | |
| "loss": 1.609, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 3.3317999839782715, | |
| "rewards/margins": 2.249804973602295, | |
| "rewards/rejected": 1.0819952487945557, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.13343799058084774, | |
| "grad_norm": 9.653645250907548, | |
| "learning_rate": 2.990068598666458e-07, | |
| "logits/chosen": 6370.8193359375, | |
| "logits/rejected": 5675.42236328125, | |
| "logps/chosen": -206.18612670898438, | |
| "logps/rejected": -194.17982482910156, | |
| "loss": 1.608, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 3.4253058433532715, | |
| "rewards/margins": 1.2780392169952393, | |
| "rewards/rejected": 2.1472671031951904, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.141287284144427, | |
| "grad_norm": 9.454673895277375, | |
| "learning_rate": 2.984785317005808e-07, | |
| "logits/chosen": 5601.3544921875, | |
| "logits/rejected": 4895.4755859375, | |
| "logps/chosen": -185.2515411376953, | |
| "logps/rejected": -175.55259704589844, | |
| "loss": 1.6091, | |
| "rewards/accuracies": 0.5750000476837158, | |
| "rewards/chosen": 3.3107008934020996, | |
| "rewards/margins": 0.6890397071838379, | |
| "rewards/rejected": 2.6216609477996826, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.14913657770800628, | |
| "grad_norm": 9.654359692410358, | |
| "learning_rate": 2.978386285137506e-07, | |
| "logits/chosen": 5670.37841796875, | |
| "logits/rejected": 4872.609375, | |
| "logps/chosen": -199.90350341796875, | |
| "logps/rejected": -190.27354431152344, | |
| "loss": 1.6056, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 3.8663978576660156, | |
| "rewards/margins": 2.9279844760894775, | |
| "rewards/rejected": 0.9384133219718933, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.15698587127158556, | |
| "grad_norm": 10.197121119934124, | |
| "learning_rate": 2.9708763116497365e-07, | |
| "logits/chosen": 5597.7919921875, | |
| "logits/rejected": 5161.28271484375, | |
| "logps/chosen": -179.18333435058594, | |
| "logps/rejected": -195.28829956054688, | |
| "loss": 1.6087, | |
| "rewards/accuracies": 0.6166666746139526, | |
| "rewards/chosen": 3.3536689281463623, | |
| "rewards/margins": 1.9326164722442627, | |
| "rewards/rejected": 1.42105233669281, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.16483516483516483, | |
| "grad_norm": 9.507119152719666, | |
| "learning_rate": 2.962261039953964e-07, | |
| "logits/chosen": 5043.921875, | |
| "logits/rejected": 4512.1884765625, | |
| "logps/chosen": -150.83897399902344, | |
| "logps/rejected": -155.26417541503906, | |
| "loss": 1.6066, | |
| "rewards/accuracies": 0.5333333611488342, | |
| "rewards/chosen": 3.090742826461792, | |
| "rewards/margins": 1.139352798461914, | |
| "rewards/rejected": 1.951390027999878, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.1726844583987441, | |
| "grad_norm": 9.378684614749979, | |
| "learning_rate": 2.952546944044155e-07, | |
| "logits/chosen": 6175.8935546875, | |
| "logits/rejected": 5279.80224609375, | |
| "logps/chosen": -192.4020233154297, | |
| "logps/rejected": -162.25759887695312, | |
| "loss": 1.6063, | |
| "rewards/accuracies": 0.6833333373069763, | |
| "rewards/chosen": 3.872145175933838, | |
| "rewards/margins": 2.17866587638855, | |
| "rewards/rejected": 1.6934791803359985, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.18053375196232338, | |
| "grad_norm": 9.79164707530873, | |
| "learning_rate": 2.941741323631859e-07, | |
| "logits/chosen": 5731.53564453125, | |
| "logits/rejected": 5152.2685546875, | |
| "logps/chosen": -183.71131896972656, | |
| "logps/rejected": -178.02699279785156, | |
| "loss": 1.6063, | |
| "rewards/accuracies": 0.6416666507720947, | |
| "rewards/chosen": 3.631937026977539, | |
| "rewards/margins": 1.891611099243164, | |
| "rewards/rejected": 1.740325927734375, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.18838304552590268, | |
| "grad_norm": 9.812812391985778, | |
| "learning_rate": 2.929852298660803e-07, | |
| "logits/chosen": 5198.8896484375, | |
| "logits/rejected": 4139.96923828125, | |
| "logps/chosen": -163.8915252685547, | |
| "logps/rejected": -153.38279724121094, | |
| "loss": 1.6049, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 3.735271453857422, | |
| "rewards/margins": 2.3994784355163574, | |
| "rewards/rejected": 1.335792899131775, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.19623233908948196, | |
| "grad_norm": 9.413138412548989, | |
| "learning_rate": 2.9168888032051097e-07, | |
| "logits/chosen": 6045.21923828125, | |
| "logits/rejected": 5354.9951171875, | |
| "logps/chosen": -163.71151733398438, | |
| "logps/rejected": -183.2251434326172, | |
| "loss": 1.6076, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": 3.4429244995117188, | |
| "rewards/margins": 2.1717002391815186, | |
| "rewards/rejected": 1.2712242603302002, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.20408163265306123, | |
| "grad_norm": 11.02606597842409, | |
| "learning_rate": 2.9028605787557423e-07, | |
| "logits/chosen": 6197.080078125, | |
| "logits/rejected": 5034.4609375, | |
| "logps/chosen": -195.53811645507812, | |
| "logps/rejected": -164.32293701171875, | |
| "loss": 1.6062, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 4.181486129760742, | |
| "rewards/margins": 2.2564964294433594, | |
| "rewards/rejected": 1.9249897003173828, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.2119309262166405, | |
| "grad_norm": 8.786474619717062, | |
| "learning_rate": 2.8877781669002034e-07, | |
| "logits/chosen": 6150.35302734375, | |
| "logits/rejected": 5960.28759765625, | |
| "logps/chosen": -192.77615356445312, | |
| "logps/rejected": -185.39297485351562, | |
| "loss": 1.6043, | |
| "rewards/accuracies": 0.5583333373069763, | |
| "rewards/chosen": 3.9804999828338623, | |
| "rewards/margins": 0.6723251938819885, | |
| "rewards/rejected": 3.3081748485565186, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.21978021978021978, | |
| "grad_norm": 11.186569406268537, | |
| "learning_rate": 2.871652901401002e-07, | |
| "logits/chosen": 5263.07763671875, | |
| "logits/rejected": 4574.97119140625, | |
| "logps/chosen": -159.4766082763672, | |
| "logps/rejected": -168.32589721679688, | |
| "loss": 1.6041, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 3.5813148021698, | |
| "rewards/margins": 2.5793395042419434, | |
| "rewards/rejected": 1.0019747018814087, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.22762951334379905, | |
| "grad_norm": 10.641068152714297, | |
| "learning_rate": 2.854496899678834e-07, | |
| "logits/chosen": 7268.1611328125, | |
| "logits/rejected": 5389.25732421875, | |
| "logps/chosen": -233.04153442382812, | |
| "logps/rejected": -181.015625, | |
| "loss": 1.606, | |
| "rewards/accuracies": 0.6583333611488342, | |
| "rewards/chosen": 4.031437397003174, | |
| "rewards/margins": 2.5249290466308594, | |
| "rewards/rejected": 1.506508469581604, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.23547880690737832, | |
| "grad_norm": 9.89515741424169, | |
| "learning_rate": 2.8363230537068793e-07, | |
| "logits/chosen": 5273.17578125, | |
| "logits/rejected": 5249.6083984375, | |
| "logps/chosen": -164.12420654296875, | |
| "logps/rejected": -168.3370819091797, | |
| "loss": 1.6072, | |
| "rewards/accuracies": 0.6166666746139526, | |
| "rewards/chosen": 3.3666069507598877, | |
| "rewards/margins": 1.4807441234588623, | |
| "rewards/rejected": 1.8858625888824463, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.24332810047095763, | |
| "grad_norm": 9.156189036074041, | |
| "learning_rate": 2.817145020323055e-07, | |
| "logits/chosen": 5941.7890625, | |
| "logits/rejected": 5263.54736328125, | |
| "logps/chosen": -173.28579711914062, | |
| "logps/rejected": -173.98643493652344, | |
| "loss": 1.6053, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": 4.284714698791504, | |
| "rewards/margins": 2.821631908416748, | |
| "rewards/rejected": 1.4630826711654663, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.25117739403453687, | |
| "grad_norm": 10.434068381012322, | |
| "learning_rate": 2.7969772109675106e-07, | |
| "logits/chosen": 5937.4228515625, | |
| "logits/rejected": 5140.39111328125, | |
| "logps/chosen": -197.049560546875, | |
| "logps/rejected": -200.171875, | |
| "loss": 1.6065, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 4.066697597503662, | |
| "rewards/margins": 2.1806929111480713, | |
| "rewards/rejected": 1.8860050439834595, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.25902668759811615, | |
| "grad_norm": 7.96980516077302, | |
| "learning_rate": 2.775834780853069e-07, | |
| "logits/chosen": 5770.4892578125, | |
| "logits/rejected": 5251.03076171875, | |
| "logps/chosen": -190.3749542236328, | |
| "logps/rejected": -175.70028686523438, | |
| "loss": 1.6057, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": 4.041922092437744, | |
| "rewards/margins": 2.4098730087280273, | |
| "rewards/rejected": 1.6320489645004272, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2668759811616955, | |
| "grad_norm": 9.327241211208603, | |
| "learning_rate": 2.75373361757676e-07, | |
| "logits/chosen": 6020.267578125, | |
| "logits/rejected": 6114.4892578125, | |
| "logps/chosen": -189.46829223632812, | |
| "logps/rejected": -218.3720703125, | |
| "loss": 1.6054, | |
| "rewards/accuracies": 0.6416666507720947, | |
| "rewards/chosen": 4.399914741516113, | |
| "rewards/margins": 2.4120266437530518, | |
| "rewards/rejected": 1.9878880977630615, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.27472527472527475, | |
| "grad_norm": 9.836682323090738, | |
| "learning_rate": 2.730690329180992e-07, | |
| "logits/chosen": 6336.00341796875, | |
| "logits/rejected": 5750.68310546875, | |
| "logps/chosen": -190.63604736328125, | |
| "logps/rejected": -205.4300537109375, | |
| "loss": 1.6049, | |
| "rewards/accuracies": 0.6833332777023315, | |
| "rewards/chosen": 4.227248191833496, | |
| "rewards/margins": 2.878089189529419, | |
| "rewards/rejected": 1.3491594791412354, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.282574568288854, | |
| "grad_norm": 10.19922545005082, | |
| "learning_rate": 2.7067222316733527e-07, | |
| "logits/chosen": 5833.3701171875, | |
| "logits/rejected": 5751.5283203125, | |
| "logps/chosen": -175.12269592285156, | |
| "logps/rejected": -198.1387481689453, | |
| "loss": 1.6072, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 2.7430808544158936, | |
| "rewards/margins": 1.5406630039215088, | |
| "rewards/rejected": 1.2024176120758057, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2904238618524333, | |
| "grad_norm": 10.34523599778714, | |
| "learning_rate": 2.6818473360143916e-07, | |
| "logits/chosen": 5338.24755859375, | |
| "logits/rejected": 5045.15087890625, | |
| "logps/chosen": -180.8574676513672, | |
| "logps/rejected": -166.79676818847656, | |
| "loss": 1.6067, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": 3.5679619312286377, | |
| "rewards/margins": 2.2969326972961426, | |
| "rewards/rejected": 1.271028995513916, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.29827315541601257, | |
| "grad_norm": 9.372139727854044, | |
| "learning_rate": 2.6560843345831893e-07, | |
| "logits/chosen": 6382.61181640625, | |
| "logits/rejected": 5298.80224609375, | |
| "logps/chosen": -223.8726348876953, | |
| "logps/rejected": -181.1964874267578, | |
| "loss": 1.6059, | |
| "rewards/accuracies": 0.6916666626930237, | |
| "rewards/chosen": 4.494563579559326, | |
| "rewards/margins": 2.7286934852600098, | |
| "rewards/rejected": 1.7658706903457642, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.30612244897959184, | |
| "grad_norm": 10.887020734305196, | |
| "learning_rate": 2.62945258713086e-07, | |
| "logits/chosen": 6511.5537109375, | |
| "logits/rejected": 4989.4306640625, | |
| "logps/chosen": -211.2510986328125, | |
| "logps/rejected": -179.0139923095703, | |
| "loss": 1.6059, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": 4.222726345062256, | |
| "rewards/margins": 3.4083054065704346, | |
| "rewards/rejected": 0.8144209980964661, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.3139717425431711, | |
| "grad_norm": 10.72240359029845, | |
| "learning_rate": 2.601972106232561e-07, | |
| "logits/chosen": 5603.4267578125, | |
| "logits/rejected": 5503.55078125, | |
| "logps/chosen": -187.75234985351562, | |
| "logps/rejected": -187.8725128173828, | |
| "loss": 1.6064, | |
| "rewards/accuracies": 0.6666666269302368, | |
| "rewards/chosen": 3.06905460357666, | |
| "rewards/margins": 2.2018256187438965, | |
| "rewards/rejected": 0.8672292828559875, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.3218210361067504, | |
| "grad_norm": 11.25468877955665, | |
| "learning_rate": 2.5736635422489343e-07, | |
| "logits/chosen": 6066.5107421875, | |
| "logits/rejected": 5429.3359375, | |
| "logps/chosen": -208.7128143310547, | |
| "logps/rejected": -207.3098907470703, | |
| "loss": 1.6046, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 3.760521650314331, | |
| "rewards/margins": 2.6000523567199707, | |
| "rewards/rejected": 1.1604695320129395, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.32967032967032966, | |
| "grad_norm": 11.575393675092053, | |
| "learning_rate": 2.5445481678082706e-07, | |
| "logits/chosen": 5705.2744140625, | |
| "logits/rejected": 5208.12353515625, | |
| "logps/chosen": -200.07925415039062, | |
| "logps/rejected": -180.0633544921875, | |
| "loss": 1.6052, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": 3.9673619270324707, | |
| "rewards/margins": 2.736542224884033, | |
| "rewards/rejected": 1.2308193445205688, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.33751962323390894, | |
| "grad_norm": 11.347255741193797, | |
| "learning_rate": 2.514647861821081e-07, | |
| "logits/chosen": 5512.1982421875, | |
| "logits/rejected": 4594.4384765625, | |
| "logps/chosen": -176.08999633789062, | |
| "logps/rejected": -160.46536254882812, | |
| "loss": 1.6052, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": 3.6505095958709717, | |
| "rewards/margins": 2.8529231548309326, | |
| "rewards/rejected": 0.7975867986679077, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3453689167974882, | |
| "grad_norm": 11.251614329473432, | |
| "learning_rate": 2.483985093039065e-07, | |
| "logits/chosen": 5866.24609375, | |
| "logits/rejected": 5800.03466796875, | |
| "logps/chosen": -178.62997436523438, | |
| "logps/rejected": -190.69210815429688, | |
| "loss": 1.6071, | |
| "rewards/accuracies": 0.6583333611488342, | |
| "rewards/chosen": 3.7587838172912598, | |
| "rewards/margins": 2.8152925968170166, | |
| "rewards/rejected": 0.9434911608695984, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3532182103610675, | |
| "grad_norm": 10.238672574135192, | |
| "learning_rate": 2.4525829031708404e-07, | |
| "logits/chosen": 6755.23828125, | |
| "logits/rejected": 4911.7578125, | |
| "logps/chosen": -204.50540161132812, | |
| "logps/rejected": -162.7772216796875, | |
| "loss": 1.6075, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": 4.24879264831543, | |
| "rewards/margins": 4.009242057800293, | |
| "rewards/rejected": 0.23955073952674866, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.36106750392464676, | |
| "grad_norm": 10.124703408047377, | |
| "learning_rate": 2.420464889567126e-07, | |
| "logits/chosen": 6971.953125, | |
| "logits/rejected": 5482.63330078125, | |
| "logps/chosen": -246.63485717773438, | |
| "logps/rejected": -189.0049591064453, | |
| "loss": 1.6062, | |
| "rewards/accuracies": 0.6583333015441895, | |
| "rewards/chosen": 3.8507399559020996, | |
| "rewards/margins": 2.7193245887756348, | |
| "rewards/rejected": 1.1314153671264648, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.36891679748822603, | |
| "grad_norm": 10.244451763866405, | |
| "learning_rate": 2.3876551874883775e-07, | |
| "logits/chosen": 5983.9541015625, | |
| "logits/rejected": 5281.8173828125, | |
| "logps/chosen": -208.52267456054688, | |
| "logps/rejected": -201.57501220703125, | |
| "loss": 1.6099, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": 3.680980682373047, | |
| "rewards/margins": 2.752244234085083, | |
| "rewards/rejected": 0.9287362098693848, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.37676609105180536, | |
| "grad_norm": 10.136642065676186, | |
| "learning_rate": 2.354178451968214e-07, | |
| "logits/chosen": 6450.548828125, | |
| "logits/rejected": 5633.7119140625, | |
| "logps/chosen": -203.6438751220703, | |
| "logps/rejected": -173.91026306152344, | |
| "loss": 1.6082, | |
| "rewards/accuracies": 0.6833333969116211, | |
| "rewards/chosen": 3.7612411975860596, | |
| "rewards/margins": 3.274782657623291, | |
| "rewards/rejected": 0.4864584505558014, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.38461538461538464, | |
| "grad_norm": 11.463940168293135, | |
| "learning_rate": 2.3200598392862542e-07, | |
| "logits/chosen": 6044.64501953125, | |
| "logits/rejected": 5817.2734375, | |
| "logps/chosen": -172.8741912841797, | |
| "logps/rejected": -201.54507446289062, | |
| "loss": 1.6072, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": 3.7329680919647217, | |
| "rewards/margins": 3.2064049243927, | |
| "rewards/rejected": 0.5265628695487976, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3924646781789639, | |
| "grad_norm": 10.258740856397358, | |
| "learning_rate": 2.2853249880642897e-07, | |
| "logits/chosen": 6601.33447265625, | |
| "logits/rejected": 5118.966796875, | |
| "logps/chosen": -219.5803985595703, | |
| "logps/rejected": -203.1927490234375, | |
| "loss": 1.6087, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": 2.4150400161743164, | |
| "rewards/margins": 3.0509958267211914, | |
| "rewards/rejected": -0.635955810546875, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.4003139717425432, | |
| "grad_norm": 9.618807221454839, | |
| "learning_rate": 2.25e-07, | |
| "logits/chosen": 5256.94482421875, | |
| "logits/rejected": 4686.318359375, | |
| "logps/chosen": -186.45138549804688, | |
| "logps/rejected": -166.85733032226562, | |
| "loss": 1.6107, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 2.9723703861236572, | |
| "rewards/margins": 3.497563123703003, | |
| "rewards/rejected": -0.5251928567886353, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.40816326530612246, | |
| "grad_norm": 10.077787546594172, | |
| "learning_rate": 2.2141114202526867e-07, | |
| "logits/chosen": 5602.5029296875, | |
| "logits/rejected": 5226.76904296875, | |
| "logps/chosen": -195.4163818359375, | |
| "logps/rejected": -194.8783721923828, | |
| "loss": 1.6119, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": 3.003880739212036, | |
| "rewards/margins": 2.041459321975708, | |
| "rewards/rejected": 0.9624212384223938, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.41601255886970173, | |
| "grad_norm": 10.362808776797175, | |
| "learning_rate": 2.1776862174957646e-07, | |
| "logits/chosen": 6315.44482421875, | |
| "logits/rejected": 5213.26953125, | |
| "logps/chosen": -208.7710723876953, | |
| "logps/rejected": -198.46128845214844, | |
| "loss": 1.6075, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": 4.163774490356445, | |
| "rewards/margins": 3.857957363128662, | |
| "rewards/rejected": 0.3058166205883026, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.423861852433281, | |
| "grad_norm": 11.129372316652974, | |
| "learning_rate": 2.1407517636510023e-07, | |
| "logits/chosen": 5997.4560546875, | |
| "logits/rejected": 5399.8779296875, | |
| "logps/chosen": -208.12722778320312, | |
| "logps/rejected": -188.67626953125, | |
| "loss": 1.6106, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 2.6769168376922607, | |
| "rewards/margins": 3.085752487182617, | |
| "rewards/rejected": -0.40883541107177734, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4317111459968603, | |
| "grad_norm": 10.874250167035393, | |
| "learning_rate": 2.1033358133197376e-07, | |
| "logits/chosen": 5797.27294921875, | |
| "logits/rejected": 5422.5419921875, | |
| "logps/chosen": -212.1249542236328, | |
| "logps/rejected": -225.52099609375, | |
| "loss": 1.6096, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": 2.68788480758667, | |
| "rewards/margins": 3.1572766304016113, | |
| "rewards/rejected": -0.46939191222190857, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.43956043956043955, | |
| "grad_norm": 11.368509816993306, | |
| "learning_rate": 2.0654664829265295e-07, | |
| "logits/chosen": 5849.35791015625, | |
| "logits/rejected": 5507.40625, | |
| "logps/chosen": -200.6465606689453, | |
| "logps/rejected": -193.24252319335938, | |
| "loss": 1.6123, | |
| "rewards/accuracies": 0.6416667103767395, | |
| "rewards/chosen": 2.6296989917755127, | |
| "rewards/margins": 2.6346523761749268, | |
| "rewards/rejected": -0.004953605122864246, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4474097331240188, | |
| "grad_norm": 9.634112543251097, | |
| "learning_rate": 2.027172229590914e-07, | |
| "logits/chosen": 6171.091796875, | |
| "logits/rejected": 5716.99951171875, | |
| "logps/chosen": -240.25399780273438, | |
| "logps/rejected": -223.2677764892578, | |
| "loss": 1.6112, | |
| "rewards/accuracies": 0.6416667103767395, | |
| "rewards/chosen": 2.6774706840515137, | |
| "rewards/margins": 2.21945858001709, | |
| "rewards/rejected": 0.45801201462745667, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.4552590266875981, | |
| "grad_norm": 10.87514770443522, | |
| "learning_rate": 1.9884818297431403e-07, | |
| "logits/chosen": 6132.4013671875, | |
| "logits/rejected": 5570.19921875, | |
| "logps/chosen": -209.4405975341797, | |
| "logps/rejected": -213.29257202148438, | |
| "loss": 1.6114, | |
| "rewards/accuracies": 0.7000001072883606, | |
| "rewards/chosen": 2.9603564739227295, | |
| "rewards/margins": 2.9691500663757324, | |
| "rewards/rejected": -0.008793527260422707, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4631083202511774, | |
| "grad_norm": 11.895191217861239, | |
| "learning_rate": 1.9494243574999645e-07, | |
| "logits/chosen": 6698.8017578125, | |
| "logits/rejected": 5693.74755859375, | |
| "logps/chosen": -228.00796508789062, | |
| "logps/rejected": -178.66822814941406, | |
| "loss": 1.6102, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": 3.6634438037872314, | |
| "rewards/margins": 3.507164716720581, | |
| "rewards/rejected": 0.15627913177013397, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.47095761381475665, | |
| "grad_norm": 10.761146722421474, | |
| "learning_rate": 1.9100291628167416e-07, | |
| "logits/chosen": 6341.4296875, | |
| "logits/rejected": 5422.83984375, | |
| "logps/chosen": -175.03768920898438, | |
| "logps/rejected": -189.64918518066406, | |
| "loss": 1.6124, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": 2.5099081993103027, | |
| "rewards/margins": 2.4614367485046387, | |
| "rewards/rejected": 0.04847153276205063, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.478806907378336, | |
| "grad_norm": 10.066536779214626, | |
| "learning_rate": 1.870325849432238e-07, | |
| "logits/chosen": 6092.59765625, | |
| "logits/rejected": 5363.5263671875, | |
| "logps/chosen": -203.3939971923828, | |
| "logps/rejected": -186.27825927734375, | |
| "loss": 1.6096, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 2.739835023880005, | |
| "rewards/margins": 3.611487627029419, | |
| "rewards/rejected": -0.8716526031494141, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.48665620094191525, | |
| "grad_norm": 9.620539433149245, | |
| "learning_rate": 1.8303442526227364e-07, | |
| "logits/chosen": 5198.9228515625, | |
| "logits/rejected": 4301.599609375, | |
| "logps/chosen": -181.21054077148438, | |
| "logps/rejected": -162.38772583007812, | |
| "loss": 1.6114, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": 2.554405689239502, | |
| "rewards/margins": 3.846482038497925, | |
| "rewards/rejected": -1.2920769453048706, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.4945054945054945, | |
| "grad_norm": 10.87064291559102, | |
| "learning_rate": 1.7901144167821507e-07, | |
| "logits/chosen": 6764.76171875, | |
| "logits/rejected": 5845.1396484375, | |
| "logps/chosen": -239.6127166748047, | |
| "logps/rejected": -195.640869140625, | |
| "loss": 1.6094, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 3.2196297645568848, | |
| "rewards/margins": 4.5956902503967285, | |
| "rewards/rejected": -1.3760606050491333, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.5023547880690737, | |
| "grad_norm": 11.572214216615475, | |
| "learning_rate": 1.749666572845e-07, | |
| "logits/chosen": 6422.68505859375, | |
| "logits/rejected": 5511.61083984375, | |
| "logps/chosen": -208.0812225341797, | |
| "logps/rejected": -215.52230834960938, | |
| "loss": 1.6109, | |
| "rewards/accuracies": 0.7083333134651184, | |
| "rewards/chosen": 2.6956992149353027, | |
| "rewards/margins": 3.72594952583313, | |
| "rewards/rejected": -1.0302503108978271, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.5102040816326531, | |
| "grad_norm": 10.924127233657837, | |
| "learning_rate": 1.7090311155692034e-07, | |
| "logits/chosen": 5531.26953125, | |
| "logits/rejected": 4566.18115234375, | |
| "logps/chosen": -194.7221221923828, | |
| "logps/rejected": -170.06761169433594, | |
| "loss": 1.6136, | |
| "rewards/accuracies": 0.6916666626930237, | |
| "rewards/chosen": 1.600431203842163, | |
| "rewards/margins": 2.939448595046997, | |
| "rewards/rejected": -1.339017391204834, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.5180533751962323, | |
| "grad_norm": 11.128560662698712, | |
| "learning_rate": 1.6682385806957695e-07, | |
| "logits/chosen": 6207.12646484375, | |
| "logits/rejected": 5023.8916015625, | |
| "logps/chosen": -215.07473754882812, | |
| "logps/rejected": -189.7795867919922, | |
| "loss": 1.6147, | |
| "rewards/accuracies": 0.7916666865348816, | |
| "rewards/chosen": 2.3060238361358643, | |
| "rewards/margins": 4.0903449058532715, | |
| "rewards/rejected": -1.7843214273452759, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.5259026687598116, | |
| "grad_norm": 11.101124136697601, | |
| "learning_rate": 1.627319622002544e-07, | |
| "logits/chosen": 6186.93798828125, | |
| "logits/rejected": 5106.5810546875, | |
| "logps/chosen": -204.43142700195312, | |
| "logps/rejected": -189.8342742919922, | |
| "loss": 1.6158, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 2.3008999824523926, | |
| "rewards/margins": 3.2730765342712402, | |
| "rewards/rejected": -0.9721772074699402, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.533751962323391, | |
| "grad_norm": 11.556695250472748, | |
| "learning_rate": 1.5863049882692565e-07, | |
| "logits/chosen": 6008.26904296875, | |
| "logits/rejected": 5051.5166015625, | |
| "logps/chosen": -212.3236083984375, | |
| "logps/rejected": -180.04847717285156, | |
| "loss": 1.6165, | |
| "rewards/accuracies": 0.6500000357627869, | |
| "rewards/chosen": 1.3081023693084717, | |
| "rewards/margins": 2.1385140419006348, | |
| "rewards/rejected": -0.8304115533828735, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5416012558869702, | |
| "grad_norm": 10.233108364900362, | |
| "learning_rate": 1.5452255001711795e-07, | |
| "logits/chosen": 5943.60302734375, | |
| "logits/rejected": 5483.03955078125, | |
| "logps/chosen": -207.3726348876953, | |
| "logps/rejected": -217.35693359375, | |
| "loss": 1.6141, | |
| "rewards/accuracies": 0.7000000476837158, | |
| "rewards/chosen": 1.7650575637817383, | |
| "rewards/margins": 2.921135425567627, | |
| "rewards/rejected": -1.1560779809951782, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5494505494505495, | |
| "grad_norm": 8.962560556788734, | |
| "learning_rate": 1.50411202711876e-07, | |
| "logits/chosen": 5653.8994140625, | |
| "logits/rejected": 5460.46044921875, | |
| "logps/chosen": -187.51516723632812, | |
| "logps/rejected": -188.84127807617188, | |
| "loss": 1.6177, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 1.7124507427215576, | |
| "rewards/margins": 3.6060166358947754, | |
| "rewards/rejected": -1.8935661315917969, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5572998430141287, | |
| "grad_norm": 10.815111342313953, | |
| "learning_rate": 1.4629954640606275e-07, | |
| "logits/chosen": 5610.15966796875, | |
| "logits/rejected": 5674.67822265625, | |
| "logps/chosen": -192.265625, | |
| "logps/rejected": -203.23452758789062, | |
| "loss": 1.6147, | |
| "rewards/accuracies": 0.6583333611488342, | |
| "rewards/chosen": 1.6721900701522827, | |
| "rewards/margins": 2.265028476715088, | |
| "rewards/rejected": -0.5928383469581604, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.565149136577708, | |
| "grad_norm": 11.454299662188049, | |
| "learning_rate": 1.4219067082674153e-07, | |
| "logits/chosen": 6028.6044921875, | |
| "logits/rejected": 5230.07373046875, | |
| "logps/chosen": -217.03543090820312, | |
| "logps/rejected": -201.26968383789062, | |
| "loss": 1.6162, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 1.9704124927520752, | |
| "rewards/margins": 3.4510631561279297, | |
| "rewards/rejected": -1.4806510210037231, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.5729984301412873, | |
| "grad_norm": 11.472362484650057, | |
| "learning_rate": 1.3808766361138294e-07, | |
| "logits/chosen": 6529.74609375, | |
| "logits/rejected": 5061.63623046875, | |
| "logps/chosen": -224.10916137695312, | |
| "logps/rejected": -176.45887756347656, | |
| "loss": 1.6142, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 1.0688775777816772, | |
| "rewards/margins": 2.4734768867492676, | |
| "rewards/rejected": -1.4045993089675903, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5808477237048666, | |
| "grad_norm": 10.32346024402139, | |
| "learning_rate": 1.3399360798764307e-07, | |
| "logits/chosen": 5415.81103515625, | |
| "logits/rejected": 4646.49267578125, | |
| "logps/chosen": -191.19082641601562, | |
| "logps/rejected": -165.0283203125, | |
| "loss": 1.6157, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": 1.2511141300201416, | |
| "rewards/margins": 3.6377835273742676, | |
| "rewards/rejected": -2.386669635772705, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5886970172684458, | |
| "grad_norm": 11.219416384074425, | |
| "learning_rate": 1.2991158045645435e-07, | |
| "logits/chosen": 5891.90380859375, | |
| "logits/rejected": 5381.7978515625, | |
| "logps/chosen": -179.5184326171875, | |
| "logps/rejected": -168.6924591064453, | |
| "loss": 1.6173, | |
| "rewards/accuracies": 0.6416667103767395, | |
| "rewards/chosen": 0.8637522459030151, | |
| "rewards/margins": 2.4437310695648193, | |
| "rewards/rejected": -1.5799791812896729, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5965463108320251, | |
| "grad_norm": 10.454823936279826, | |
| "learning_rate": 1.2584464848017214e-07, | |
| "logits/chosen": 6189.07373046875, | |
| "logits/rejected": 4946.9921875, | |
| "logps/chosen": -201.6912078857422, | |
| "logps/rejected": -176.08975219726562, | |
| "loss": 1.6189, | |
| "rewards/accuracies": 0.708333432674408, | |
| "rewards/chosen": 1.5779491662979126, | |
| "rewards/margins": 3.1704206466674805, | |
| "rewards/rejected": -1.592471718788147, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.6043956043956044, | |
| "grad_norm": 11.144568917516514, | |
| "learning_rate": 1.2179586817751263e-07, | |
| "logits/chosen": 6398.5439453125, | |
| "logits/rejected": 5549.02197265625, | |
| "logps/chosen": -229.02297973632812, | |
| "logps/rejected": -226.2268829345703, | |
| "loss": 1.6196, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.6905540227890015, | |
| "rewards/margins": 3.0377731323242188, | |
| "rewards/rejected": -2.3472187519073486, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.6122448979591837, | |
| "grad_norm": 11.784984283909536, | |
| "learning_rate": 1.177682820270159e-07, | |
| "logits/chosen": 6260.98388671875, | |
| "logits/rejected": 5981.9970703125, | |
| "logps/chosen": -215.8638153076172, | |
| "logps/rejected": -239.6117706298828, | |
| "loss": 1.62, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.24781355261802673, | |
| "rewards/margins": 2.456618309020996, | |
| "rewards/rejected": -2.2088046073913574, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.6200941915227629, | |
| "grad_norm": 11.336513282533264, | |
| "learning_rate": 1.13764916580758e-07, | |
| "logits/chosen": 5813.20849609375, | |
| "logits/rejected": 5193.22705078125, | |
| "logps/chosen": -197.9930877685547, | |
| "logps/rejected": -192.1040496826172, | |
| "loss": 1.6205, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.16777479648590088, | |
| "rewards/margins": 3.6429495811462402, | |
| "rewards/rejected": -3.47517466545105, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.6279434850863422, | |
| "grad_norm": 11.37275833046531, | |
| "learning_rate": 1.0978878019003183e-07, | |
| "logits/chosen": 5823.0517578125, | |
| "logits/rejected": 5127.9541015625, | |
| "logps/chosen": -216.37841796875, | |
| "logps/rejected": -215.0908203125, | |
| "loss": 1.6191, | |
| "rewards/accuracies": 0.6333333253860474, | |
| "rewards/chosen": 0.3729097247123718, | |
| "rewards/margins": 2.810680866241455, | |
| "rewards/rejected": -2.4377713203430176, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.6357927786499215, | |
| "grad_norm": 11.92843313737738, | |
| "learning_rate": 1.0584286074470446e-07, | |
| "logits/chosen": 5377.32421875, | |
| "logits/rejected": 4863.58154296875, | |
| "logps/chosen": -201.692626953125, | |
| "logps/rejected": -224.701171875, | |
| "loss": 1.6182, | |
| "rewards/accuracies": 0.7666667103767395, | |
| "rewards/chosen": 0.7583420872688293, | |
| "rewards/margins": 4.198758602142334, | |
| "rewards/rejected": -3.4404168128967285, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6436420722135008, | |
| "grad_norm": 11.620823490647497, | |
| "learning_rate": 1.0193012342795095e-07, | |
| "logits/chosen": 5848.99462890625, | |
| "logits/rejected": 4559.11669921875, | |
| "logps/chosen": -219.4331512451172, | |
| "logps/rejected": -170.939697265625, | |
| "loss": 1.6184, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 1.7729305028915405, | |
| "rewards/margins": 4.044053077697754, | |
| "rewards/rejected": -2.271122694015503, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6514913657770801, | |
| "grad_norm": 10.399323259266707, | |
| "learning_rate": 9.805350848805078e-08, | |
| "logits/chosen": 5789.95068359375, | |
| "logits/rejected": 5189.4072265625, | |
| "logps/chosen": -203.59767150878906, | |
| "logps/rejected": -195.38893127441406, | |
| "loss": 1.6199, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.5932119488716125, | |
| "rewards/margins": 3.508798122406006, | |
| "rewards/rejected": -2.915585994720459, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.6593406593406593, | |
| "grad_norm": 10.035390499048138, | |
| "learning_rate": 9.421592902892184e-08, | |
| "logits/chosen": 6314.8837890625, | |
| "logits/rejected": 5214.59326171875, | |
| "logps/chosen": -227.4573974609375, | |
| "logps/rejected": -215.1611785888672, | |
| "loss": 1.6204, | |
| "rewards/accuracies": 0.5833333134651184, | |
| "rewards/chosen": -0.09185920655727386, | |
| "rewards/margins": 2.265404462814331, | |
| "rewards/rejected": -2.3572635650634766, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6671899529042387, | |
| "grad_norm": 11.544156642460049, | |
| "learning_rate": 9.042026882105215e-08, | |
| "logits/chosen": 5835.57275390625, | |
| "logits/rejected": 5284.0244140625, | |
| "logps/chosen": -214.9558563232422, | |
| "logps/rejected": -203.69520568847656, | |
| "loss": 1.618, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": 0.5202548503875732, | |
| "rewards/margins": 3.6990485191345215, | |
| "rewards/rejected": -3.1787941455841064, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6750392464678179, | |
| "grad_norm": 11.190718842900319, | |
| "learning_rate": 8.666938013447484e-08, | |
| "logits/chosen": 6286.025390625, | |
| "logits/rejected": 5394.0, | |
| "logps/chosen": -247.1281280517578, | |
| "logps/rejected": -205.7931671142578, | |
| "loss": 1.6199, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": 0.6725990772247314, | |
| "rewards/margins": 3.7918269634246826, | |
| "rewards/rejected": -3.119227886199951, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6828885400313972, | |
| "grad_norm": 11.069681395938431, | |
| "learning_rate": 8.29660815954134e-08, | |
| "logits/chosen": 6258.58837890625, | |
| "logits/rejected": 5223.68017578125, | |
| "logps/chosen": -229.8612518310547, | |
| "logps/rejected": -203.4960174560547, | |
| "loss": 1.6203, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 0.273890882730484, | |
| "rewards/margins": 3.0515148639678955, | |
| "rewards/rejected": -2.7776238918304443, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6907378335949764, | |
| "grad_norm": 10.760516756235438, | |
| "learning_rate": 7.931315606820979e-08, | |
| "logits/chosen": 5503.98388671875, | |
| "logits/rejected": 5300.1279296875, | |
| "logps/chosen": -210.29623413085938, | |
| "logps/rejected": -234.48428344726562, | |
| "loss": 1.6209, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.23874196410179138, | |
| "rewards/margins": 3.12656831741333, | |
| "rewards/rejected": -2.887826442718506, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6985871271585558, | |
| "grad_norm": 11.62760722742798, | |
| "learning_rate": 7.571334856412503e-08, | |
| "logits/chosen": 6500.73779296875, | |
| "logits/rejected": 5318.8017578125, | |
| "logps/chosen": -257.2298278808594, | |
| "logps/rejected": -215.5242156982422, | |
| "loss": 1.6219, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.3969727158546448, | |
| "rewards/margins": 4.151793003082275, | |
| "rewards/rejected": -3.754819869995117, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.706436420722135, | |
| "grad_norm": 9.458870050953964, | |
| "learning_rate": 7.216936417858549e-08, | |
| "logits/chosen": 5499.93701171875, | |
| "logits/rejected": 5135.26318359375, | |
| "logps/chosen": -204.2237548828125, | |
| "logps/rejected": -214.3614959716797, | |
| "loss": 1.622, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": 0.10146790742874146, | |
| "rewards/margins": 4.020596027374268, | |
| "rewards/rejected": -3.919127941131592, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.7142857142857143, | |
| "grad_norm": 11.195584952901939, | |
| "learning_rate": 6.868386605842379e-08, | |
| "logits/chosen": 5412.22216796875, | |
| "logits/rejected": 5166.134765625, | |
| "logps/chosen": -207.86279296875, | |
| "logps/rejected": -212.60983276367188, | |
| "loss": 1.6223, | |
| "rewards/accuracies": 0.7666666507720947, | |
| "rewards/chosen": 0.4062539041042328, | |
| "rewards/margins": 5.072477340698242, | |
| "rewards/rejected": -4.666223049163818, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.7221350078492935, | |
| "grad_norm": 10.447859698886178, | |
| "learning_rate": 6.525947340064232e-08, | |
| "logits/chosen": 5988.22265625, | |
| "logits/rejected": 4997.0390625, | |
| "logps/chosen": -232.16281127929688, | |
| "logps/rejected": -202.34010314941406, | |
| "loss": 1.6219, | |
| "rewards/accuracies": 0.7583333849906921, | |
| "rewards/chosen": 0.2832966446876526, | |
| "rewards/margins": 4.437819480895996, | |
| "rewards/rejected": -4.154522895812988, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.7299843014128728, | |
| "grad_norm": 11.894379056711179, | |
| "learning_rate": 6.189875948420276e-08, | |
| "logits/chosen": 5893.60205078125, | |
| "logits/rejected": 4513.279296875, | |
| "logps/chosen": -236.3983612060547, | |
| "logps/rejected": -190.51617431640625, | |
| "loss": 1.6224, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.3916471600532532, | |
| "rewards/margins": 4.208518981933594, | |
| "rewards/rejected": -4.600167274475098, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.7378335949764521, | |
| "grad_norm": 10.98098284583286, | |
| "learning_rate": 5.86042497363215e-08, | |
| "logits/chosen": 5721.9482421875, | |
| "logits/rejected": 5032.0947265625, | |
| "logps/chosen": -218.5546875, | |
| "logps/rejected": -209.4766387939453, | |
| "loss": 1.6227, | |
| "rewards/accuracies": 0.6916666626930237, | |
| "rewards/chosen": -0.21962371468544006, | |
| "rewards/margins": 4.292196273803711, | |
| "rewards/rejected": -4.511819362640381, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.7456828885400314, | |
| "grad_norm": 10.09472856202966, | |
| "learning_rate": 5.537841983472337e-08, | |
| "logits/chosen": 5903.6142578125, | |
| "logits/rejected": 5460.48779296875, | |
| "logps/chosen": -217.61929321289062, | |
| "logps/rejected": -217.16525268554688, | |
| "loss": 1.6233, | |
| "rewards/accuracies": 0.6833333373069763, | |
| "rewards/chosen": 0.16755667328834534, | |
| "rewards/margins": 3.2451682090759277, | |
| "rewards/rejected": -3.0776114463806152, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7535321821036107, | |
| "grad_norm": 11.803340128971255, | |
| "learning_rate": 5.222369384727954e-08, | |
| "logits/chosen": 5730.85205078125, | |
| "logits/rejected": 5144.63037109375, | |
| "logps/chosen": -218.22622680664062, | |
| "logps/rejected": -206.0850372314453, | |
| "loss": 1.6222, | |
| "rewards/accuracies": 0.7916666269302368, | |
| "rewards/chosen": 0.27254268527030945, | |
| "rewards/margins": 3.8942532539367676, | |
| "rewards/rejected": -3.6217105388641357, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7613814756671899, | |
| "grad_norm": 11.454679795991241, | |
| "learning_rate": 4.914244241042832e-08, | |
| "logits/chosen": 5840.7158203125, | |
| "logits/rejected": 4928.1845703125, | |
| "logps/chosen": -240.2581787109375, | |
| "logps/rejected": -210.8507537841797, | |
| "loss": 1.6232, | |
| "rewards/accuracies": 0.7249999642372131, | |
| "rewards/chosen": -0.018189292401075363, | |
| "rewards/margins": 4.160759925842285, | |
| "rewards/rejected": -4.178948402404785, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7692307692307693, | |
| "grad_norm": 11.99470899522862, | |
| "learning_rate": 4.613698094774715e-08, | |
| "logits/chosen": 5542.8330078125, | |
| "logits/rejected": 4527.66015625, | |
| "logps/chosen": -228.67837524414062, | |
| "logps/rejected": -195.180419921875, | |
| "loss": 1.622, | |
| "rewards/accuracies": 0.7583333253860474, | |
| "rewards/chosen": 0.15695662796497345, | |
| "rewards/margins": 4.860254764556885, | |
| "rewards/rejected": -4.703298091888428, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7770800627943485, | |
| "grad_norm": 13.577338184519638, | |
| "learning_rate": 4.3209567930014746e-08, | |
| "logits/chosen": 5864.568359375, | |
| "logits/rejected": 4959.56396484375, | |
| "logps/chosen": -226.41036987304688, | |
| "logps/rejected": -198.62289428710938, | |
| "loss": 1.622, | |
| "rewards/accuracies": 0.6416667103767395, | |
| "rewards/chosen": -0.5170873403549194, | |
| "rewards/margins": 4.038225173950195, | |
| "rewards/rejected": -4.555312156677246, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7849293563579278, | |
| "grad_norm": 11.227134966757461, | |
| "learning_rate": 4.036240317807059e-08, | |
| "logits/chosen": 5562.1455078125, | |
| "logits/rejected": 4746.4765625, | |
| "logps/chosen": -213.9759979248047, | |
| "logps/rejected": -184.7088165283203, | |
| "loss": 1.6223, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": -0.1758609116077423, | |
| "rewards/margins": 4.17446231842041, | |
| "rewards/rejected": -4.35032320022583, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.792778649921507, | |
| "grad_norm": 10.788480554925878, | |
| "learning_rate": 3.759762620974716e-08, | |
| "logits/chosen": 5739.294921875, | |
| "logits/rejected": 4851.3935546875, | |
| "logps/chosen": -214.68118286132812, | |
| "logps/rejected": -195.8408203125, | |
| "loss": 1.6224, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.02830815315246582, | |
| "rewards/margins": 4.474509239196777, | |
| "rewards/rejected": -4.446200847625732, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.8006279434850864, | |
| "grad_norm": 10.560717225935313, | |
| "learning_rate": 3.4917314632117423e-08, | |
| "logits/chosen": 5831.240234375, | |
| "logits/rejected": 5222.7080078125, | |
| "logps/chosen": -248.1968994140625, | |
| "logps/rejected": -230.4774932861328, | |
| "loss": 1.6229, | |
| "rewards/accuracies": 0.783333420753479, | |
| "rewards/chosen": -0.018525028601288795, | |
| "rewards/margins": 4.307133674621582, | |
| "rewards/rejected": -4.325658798217773, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.8084772370486656, | |
| "grad_norm": 11.450889100406535, | |
| "learning_rate": 3.232348258026552e-08, | |
| "logits/chosen": 6223.50048828125, | |
| "logits/rejected": 5888.0341796875, | |
| "logps/chosen": -231.3888702392578, | |
| "logps/rejected": -237.2459259033203, | |
| "loss": 1.6228, | |
| "rewards/accuracies": 0.7833333611488342, | |
| "rewards/chosen": 0.0547012984752655, | |
| "rewards/margins": 4.181652545928955, | |
| "rewards/rejected": -4.126951217651367, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.8163265306122449, | |
| "grad_norm": 11.896899452824945, | |
| "learning_rate": 2.9818079203753654e-08, | |
| "logits/chosen": 5537.16064453125, | |
| "logits/rejected": 5094.0849609375, | |
| "logps/chosen": -220.12808227539062, | |
| "logps/rejected": -212.80624389648438, | |
| "loss": 1.6234, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": -0.038040757179260254, | |
| "rewards/margins": 3.5793395042419434, | |
| "rewards/rejected": -3.617380142211914, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.8241758241758241, | |
| "grad_norm": 12.623914170332412, | |
| "learning_rate": 2.740298720192245e-08, | |
| "logits/chosen": 5572.4052734375, | |
| "logits/rejected": 5124.4365234375, | |
| "logps/chosen": -237.76119995117188, | |
| "logps/rejected": -219.5039825439453, | |
| "loss": 1.6226, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -0.2121574878692627, | |
| "rewards/margins": 3.454202175140381, | |
| "rewards/rejected": -3.6663596630096436, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.8320251177394035, | |
| "grad_norm": 11.913054191421269, | |
| "learning_rate": 2.5080021409126655e-08, | |
| "logits/chosen": 6892.4423828125, | |
| "logits/rejected": 5450.9912109375, | |
| "logps/chosen": -261.4605407714844, | |
| "logps/rejected": -224.04019165039062, | |
| "loss": 1.6228, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -0.4351968765258789, | |
| "rewards/margins": 3.5411152839660645, | |
| "rewards/rejected": -3.9763119220733643, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.8398744113029827, | |
| "grad_norm": 11.445993088464819, | |
| "learning_rate": 2.2850927430967287e-08, | |
| "logits/chosen": 5742.19580078125, | |
| "logits/rejected": 5310.591796875, | |
| "logps/chosen": -225.3037567138672, | |
| "logps/rejected": -197.18203735351562, | |
| "loss": 1.624, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": -0.027328431606292725, | |
| "rewards/margins": 4.017100811004639, | |
| "rewards/rejected": -4.044428825378418, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.847723704866562, | |
| "grad_norm": 11.059510886657053, | |
| "learning_rate": 2.0717380332546768e-08, | |
| "logits/chosen": 6291.6181640625, | |
| "logits/rejected": 4913.92919921875, | |
| "logps/chosen": -266.40643310546875, | |
| "logps/rejected": -206.01705932617188, | |
| "loss": 1.6219, | |
| "rewards/accuracies": 0.8166667222976685, | |
| "rewards/chosen": 0.1874084770679474, | |
| "rewards/margins": 5.219174385070801, | |
| "rewards/rejected": -5.031765937805176, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8555729984301413, | |
| "grad_norm": 11.093065022674237, | |
| "learning_rate": 1.8680983379731618e-08, | |
| "logits/chosen": 6390.0458984375, | |
| "logits/rejected": 4830.373046875, | |
| "logps/chosen": -228.0162353515625, | |
| "logps/rejected": -217.597900390625, | |
| "loss": 1.6244, | |
| "rewards/accuracies": 0.7333334684371948, | |
| "rewards/chosen": -0.5020327568054199, | |
| "rewards/margins": 4.1861772537231445, | |
| "rewards/rejected": -4.688210487365723, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8634222919937206, | |
| "grad_norm": 11.004461382075167, | |
| "learning_rate": 1.6743266834369856e-08, | |
| "logits/chosen": 6145.6806640625, | |
| "logits/rejected": 5110.62548828125, | |
| "logps/chosen": -239.3874053955078, | |
| "logps/rejected": -194.05117797851562, | |
| "loss": 1.6246, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -0.8312684893608093, | |
| "rewards/margins": 3.724156141281128, | |
| "rewards/rejected": -4.555424690246582, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8712715855572999, | |
| "grad_norm": 11.169359353898571, | |
| "learning_rate": 1.4905686804366853e-08, | |
| "logits/chosen": 6507.7529296875, | |
| "logits/rejected": 5290.05908203125, | |
| "logps/chosen": -265.2042236328125, | |
| "logps/rejected": -227.3978729248047, | |
| "loss": 1.6229, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -0.6048145294189453, | |
| "rewards/margins": 4.526320457458496, | |
| "rewards/rejected": -5.131134986877441, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8791208791208791, | |
| "grad_norm": 10.36627141349426, | |
| "learning_rate": 1.3169624149485037e-08, | |
| "logits/chosen": 6368.01416015625, | |
| "logits/rejected": 5209.26611328125, | |
| "logps/chosen": -246.00863647460938, | |
| "logps/rejected": -221.8163604736328, | |
| "loss": 1.6232, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.8775882720947266, | |
| "rewards/margins": 4.164261817932129, | |
| "rewards/rejected": -5.041850566864014, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8869701726844584, | |
| "grad_norm": 11.689309702253437, | |
| "learning_rate": 1.1536383443688968e-08, | |
| "logits/chosen": 5779.93896484375, | |
| "logits/rejected": 4615.59326171875, | |
| "logps/chosen": -234.68215942382812, | |
| "logps/rejected": -184.45750427246094, | |
| "loss": 1.6232, | |
| "rewards/accuracies": 0.7333333492279053, | |
| "rewards/chosen": 0.10495586693286896, | |
| "rewards/margins": 4.333970546722412, | |
| "rewards/rejected": -4.2290143966674805, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8948194662480377, | |
| "grad_norm": 11.69568286023684, | |
| "learning_rate": 1.0007191994816272e-08, | |
| "logits/chosen": 6343.60009765625, | |
| "logits/rejected": 5211.26416015625, | |
| "logps/chosen": -258.3962097167969, | |
| "logps/rejected": -225.07742309570312, | |
| "loss": 1.6226, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.7535941004753113, | |
| "rewards/margins": 3.4583687782287598, | |
| "rewards/rejected": -4.211962699890137, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.902668759811617, | |
| "grad_norm": 12.01611350837566, | |
| "learning_rate": 8.583198922310031e-09, | |
| "logits/chosen": 5625.16796875, | |
| "logits/rejected": 5218.8232421875, | |
| "logps/chosen": -227.22756958007812, | |
| "logps/rejected": -223.5425262451172, | |
| "loss": 1.623, | |
| "rewards/accuracies": 0.7166666984558105, | |
| "rewards/chosen": -0.24080140888690948, | |
| "rewards/margins": 3.7667922973632812, | |
| "rewards/rejected": -4.007594108581543, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.9105180533751962, | |
| "grad_norm": 11.149096309955521, | |
| "learning_rate": 7.2654742937067916e-09, | |
| "logits/chosen": 5776.7119140625, | |
| "logits/rejected": 4638.9755859375, | |
| "logps/chosen": -228.9247283935547, | |
| "logps/rejected": -199.80224609375, | |
| "loss": 1.624, | |
| "rewards/accuracies": 0.7083333134651184, | |
| "rewards/chosen": -0.35184532403945923, | |
| "rewards/margins": 3.520703077316284, | |
| "rewards/rejected": -3.8725483417510986, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.9183673469387755, | |
| "grad_norm": 11.730310171305202, | |
| "learning_rate": 6.0550083205280325e-09, | |
| "logits/chosen": 6331.5869140625, | |
| "logits/rejected": 6107.18408203125, | |
| "logps/chosen": -243.18026733398438, | |
| "logps/rejected": -251.1663360595703, | |
| "loss": 1.6231, | |
| "rewards/accuracies": 0.6666666865348816, | |
| "rewards/chosen": -0.2250463217496872, | |
| "rewards/margins": 4.151488780975342, | |
| "rewards/rejected": -4.376535415649414, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.9262166405023547, | |
| "grad_norm": 12.532143717176293, | |
| "learning_rate": 4.952710614180567e-09, | |
| "logits/chosen": 5591.78173828125, | |
| "logits/rejected": 5125.27197265625, | |
| "logps/chosen": -240.2451629638672, | |
| "logps/rejected": -220.843505859375, | |
| "loss": 1.6247, | |
| "rewards/accuracies": 0.7416667342185974, | |
| "rewards/chosen": -0.33638685941696167, | |
| "rewards/margins": 3.716320753097534, | |
| "rewards/rejected": -4.052707672119141, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.9340659340659341, | |
| "grad_norm": 11.073947353465151, | |
| "learning_rate": 3.959409502423744e-09, | |
| "logits/chosen": 6082.80712890625, | |
| "logits/rejected": 5334.0517578125, | |
| "logps/chosen": -250.0338897705078, | |
| "logps/rejected": -221.93896484375, | |
| "loss": 1.6231, | |
| "rewards/accuracies": 0.6750000715255737, | |
| "rewards/chosen": -1.1701091527938843, | |
| "rewards/margins": 3.618891477584839, | |
| "rewards/rejected": -4.789000511169434, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.9419152276295133, | |
| "grad_norm": 11.478913045969124, | |
| "learning_rate": 3.075851406917995e-09, | |
| "logits/chosen": 6255.9599609375, | |
| "logits/rejected": 5368.0625, | |
| "logps/chosen": -256.30841064453125, | |
| "logps/rejected": -224.0297393798828, | |
| "loss": 1.6228, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": -0.6638107299804688, | |
| "rewards/margins": 4.6817946434021, | |
| "rewards/rejected": -5.345605373382568, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.9497645211930926, | |
| "grad_norm": 11.772570086955174, | |
| "learning_rate": 2.3027002823221064e-09, | |
| "logits/chosen": 5870.1064453125, | |
| "logits/rejected": 5151.9208984375, | |
| "logps/chosen": -240.24136352539062, | |
| "logps/rejected": -222.8927764892578, | |
| "loss": 1.6242, | |
| "rewards/accuracies": 0.7916667461395264, | |
| "rewards/chosen": -0.034066688269376755, | |
| "rewards/margins": 4.694322109222412, | |
| "rewards/rejected": -4.728388786315918, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.957613814756672, | |
| "grad_norm": 11.341650876448142, | |
| "learning_rate": 1.6405371173608073e-09, | |
| "logits/chosen": 7316.70556640625, | |
| "logits/rejected": 5417.42626953125, | |
| "logps/chosen": -297.10943603515625, | |
| "logps/rejected": -250.65249633789062, | |
| "loss": 1.6226, | |
| "rewards/accuracies": 0.7249999642372131, | |
| "rewards/chosen": -0.27967390418052673, | |
| "rewards/margins": 5.194223880767822, | |
| "rewards/rejected": -5.4738969802856445, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9654631083202512, | |
| "grad_norm": 12.25958376489255, | |
| "learning_rate": 1.0898594982375387e-09, | |
| "logits/chosen": 6507.357421875, | |
| "logits/rejected": 5199.5634765625, | |
| "logps/chosen": -260.1299133300781, | |
| "logps/rejected": -217.67202758789062, | |
| "loss": 1.6232, | |
| "rewards/accuracies": 0.7416666746139526, | |
| "rewards/chosen": 0.43730440735816956, | |
| "rewards/margins": 5.117526054382324, | |
| "rewards/rejected": -4.680222511291504, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9733124018838305, | |
| "grad_norm": 12.320721462602215, | |
| "learning_rate": 6.510812347205185e-10, | |
| "logits/chosen": 5608.94189453125, | |
| "logits/rejected": 5175.58935546875, | |
| "logps/chosen": -232.12710571289062, | |
| "logps/rejected": -239.9905242919922, | |
| "loss": 1.6223, | |
| "rewards/accuracies": 0.6916667222976685, | |
| "rewards/chosen": -0.5330235362052917, | |
| "rewards/margins": 4.260423183441162, | |
| "rewards/rejected": -4.7934465408325195, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9811616954474097, | |
| "grad_norm": 11.623382089715227, | |
| "learning_rate": 3.245320491830683e-10, | |
| "logits/chosen": 5712.59912109375, | |
| "logits/rejected": 4849.1201171875, | |
| "logps/chosen": -230.5782928466797, | |
| "logps/rejected": -193.20933532714844, | |
| "loss": 1.6227, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -0.2468281090259552, | |
| "rewards/margins": 3.513197660446167, | |
| "rewards/rejected": -3.7600255012512207, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.989010989010989, | |
| "grad_norm": 10.941217530065499, | |
| "learning_rate": 1.1045732883197545e-10, | |
| "logits/chosen": 5392.3876953125, | |
| "logits/rejected": 5146.81103515625, | |
| "logps/chosen": -233.12521362304688, | |
| "logps/rejected": -220.48587036132812, | |
| "loss": 1.6232, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": -0.9743496775627136, | |
| "rewards/margins": 3.1852924823760986, | |
| "rewards/rejected": -4.159642219543457, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9968602825745683, | |
| "grad_norm": 11.869179900549053, | |
| "learning_rate": 9.017941309813482e-12, | |
| "logits/chosen": 5663.7666015625, | |
| "logits/rejected": 4646.2197265625, | |
| "logps/chosen": -224.53286743164062, | |
| "logps/rejected": -189.52761840820312, | |
| "loss": 1.6221, | |
| "rewards/accuracies": 0.7083333730697632, | |
| "rewards/chosen": -0.044149044901132584, | |
| "rewards/margins": 3.1750922203063965, | |
| "rewards/rejected": -3.2192413806915283, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 1274, | |
| "total_flos": 0.0, | |
| "train_loss": 1.6157182269792543, | |
| "train_runtime": 13119.408, | |
| "train_samples_per_second": 4.66, | |
| "train_steps_per_second": 0.097 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1274, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 2000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": false, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 6, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |