| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.409626216077829, |
| "eval_steps": 300, |
| "global_step": 50, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.00819252432155658, |
| "grad_norm": 6.9504547119140625, |
| "learning_rate": 1.5384615384615387e-05, |
| "logits/chosen": -0.368751585483551, |
| "logits/rejected": -0.5192137956619263, |
| "logps/chosen": -6.009060859680176, |
| "logps/rejected": -15.29100513458252, |
| "loss": 1.2444509267807007, |
| "memory(GiB)": 45.94, |
| "nll_loss": 0.5513038039207458, |
| "rewards/accuracies": 0.0, |
| "rewards/chosen": 0.0, |
| "rewards/margins": 0.0, |
| "rewards/rejected": 0.0, |
| "step": 1, |
| "train_speed(iter/s)": 0.005419 |
| }, |
| { |
| "epoch": 0.01638504864311316, |
| "grad_norm": 4.338643550872803, |
| "learning_rate": 3.0769230769230774e-05, |
| "logits/chosen": -0.28789153695106506, |
| "logits/rejected": -0.44838714599609375, |
| "logps/chosen": -7.449802398681641, |
| "logps/rejected": -18.95979881286621, |
| "loss": 1.2916792631149292, |
| "memory(GiB)": 45.94, |
| "nll_loss": 0.59853196144104, |
| "rewards/accuracies": 0.0, |
| "rewards/chosen": 0.0, |
| "rewards/margins": 0.0, |
| "rewards/rejected": 0.0, |
| "step": 2, |
| "train_speed(iter/s)": 0.005497 |
| }, |
| { |
| "epoch": 0.02457757296466974, |
| "grad_norm": 5.146463394165039, |
| "learning_rate": 4.615384615384616e-05, |
| "logits/chosen": -0.37533068656921387, |
| "logits/rejected": -0.5269488096237183, |
| "logps/chosen": -4.474628925323486, |
| "logps/rejected": -17.257675170898438, |
| "loss": 1.0800914764404297, |
| "memory(GiB)": 45.94, |
| "nll_loss": 0.3937700092792511, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": 0.005001109093427658, |
| "rewards/margins": 0.01398632675409317, |
| "rewards/rejected": -0.008985218591988087, |
| "step": 3, |
| "train_speed(iter/s)": 0.005544 |
| }, |
| { |
| "epoch": 0.03277009728622632, |
| "grad_norm": 3.9697017669677734, |
| "learning_rate": 6.153846153846155e-05, |
| "logits/chosen": -0.3554455041885376, |
| "logits/rejected": -0.4836651384830475, |
| "logps/chosen": -7.087220191955566, |
| "logps/rejected": -16.969388961791992, |
| "loss": 1.2551683187484741, |
| "memory(GiB)": 45.94, |
| "nll_loss": 0.5823322534561157, |
| "rewards/accuracies": 0.59375, |
| "rewards/chosen": -0.01054135151207447, |
| "rewards/margins": 0.04771733283996582, |
| "rewards/rejected": -0.05825868248939514, |
| "step": 4, |
| "train_speed(iter/s)": 0.005563 |
| }, |
| { |
| "epoch": 0.0409626216077829, |
| "grad_norm": 7.601996421813965, |
| "learning_rate": 7.692307692307693e-05, |
| "logits/chosen": -0.3477073013782501, |
| "logits/rejected": -0.46865567564964294, |
| "logps/chosen": -8.432049751281738, |
| "logps/rejected": -20.41954231262207, |
| "loss": 1.2856180667877197, |
| "memory(GiB)": 45.94, |
| "nll_loss": 0.6360925436019897, |
| "rewards/accuracies": 0.609375, |
| "rewards/chosen": -0.10297518223524094, |
| "rewards/margins": 0.15188881754875183, |
| "rewards/rejected": -0.25486400723457336, |
| "step": 5, |
| "train_speed(iter/s)": 0.005571 |
| }, |
| { |
| "epoch": 0.04915514592933948, |
| "grad_norm": 6.308995246887207, |
| "learning_rate": 9.230769230769232e-05, |
| "logits/chosen": -0.3014591932296753, |
| "logits/rejected": -0.4142557978630066, |
| "logps/chosen": -8.621772766113281, |
| "logps/rejected": -19.327842712402344, |
| "loss": 1.2817319631576538, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.6402445435523987, |
| "rewards/accuracies": 0.609375, |
| "rewards/chosen": -0.10090938210487366, |
| "rewards/margins": 0.18325890600681305, |
| "rewards/rejected": -0.2841682732105255, |
| "step": 6, |
| "train_speed(iter/s)": 0.005566 |
| }, |
| { |
| "epoch": 0.05734767025089606, |
| "grad_norm": 4.3832902908325195, |
| "learning_rate": 0.0001076923076923077, |
| "logits/chosen": -0.3800317347049713, |
| "logits/rejected": -0.4671918749809265, |
| "logps/chosen": -5.921639442443848, |
| "logps/rejected": -15.812642097473145, |
| "loss": 1.0748225450515747, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.42862600088119507, |
| "rewards/accuracies": 0.609375, |
| "rewards/chosen": -0.07157032936811447, |
| "rewards/margins": 0.1520296335220337, |
| "rewards/rejected": -0.22359995543956757, |
| "step": 7, |
| "train_speed(iter/s)": 0.005576 |
| }, |
| { |
| "epoch": 0.06554019457245264, |
| "grad_norm": 2.131096601486206, |
| "learning_rate": 0.0001230769230769231, |
| "logits/chosen": -0.326383113861084, |
| "logits/rejected": -0.4235616624355316, |
| "logps/chosen": -7.039735317230225, |
| "logps/rejected": -15.89234733581543, |
| "loss": 1.1329768896102905, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4950546622276306, |
| "rewards/accuracies": 0.640625, |
| "rewards/chosen": 0.02736004814505577, |
| "rewards/margins": 0.13636186718940735, |
| "rewards/rejected": -0.10900183022022247, |
| "step": 8, |
| "train_speed(iter/s)": 0.00558 |
| }, |
| { |
| "epoch": 0.07373271889400922, |
| "grad_norm": 3.154919147491455, |
| "learning_rate": 0.00013846153846153847, |
| "logits/chosen": -0.3660247027873993, |
| "logits/rejected": -0.4395582675933838, |
| "logps/chosen": -7.082914352416992, |
| "logps/rejected": -12.765766143798828, |
| "loss": 1.1050751209259033, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.44545602798461914, |
| "rewards/accuracies": 0.6875, |
| "rewards/chosen": 0.05931762978434563, |
| "rewards/margins": 0.0759875625371933, |
| "rewards/rejected": -0.016669925302267075, |
| "step": 9, |
| "train_speed(iter/s)": 0.005582 |
| }, |
| { |
| "epoch": 0.0819252432155658, |
| "grad_norm": 2.322492837905884, |
| "learning_rate": 0.00015384615384615385, |
| "logits/chosen": -0.3979243040084839, |
| "logits/rejected": -0.4425738453865051, |
| "logps/chosen": -8.025144577026367, |
| "logps/rejected": -15.784225463867188, |
| "loss": 1.075645089149475, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4679878056049347, |
| "rewards/accuracies": 0.828125, |
| "rewards/chosen": 0.048429328948259354, |
| "rewards/margins": 0.1987656205892563, |
| "rewards/rejected": -0.15033629536628723, |
| "step": 10, |
| "train_speed(iter/s)": 0.005587 |
| }, |
| { |
| "epoch": 0.09011776753712238, |
| "grad_norm": 2.226271152496338, |
| "learning_rate": 0.00016923076923076923, |
| "logits/chosen": -0.2428550124168396, |
| "logits/rejected": -0.4068838357925415, |
| "logps/chosen": -5.73712682723999, |
| "logps/rejected": -23.750782012939453, |
| "loss": 0.9072864651679993, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4175054132938385, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.0929398313164711, |
| "rewards/margins": 0.5025829672813416, |
| "rewards/rejected": -0.40964314341545105, |
| "step": 11, |
| "train_speed(iter/s)": 0.005584 |
| }, |
| { |
| "epoch": 0.09831029185867896, |
| "grad_norm": 9.130071640014648, |
| "learning_rate": 0.00018461538461538463, |
| "logits/chosen": -0.3092227280139923, |
| "logits/rejected": -0.4145185053348541, |
| "logps/chosen": -9.414848327636719, |
| "logps/rejected": -22.3564395904541, |
| "loss": 1.266426920890808, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.6574685573577881, |
| "rewards/accuracies": 0.65625, |
| "rewards/chosen": -0.27316582202911377, |
| "rewards/margins": 0.5058421492576599, |
| "rewards/rejected": -0.7790080308914185, |
| "step": 12, |
| "train_speed(iter/s)": 0.005586 |
| }, |
| { |
| "epoch": 0.10650281618023553, |
| "grad_norm": 8.017046928405762, |
| "learning_rate": 0.0002, |
| "logits/chosen": -0.3036244809627533, |
| "logits/rejected": -0.43275412917137146, |
| "logps/chosen": -8.467231750488281, |
| "logps/rejected": -26.61805534362793, |
| "loss": 1.2604392766952515, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.7518106698989868, |
| "rewards/accuracies": 0.78125, |
| "rewards/chosen": -0.2612573206424713, |
| "rewards/margins": 0.6518616676330566, |
| "rewards/rejected": -0.9131190180778503, |
| "step": 13, |
| "train_speed(iter/s)": 0.005587 |
| }, |
| { |
| "epoch": 0.11469534050179211, |
| "grad_norm": 2.1487388610839844, |
| "learning_rate": 0.00019999075218579183, |
| "logits/chosen": -0.2624781131744385, |
| "logits/rejected": -0.3863448202610016, |
| "logps/chosen": -8.31130313873291, |
| "logps/rejected": -26.065853118896484, |
| "loss": 0.9567878246307373, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4835919439792633, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": 0.02007860317826271, |
| "rewards/margins": 0.6282916069030762, |
| "rewards/rejected": -0.6082130074501038, |
| "step": 14, |
| "train_speed(iter/s)": 0.005584 |
| }, |
| { |
| "epoch": 0.1228878648233487, |
| "grad_norm": 3.114581823348999, |
| "learning_rate": 0.00019996301045360873, |
| "logits/chosen": -0.32559511065483093, |
| "logits/rejected": -0.40920543670654297, |
| "logps/chosen": -7.523581027984619, |
| "logps/rejected": -18.561134338378906, |
| "loss": 1.0775632858276367, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5253046751022339, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": -0.007055876310914755, |
| "rewards/margins": 0.419503390789032, |
| "rewards/rejected": -0.42655929923057556, |
| "step": 15, |
| "train_speed(iter/s)": 0.005586 |
| }, |
| { |
| "epoch": 0.13108038914490527, |
| "grad_norm": 3.7939462661743164, |
| "learning_rate": 0.0001999167799344583, |
| "logits/chosen": -0.2907140851020813, |
| "logits/rejected": -0.3809354901313782, |
| "logps/chosen": -6.864745616912842, |
| "logps/rejected": -20.829776763916016, |
| "loss": 1.0188764333724976, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5191283822059631, |
| "rewards/accuracies": 0.765625, |
| "rewards/chosen": 0.03652028366923332, |
| "rewards/margins": 0.6273013353347778, |
| "rewards/rejected": -0.5907810926437378, |
| "step": 16, |
| "train_speed(iter/s)": 0.005586 |
| }, |
| { |
| "epoch": 0.13927291346646187, |
| "grad_norm": 2.7766926288604736, |
| "learning_rate": 0.00019985206917896563, |
| "logits/chosen": -0.3052994906902313, |
| "logits/rejected": -0.37751495838165283, |
| "logps/chosen": -8.09772777557373, |
| "logps/rejected": -18.549510955810547, |
| "loss": 1.1308537721633911, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.564860999584198, |
| "rewards/accuracies": 0.71875, |
| "rewards/chosen": 0.022659722715616226, |
| "rewards/margins": 0.4622555077075958, |
| "rewards/rejected": -0.4395957589149475, |
| "step": 17, |
| "train_speed(iter/s)": 0.005587 |
| }, |
| { |
| "epoch": 0.14746543778801843, |
| "grad_norm": 3.9402501583099365, |
| "learning_rate": 0.00019976889015579165, |
| "logits/chosen": -0.2790537476539612, |
| "logits/rejected": -0.3640768229961395, |
| "logps/chosen": -10.8331298828125, |
| "logps/rejected": -25.086530685424805, |
| "loss": 1.1370517015457153, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.6662458777427673, |
| "rewards/accuracies": 0.796875, |
| "rewards/chosen": -0.06523045897483826, |
| "rewards/margins": 0.8811236619949341, |
| "rewards/rejected": -0.9463540315628052, |
| "step": 18, |
| "train_speed(iter/s)": 0.005586 |
| }, |
| { |
| "epoch": 0.15565796210957503, |
| "grad_norm": 4.501293659210205, |
| "learning_rate": 0.00019966725824941932, |
| "logits/chosen": -0.3320932686328888, |
| "logits/rejected": -0.44169852137565613, |
| "logps/chosen": -10.038064956665039, |
| "logps/rejected": -25.767019271850586, |
| "loss": 1.2291309833526611, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.7401180267333984, |
| "rewards/accuracies": 0.765625, |
| "rewards/chosen": -0.24555523693561554, |
| "rewards/margins": 0.8229343891143799, |
| "rewards/rejected": -1.068489670753479, |
| "step": 19, |
| "train_speed(iter/s)": 0.005589 |
| }, |
| { |
| "epoch": 0.1638504864311316, |
| "grad_norm": 1.8677138090133667, |
| "learning_rate": 0.00019954719225730847, |
| "logits/chosen": -0.28071457147598267, |
| "logits/rejected": -0.4227759838104248, |
| "logps/chosen": -7.934963703155518, |
| "logps/rejected": -24.492511749267578, |
| "loss": 0.956333339214325, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4845323860645294, |
| "rewards/accuracies": 0.765625, |
| "rewards/chosen": -0.04675678163766861, |
| "rewards/margins": 0.7908064126968384, |
| "rewards/rejected": -0.8375633358955383, |
| "step": 20, |
| "train_speed(iter/s)": 0.005589 |
| }, |
| { |
| "epoch": 0.17204301075268819, |
| "grad_norm": 1.5877766609191895, |
| "learning_rate": 0.00019940871438641882, |
| "logits/chosen": -0.3024439215660095, |
| "logits/rejected": -0.43745386600494385, |
| "logps/chosen": -6.4543776512146, |
| "logps/rejected": -22.530338287353516, |
| "loss": 0.9309813380241394, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4551314115524292, |
| "rewards/accuracies": 0.828125, |
| "rewards/chosen": 0.04532763361930847, |
| "rewards/margins": 0.6945431232452393, |
| "rewards/rejected": -0.6492154598236084, |
| "step": 21, |
| "train_speed(iter/s)": 0.005589 |
| }, |
| { |
| "epoch": 0.18023553507424475, |
| "grad_norm": 1.7120730876922607, |
| "learning_rate": 0.00019925185024910277, |
| "logits/chosen": -0.2665388584136963, |
| "logits/rejected": -0.39961740374565125, |
| "logps/chosen": -7.612581729888916, |
| "logps/rejected": -24.12154769897461, |
| "loss": 1.017120599746704, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5679780840873718, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": 0.12790349125862122, |
| "rewards/margins": 0.8537773489952087, |
| "rewards/rejected": -0.7258739471435547, |
| "step": 22, |
| "train_speed(iter/s)": 0.005587 |
| }, |
| { |
| "epoch": 0.18842805939580134, |
| "grad_norm": 1.8607813119888306, |
| "learning_rate": 0.00019907662885836836, |
| "logits/chosen": -0.29434627294540405, |
| "logits/rejected": -0.38500675559043884, |
| "logps/chosen": -9.917289733886719, |
| "logps/rejected": -20.862939834594727, |
| "loss": 1.072438359260559, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5891656875610352, |
| "rewards/accuracies": 0.84375, |
| "rewards/chosen": 0.050024691969156265, |
| "rewards/margins": 0.6231139302253723, |
| "rewards/rejected": -0.5730893015861511, |
| "step": 23, |
| "train_speed(iter/s)": 0.005586 |
| }, |
| { |
| "epoch": 0.1966205837173579, |
| "grad_norm": 2.2947323322296143, |
| "learning_rate": 0.00019888308262251285, |
| "logits/chosen": -0.2984248995780945, |
| "logits/rejected": -0.4263286292552948, |
| "logps/chosen": -5.695917129516602, |
| "logps/rejected": -27.542821884155273, |
| "loss": 0.8626710176467896, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4990408718585968, |
| "rewards/accuracies": 0.859375, |
| "rewards/chosen": 0.05305299907922745, |
| "rewards/margins": 1.2556676864624023, |
| "rewards/rejected": -1.2026147842407227, |
| "step": 24, |
| "train_speed(iter/s)": 0.005587 |
| }, |
| { |
| "epoch": 0.2048131080389145, |
| "grad_norm": 2.453016996383667, |
| "learning_rate": 0.0001986712473391289, |
| "logits/chosen": -0.37013769149780273, |
| "logits/rejected": -0.4342552423477173, |
| "logps/chosen": -5.276243209838867, |
| "logps/rejected": -23.596858978271484, |
| "loss": 0.8703887462615967, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.45367568731307983, |
| "rewards/accuracies": 0.78125, |
| "rewards/chosen": -0.010003786534070969, |
| "rewards/margins": 1.1414563655853271, |
| "rewards/rejected": -1.1514601707458496, |
| "step": 25, |
| "train_speed(iter/s)": 0.005589 |
| }, |
| { |
| "epoch": 0.21300563236047107, |
| "grad_norm": 3.131953477859497, |
| "learning_rate": 0.00019844116218848334, |
| "logits/chosen": -0.3483574688434601, |
| "logits/rejected": -0.4867978096008301, |
| "logps/chosen": -6.131460666656494, |
| "logps/rejected": -25.60129165649414, |
| "loss": 0.8633096814155579, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4731563329696655, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.08134596049785614, |
| "rewards/margins": 1.066148042678833, |
| "rewards/rejected": -0.9848020076751709, |
| "step": 26, |
| "train_speed(iter/s)": 0.005591 |
| }, |
| { |
| "epoch": 0.22119815668202766, |
| "grad_norm": 2.4780352115631104, |
| "learning_rate": 0.00019819286972627066, |
| "logits/chosen": -0.36855828762054443, |
| "logits/rejected": -0.49951156973838806, |
| "logps/chosen": -5.627861499786377, |
| "logps/rejected": -31.835105895996094, |
| "loss": 0.7191118597984314, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.39769047498703003, |
| "rewards/accuracies": 0.859375, |
| "rewards/chosen": -0.007685612887144089, |
| "rewards/margins": 1.6878087520599365, |
| "rewards/rejected": -1.6954944133758545, |
| "step": 27, |
| "train_speed(iter/s)": 0.005593 |
| }, |
| { |
| "epoch": 0.22939068100358423, |
| "grad_norm": 2.9881274700164795, |
| "learning_rate": 0.00019792641587574212, |
| "logits/chosen": -0.2455301731824875, |
| "logits/rejected": -0.39179977774620056, |
| "logps/chosen": -6.589425563812256, |
| "logps/rejected": -36.522743225097656, |
| "loss": 0.7601451277732849, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4516022801399231, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.009008985012769699, |
| "rewards/margins": 1.831702709197998, |
| "rewards/rejected": -1.8226937055587769, |
| "step": 28, |
| "train_speed(iter/s)": 0.005593 |
| }, |
| { |
| "epoch": 0.23758320532514082, |
| "grad_norm": 2.97683048248291, |
| "learning_rate": 0.00019764184991921178, |
| "logits/chosen": -0.30706074833869934, |
| "logits/rejected": -0.44865599274635315, |
| "logps/chosen": -10.063987731933594, |
| "logps/rejected": -31.307193756103516, |
| "loss": 1.0377051830291748, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.6944084763526917, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": -0.10418727993965149, |
| "rewards/margins": 1.5954655408859253, |
| "rewards/rejected": -1.6996527910232544, |
| "step": 29, |
| "train_speed(iter/s)": 0.005594 |
| }, |
| { |
| "epoch": 0.2457757296466974, |
| "grad_norm": 1.9776651859283447, |
| "learning_rate": 0.0001973392244889415, |
| "logits/chosen": -0.32163313031196594, |
| "logits/rejected": -0.4449259042739868, |
| "logps/chosen": -8.357927322387695, |
| "logps/rejected": -33.13959884643555, |
| "loss": 0.880458414554596, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5416579246520996, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": -0.04613646864891052, |
| "rewards/margins": 1.6892749071121216, |
| "rewards/rejected": -1.735411286354065, |
| "step": 30, |
| "train_speed(iter/s)": 0.005595 |
| }, |
| { |
| "epoch": 0.25396825396825395, |
| "grad_norm": 2.918856620788574, |
| "learning_rate": 0.00019701859555740648, |
| "logits/chosen": -0.3250303566455841, |
| "logits/rejected": -0.42262473702430725, |
| "logps/chosen": -7.8114013671875, |
| "logps/rejected": -27.219764709472656, |
| "loss": 0.9608601331710815, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5916098356246948, |
| "rewards/accuracies": 0.859375, |
| "rewards/chosen": 0.02971716970205307, |
| "rewards/margins": 1.5878727436065674, |
| "rewards/rejected": -1.5581557750701904, |
| "step": 31, |
| "train_speed(iter/s)": 0.005596 |
| }, |
| { |
| "epoch": 0.26216077828981055, |
| "grad_norm": 1.8905093669891357, |
| "learning_rate": 0.00019668002242694238, |
| "logits/chosen": -0.2707211673259735, |
| "logits/rejected": -0.4425659477710724, |
| "logps/chosen": -5.811425685882568, |
| "logps/rejected": -30.51934814453125, |
| "loss": 0.7490611672401428, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.3751787841320038, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": 0.11003397405147552, |
| "rewards/margins": 1.2447575330734253, |
| "rewards/rejected": -1.134723424911499, |
| "step": 32, |
| "train_speed(iter/s)": 0.005596 |
| }, |
| { |
| "epoch": 0.27035330261136714, |
| "grad_norm": 3.2241005897521973, |
| "learning_rate": 0.00019632356771877735, |
| "logits/chosen": -0.31523844599723816, |
| "logits/rejected": -0.48877376317977905, |
| "logps/chosen": -7.238929271697998, |
| "logps/rejected": -24.841379165649414, |
| "loss": 0.9657001495361328, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5450810194015503, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.09001664817333221, |
| "rewards/margins": 0.8783948421478271, |
| "rewards/rejected": -0.7883782386779785, |
| "step": 33, |
| "train_speed(iter/s)": 0.005596 |
| }, |
| { |
| "epoch": 0.27854582693292373, |
| "grad_norm": 2.6415717601776123, |
| "learning_rate": 0.00019594929736144976, |
| "logits/chosen": -0.370350182056427, |
| "logits/rejected": -0.48758062720298767, |
| "logps/chosen": -6.904268741607666, |
| "logps/rejected": -27.709651947021484, |
| "loss": 0.7636420130729675, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.37949565052986145, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": 0.012274645268917084, |
| "rewards/margins": 1.4130356311798096, |
| "rewards/rejected": -1.4007611274719238, |
| "step": 34, |
| "train_speed(iter/s)": 0.005596 |
| }, |
| { |
| "epoch": 0.2867383512544803, |
| "grad_norm": 1.870375156402588, |
| "learning_rate": 0.0001955572805786141, |
| "logits/chosen": -0.32100898027420044, |
| "logits/rejected": -0.5096184611320496, |
| "logps/chosen": -5.18679666519165, |
| "logps/rejected": -32.468994140625, |
| "loss": 0.7190079689025879, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4115956127643585, |
| "rewards/accuracies": 0.890625, |
| "rewards/chosen": 0.07185505330562592, |
| "rewards/margins": 1.6462374925613403, |
| "rewards/rejected": -1.5743825435638428, |
| "step": 35, |
| "train_speed(iter/s)": 0.005597 |
| }, |
| { |
| "epoch": 0.29493087557603687, |
| "grad_norm": 6.671431064605713, |
| "learning_rate": 0.00019514758987623784, |
| "logits/chosen": -0.34625619649887085, |
| "logits/rejected": -0.4750059247016907, |
| "logps/chosen": -10.31639575958252, |
| "logps/rejected": -36.582740783691406, |
| "loss": 1.0590507984161377, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.6509675979614258, |
| "rewards/accuracies": 0.796875, |
| "rewards/chosen": -0.20210294425487518, |
| "rewards/margins": 1.9486511945724487, |
| "rewards/rejected": -2.15075421333313, |
| "step": 36, |
| "train_speed(iter/s)": 0.005597 |
| }, |
| { |
| "epoch": 0.30312339989759346, |
| "grad_norm": 7.133448123931885, |
| "learning_rate": 0.000194720301029191, |
| "logits/chosen": -0.3258736729621887, |
| "logits/rejected": -0.5075902938842773, |
| "logps/chosen": -9.880843162536621, |
| "logps/rejected": -39.184967041015625, |
| "loss": 1.0690901279449463, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.7646387815475464, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": -0.22141867876052856, |
| "rewards/margins": 2.19596004486084, |
| "rewards/rejected": -2.4173786640167236, |
| "step": 37, |
| "train_speed(iter/s)": 0.005598 |
| }, |
| { |
| "epoch": 0.31131592421915005, |
| "grad_norm": 3.9345178604125977, |
| "learning_rate": 0.00019427549306723097, |
| "logits/chosen": -0.32678377628326416, |
| "logits/rejected": -0.5091384649276733, |
| "logps/chosen": -6.220644950866699, |
| "logps/rejected": -38.1715087890625, |
| "loss": 0.7903810143470764, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5349410772323608, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": 0.04198794811964035, |
| "rewards/margins": 2.360246419906616, |
| "rewards/rejected": -2.31825852394104, |
| "step": 38, |
| "train_speed(iter/s)": 0.005599 |
| }, |
| { |
| "epoch": 0.3195084485407066, |
| "grad_norm": 2.339552402496338, |
| "learning_rate": 0.0001938132482603856, |
| "logits/chosen": -0.24855449795722961, |
| "logits/rejected": -0.46237537264823914, |
| "logps/chosen": -4.8696136474609375, |
| "logps/rejected": -35.97483825683594, |
| "loss": 0.6593115925788879, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.3662545680999756, |
| "rewards/accuracies": 0.9375, |
| "rewards/chosen": 0.0723145455121994, |
| "rewards/margins": 1.8895220756530762, |
| "rewards/rejected": -1.8172073364257812, |
| "step": 39, |
| "train_speed(iter/s)": 0.005599 |
| }, |
| { |
| "epoch": 0.3277009728622632, |
| "grad_norm": 2.3479816913604736, |
| "learning_rate": 0.0001933336521037367, |
| "logits/chosen": -0.3132964074611664, |
| "logits/rejected": -0.4357830882072449, |
| "logps/chosen": -7.038135528564453, |
| "logps/rejected": -29.72612190246582, |
| "loss": 0.7142176628112793, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4181695878505707, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": 0.04734823852777481, |
| "rewards/margins": 1.5952571630477905, |
| "rewards/rejected": -1.5479090213775635, |
| "step": 40, |
| "train_speed(iter/s)": 0.005599 |
| }, |
| { |
| "epoch": 0.3358934971838198, |
| "grad_norm": 2.551851272583008, |
| "learning_rate": 0.00019283679330160726, |
| "logits/chosen": -0.32567888498306274, |
| "logits/rejected": -0.4566379487514496, |
| "logps/chosen": -7.564522743225098, |
| "logps/rejected": -29.297645568847656, |
| "loss": 0.8506546020507812, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4627075493335724, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": -0.018498025834560394, |
| "rewards/margins": 1.5199252367019653, |
| "rewards/rejected": -1.5384231805801392, |
| "step": 41, |
| "train_speed(iter/s)": 0.005599 |
| }, |
| { |
| "epoch": 0.34408602150537637, |
| "grad_norm": 1.9277410507202148, |
| "learning_rate": 0.00019232276375115515, |
| "logits/chosen": -0.33336636424064636, |
| "logits/rejected": -0.4738547205924988, |
| "logps/chosen": -6.605499267578125, |
| "logps/rejected": -26.275497436523438, |
| "loss": 0.7640136480331421, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.45767274498939514, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": 0.18768644332885742, |
| "rewards/margins": 1.589571237564087, |
| "rewards/rejected": -1.4018847942352295, |
| "step": 42, |
| "train_speed(iter/s)": 0.0056 |
| }, |
| { |
| "epoch": 0.3522785458269329, |
| "grad_norm": 2.7057321071624756, |
| "learning_rate": 0.00019179165852537596, |
| "logits/chosen": -0.24734170734882355, |
| "logits/rejected": -0.4288603663444519, |
| "logps/chosen": -8.57066822052002, |
| "logps/rejected": -35.444984436035156, |
| "loss": 0.9440370798110962, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.6365867853164673, |
| "rewards/accuracies": 0.8125, |
| "rewards/chosen": -0.04400603100657463, |
| "rewards/margins": 1.8436404466629028, |
| "rewards/rejected": -1.8876465559005737, |
| "step": 43, |
| "train_speed(iter/s)": 0.0056 |
| }, |
| { |
| "epoch": 0.3604710701484895, |
| "grad_norm": 2.4904823303222656, |
| "learning_rate": 0.00019124357585551872, |
| "logits/chosen": -0.296947717666626, |
| "logits/rejected": -0.43239718675613403, |
| "logps/chosen": -8.56611442565918, |
| "logps/rejected": -36.052391052246094, |
| "loss": 0.8505915403366089, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.5557512044906616, |
| "rewards/accuracies": 0.875, |
| "rewards/chosen": 0.0323612280189991, |
| "rewards/margins": 2.1337056159973145, |
| "rewards/rejected": -2.1013448238372803, |
| "step": 44, |
| "train_speed(iter/s)": 0.0056 |
| }, |
| { |
| "epoch": 0.3686635944700461, |
| "grad_norm": 2.8696656227111816, |
| "learning_rate": 0.00019067861711291744, |
| "logits/chosen": -0.255009263753891, |
| "logits/rejected": -0.4662845730781555, |
| "logps/chosen": -4.24545431137085, |
| "logps/rejected": -36.80644607543945, |
| "loss": 0.642361044883728, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.373349666595459, |
| "rewards/accuracies": 0.921875, |
| "rewards/chosen": 0.03832457214593887, |
| "rewards/margins": 2.0462074279785156, |
| "rewards/rejected": -2.007882833480835, |
| "step": 45, |
| "train_speed(iter/s)": 0.0056 |
| }, |
| { |
| "epoch": 0.3768561187916027, |
| "grad_norm": 2.3527863025665283, |
| "learning_rate": 0.0001900968867902419, |
| "logits/chosen": -0.29511338472366333, |
| "logits/rejected": -0.4300462305545807, |
| "logps/chosen": -6.108550071716309, |
| "logps/rejected": -30.907669067382812, |
| "loss": 0.8321690559387207, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.4702669382095337, |
| "rewards/accuracies": 0.84375, |
| "rewards/chosen": -0.013932769186794758, |
| "rewards/margins": 1.64110267162323, |
| "rewards/rejected": -1.6550354957580566, |
| "step": 46, |
| "train_speed(iter/s)": 0.0056 |
| }, |
| { |
| "epoch": 0.38504864311315923, |
| "grad_norm": 1.8884741067886353, |
| "learning_rate": 0.000189498492482171, |
| "logits/chosen": -0.28806138038635254, |
| "logits/rejected": -0.43808045983314514, |
| "logps/chosen": -5.240577697753906, |
| "logps/rejected": -29.244850158691406, |
| "loss": 0.6651690602302551, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.3822129964828491, |
| "rewards/accuracies": 0.90625, |
| "rewards/chosen": 0.10050319880247116, |
| "rewards/margins": 1.6757993698120117, |
| "rewards/rejected": -1.5752960443496704, |
| "step": 47, |
| "train_speed(iter/s)": 0.005601 |
| }, |
| { |
| "epoch": 0.3932411674347158, |
| "grad_norm": 1.7884507179260254, |
| "learning_rate": 0.00018888354486549237, |
| "logits/chosen": -0.29420942068099976, |
| "logits/rejected": -0.43043237924575806, |
| "logps/chosen": -6.428597450256348, |
| "logps/rejected": -32.303741455078125, |
| "loss": 0.700837254524231, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.45000576972961426, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": 0.12149131298065186, |
| "rewards/margins": 1.8844413757324219, |
| "rewards/rejected": -1.7629501819610596, |
| "step": 48, |
| "train_speed(iter/s)": 0.005601 |
| }, |
| { |
| "epoch": 0.4014336917562724, |
| "grad_norm": 4.318137168884277, |
| "learning_rate": 0.00018825215767863214, |
| "logits/chosen": -0.18930676579475403, |
| "logits/rejected": -0.40716230869293213, |
| "logps/chosen": -5.259353160858154, |
| "logps/rejected": -37.35228729248047, |
| "loss": 0.7124305963516235, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.43444421887397766, |
| "rewards/accuracies": 0.953125, |
| "rewards/chosen": 0.09258248656988144, |
| "rewards/margins": 2.3026676177978516, |
| "rewards/rejected": -2.210084915161133, |
| "step": 49, |
| "train_speed(iter/s)": 0.005601 |
| }, |
| { |
| "epoch": 0.409626216077829, |
| "grad_norm": 4.764008522033691, |
| "learning_rate": 0.0001876044477006183, |
| "logits/chosen": -0.14179718494415283, |
| "logits/rejected": -0.313814252614975, |
| "logps/chosen": -8.565025329589844, |
| "logps/rejected": -39.398197174072266, |
| "loss": 1.0041635036468506, |
| "memory(GiB)": 46.69, |
| "nll_loss": 0.6235343217849731, |
| "rewards/accuracies": 0.859375, |
| "rewards/chosen": -0.16454477608203888, |
| "rewards/margins": 2.0170154571533203, |
| "rewards/rejected": -2.1815602779388428, |
| "step": 50, |
| "train_speed(iter/s)": 0.0056 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 244, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 2, |
| "save_steps": 50, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 5.990563570096538e+16, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
