| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 50, | |
| "global_step": 478, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02092050209205021, | |
| "grad_norm": 6.4084427221095295, | |
| "learning_rate": 1.0416666666666667e-07, | |
| "logits/chosen": -2.630444049835205, | |
| "logits/rejected": -2.576719045639038, | |
| "logps/chosen": -288.65911865234375, | |
| "logps/rejected": -275.90252685546875, | |
| "loss": 0.6931, | |
| "rewards/accuracies": 0.4312500059604645, | |
| "rewards/chosen": 0.00032657841802574694, | |
| "rewards/margins": 0.0008425033884122968, | |
| "rewards/rejected": -0.0005159247666597366, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04184100418410042, | |
| "grad_norm": 5.7047484713043755, | |
| "learning_rate": 2.0833333333333333e-07, | |
| "logits/chosen": -2.644824981689453, | |
| "logits/rejected": -2.6137185096740723, | |
| "logps/chosen": -293.5597839355469, | |
| "logps/rejected": -259.2336730957031, | |
| "loss": 0.6926, | |
| "rewards/accuracies": 0.606249988079071, | |
| "rewards/chosen": 0.002616675803437829, | |
| "rewards/margins": 0.001477306941524148, | |
| "rewards/rejected": 0.0011393685126677155, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.06276150627615062, | |
| "grad_norm": 5.335002677652247, | |
| "learning_rate": 3.1249999999999997e-07, | |
| "logits/chosen": -2.666045665740967, | |
| "logits/rejected": -2.5890631675720215, | |
| "logps/chosen": -294.64007568359375, | |
| "logps/rejected": -287.18695068359375, | |
| "loss": 0.6901, | |
| "rewards/accuracies": 0.59375, | |
| "rewards/chosen": 0.015856895595788956, | |
| "rewards/margins": 0.008159220218658447, | |
| "rewards/rejected": 0.0076976739801466465, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.08368200836820083, | |
| "grad_norm": 5.361991577232885, | |
| "learning_rate": 4.1666666666666667e-07, | |
| "logits/chosen": -2.6365890502929688, | |
| "logits/rejected": -2.5537407398223877, | |
| "logps/chosen": -270.41375732421875, | |
| "logps/rejected": -240.17959594726562, | |
| "loss": 0.6825, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": 0.03482341766357422, | |
| "rewards/margins": 0.02616509422659874, | |
| "rewards/rejected": 0.008658323436975479, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.10460251046025104, | |
| "grad_norm": 6.225003725839748, | |
| "learning_rate": 4.999733114418725e-07, | |
| "logits/chosen": -2.5800583362579346, | |
| "logits/rejected": -2.571406364440918, | |
| "logps/chosen": -264.10205078125, | |
| "logps/rejected": -246.74868774414062, | |
| "loss": 0.6687, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": 0.013260206207633018, | |
| "rewards/margins": 0.07922474294900894, | |
| "rewards/rejected": -0.06596453487873077, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.10460251046025104, | |
| "eval_logits/chosen": -2.615793228149414, | |
| "eval_logits/rejected": -2.5782690048217773, | |
| "eval_logps/chosen": -260.0484313964844, | |
| "eval_logps/rejected": -271.3634033203125, | |
| "eval_loss": 0.649348258972168, | |
| "eval_rewards/accuracies": 0.70703125, | |
| "eval_rewards/chosen": 0.025814848020672798, | |
| "eval_rewards/margins": 0.11282772570848465, | |
| "eval_rewards/rejected": -0.0870128720998764, | |
| "eval_runtime": 103.335, | |
| "eval_samples_per_second": 19.355, | |
| "eval_steps_per_second": 0.31, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.12552301255230125, | |
| "grad_norm": 9.898033994957384, | |
| "learning_rate": 4.990398100856366e-07, | |
| "logits/chosen": -2.543462038040161, | |
| "logits/rejected": -2.50410795211792, | |
| "logps/chosen": -268.82574462890625, | |
| "logps/rejected": -262.0675964355469, | |
| "loss": 0.6502, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.010042434558272362, | |
| "rewards/margins": 0.14874781668186188, | |
| "rewards/rejected": -0.15879027545452118, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.14644351464435146, | |
| "grad_norm": 9.743629526270734, | |
| "learning_rate": 4.967775735898179e-07, | |
| "logits/chosen": -2.6071839332580566, | |
| "logits/rejected": -2.5465664863586426, | |
| "logps/chosen": -304.88092041015625, | |
| "logps/rejected": -310.2099609375, | |
| "loss": 0.6315, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.198007270693779, | |
| "rewards/margins": 0.2037646323442459, | |
| "rewards/rejected": -0.4017719328403473, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.16736401673640167, | |
| "grad_norm": 15.248138814448389, | |
| "learning_rate": 4.931986719649298e-07, | |
| "logits/chosen": -1.8189195394515991, | |
| "logits/rejected": -1.8181097507476807, | |
| "logps/chosen": -303.1712951660156, | |
| "logps/rejected": -337.1569519042969, | |
| "loss": 0.6003, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": -0.4402276873588562, | |
| "rewards/margins": 0.30723443627357483, | |
| "rewards/rejected": -0.7474621534347534, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18828451882845187, | |
| "grad_norm": 14.823998474977419, | |
| "learning_rate": 4.883222001996351e-07, | |
| "logits/chosen": -0.9793803095817566, | |
| "logits/rejected": -0.8350278735160828, | |
| "logps/chosen": -324.31463623046875, | |
| "logps/rejected": -369.8716735839844, | |
| "loss": 0.5849, | |
| "rewards/accuracies": 0.65625, | |
| "rewards/chosen": -0.5321702361106873, | |
| "rewards/margins": 0.5584505796432495, | |
| "rewards/rejected": -1.090620756149292, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20920502092050208, | |
| "grad_norm": 22.816637306761073, | |
| "learning_rate": 4.821741763807186e-07, | |
| "logits/chosen": -0.4131926894187927, | |
| "logits/rejected": 0.20026779174804688, | |
| "logps/chosen": -381.84002685546875, | |
| "logps/rejected": -387.60943603515625, | |
| "loss": 0.5614, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.8785870671272278, | |
| "rewards/margins": 0.7347825765609741, | |
| "rewards/rejected": -1.6133695840835571, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.20920502092050208, | |
| "eval_logits/chosen": -0.5290488004684448, | |
| "eval_logits/rejected": -0.22265684604644775, | |
| "eval_logps/chosen": -343.2120361328125, | |
| "eval_logps/rejected": -421.86474609375, | |
| "eval_loss": 0.5806925892829895, | |
| "eval_rewards/accuracies": 0.7109375, | |
| "eval_rewards/chosen": -0.8058211803436279, | |
| "eval_rewards/margins": 0.7862052321434021, | |
| "eval_rewards/rejected": -1.5920264720916748, | |
| "eval_runtime": 104.2044, | |
| "eval_samples_per_second": 19.193, | |
| "eval_steps_per_second": 0.307, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2301255230125523, | |
| "grad_norm": 19.02164952454875, | |
| "learning_rate": 4.747874028753375e-07, | |
| "logits/chosen": -0.43088096380233765, | |
| "logits/rejected": -0.01379423774778843, | |
| "logps/chosen": -391.91552734375, | |
| "logps/rejected": -446.66961669921875, | |
| "loss": 0.5589, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.9717921018600464, | |
| "rewards/margins": 0.7386372089385986, | |
| "rewards/rejected": -1.7104294300079346, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2510460251046025, | |
| "grad_norm": 23.262987016748436, | |
| "learning_rate": 4.662012913161997e-07, | |
| "logits/chosen": 0.030251333490014076, | |
| "logits/rejected": 0.5855879187583923, | |
| "logps/chosen": -378.17999267578125, | |
| "logps/rejected": -441.85406494140625, | |
| "loss": 0.5556, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.0160324573516846, | |
| "rewards/margins": 1.018049955368042, | |
| "rewards/rejected": -2.0340826511383057, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2719665271966527, | |
| "grad_norm": 21.032912122052533, | |
| "learning_rate": 4.5646165232345103e-07, | |
| "logits/chosen": -0.32387033104896545, | |
| "logits/rejected": 0.246691033244133, | |
| "logps/chosen": -401.41961669921875, | |
| "logps/rejected": -456.65216064453125, | |
| "loss": 0.542, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -1.039526343345642, | |
| "rewards/margins": 0.8820648193359375, | |
| "rewards/rejected": -1.9215911626815796, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2928870292887029, | |
| "grad_norm": 18.49329811567727, | |
| "learning_rate": 4.456204510851956e-07, | |
| "logits/chosen": -0.7103713154792786, | |
| "logits/rejected": -0.07352075725793839, | |
| "logps/chosen": -398.73785400390625, | |
| "logps/rejected": -456.5108337402344, | |
| "loss": 0.5449, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -0.9843541979789734, | |
| "rewards/margins": 0.8879961967468262, | |
| "rewards/rejected": -1.8723503351211548, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3138075313807531, | |
| "grad_norm": 22.991798597203633, | |
| "learning_rate": 4.337355301007335e-07, | |
| "logits/chosen": -0.8071734309196472, | |
| "logits/rejected": 0.08783279359340668, | |
| "logps/chosen": -380.7442321777344, | |
| "logps/rejected": -434.64959716796875, | |
| "loss": 0.5419, | |
| "rewards/accuracies": 0.6937500238418579, | |
| "rewards/chosen": -0.8969854116439819, | |
| "rewards/margins": 0.7560001611709595, | |
| "rewards/rejected": -1.6529855728149414, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3138075313807531, | |
| "eval_logits/chosen": -0.0013768002390861511, | |
| "eval_logits/rejected": 0.6415377259254456, | |
| "eval_logps/chosen": -367.39569091796875, | |
| "eval_logps/rejected": -469.2164611816406, | |
| "eval_loss": 0.5584754347801208, | |
| "eval_rewards/accuracies": 0.74609375, | |
| "eval_rewards/chosen": -1.0476573705673218, | |
| "eval_rewards/margins": 1.0178861618041992, | |
| "eval_rewards/rejected": -2.0655436515808105, | |
| "eval_runtime": 102.1901, | |
| "eval_samples_per_second": 19.571, | |
| "eval_steps_per_second": 0.313, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33472803347280333, | |
| "grad_norm": 17.28842262569855, | |
| "learning_rate": 4.2087030056579986e-07, | |
| "logits/chosen": 0.42611104249954224, | |
| "logits/rejected": 1.240443468093872, | |
| "logps/chosen": -371.35699462890625, | |
| "logps/rejected": -441.8954162597656, | |
| "loss": 0.5371, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.0083401203155518, | |
| "rewards/margins": 1.0140050649642944, | |
| "rewards/rejected": -2.0223450660705566, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.35564853556485354, | |
| "grad_norm": 23.711859022455013, | |
| "learning_rate": 4.070934040463998e-07, | |
| "logits/chosen": 0.9712627530097961, | |
| "logits/rejected": 1.8879001140594482, | |
| "logps/chosen": -393.61029052734375, | |
| "logps/rejected": -474.1710510253906, | |
| "loss": 0.5201, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.2388116121292114, | |
| "rewards/margins": 0.9170275926589966, | |
| "rewards/rejected": -2.155838966369629, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.37656903765690375, | |
| "grad_norm": 20.4986556938231, | |
| "learning_rate": 3.9247834624635404e-07, | |
| "logits/chosen": 1.3909448385238647, | |
| "logits/rejected": 2.293593406677246, | |
| "logps/chosen": -461.04046630859375, | |
| "logps/rejected": -531.3743286132812, | |
| "loss": 0.534, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.5673408508300781, | |
| "rewards/margins": 1.0537548065185547, | |
| "rewards/rejected": -2.621096134185791, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.39748953974895396, | |
| "grad_norm": 29.28443173248426, | |
| "learning_rate": 3.7710310482256523e-07, | |
| "logits/chosen": 0.8631278276443481, | |
| "logits/rejected": 1.6216942071914673, | |
| "logps/chosen": -393.5219421386719, | |
| "logps/rejected": -465.5166015625, | |
| "loss": 0.5181, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.397423505783081, | |
| "rewards/margins": 0.8221076726913452, | |
| "rewards/rejected": -2.2195310592651367, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.41841004184100417, | |
| "grad_norm": 24.61839368474693, | |
| "learning_rate": 3.610497133404795e-07, | |
| "logits/chosen": 1.1675374507904053, | |
| "logits/rejected": 2.012094020843506, | |
| "logps/chosen": -408.1467590332031, | |
| "logps/rejected": -462.46875, | |
| "loss": 0.526, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.4781768321990967, | |
| "rewards/margins": 0.8430646657943726, | |
| "rewards/rejected": -2.3212413787841797, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.41841004184100417, | |
| "eval_logits/chosen": 0.9801958799362183, | |
| "eval_logits/rejected": 1.7427237033843994, | |
| "eval_logps/chosen": -402.51995849609375, | |
| "eval_logps/rejected": -517.015625, | |
| "eval_loss": 0.5562114715576172, | |
| "eval_rewards/accuracies": 0.76171875, | |
| "eval_rewards/chosen": -1.3989005088806152, | |
| "eval_rewards/margins": 1.1446349620819092, | |
| "eval_rewards/rejected": -2.5435354709625244, | |
| "eval_runtime": 106.3821, | |
| "eval_samples_per_second": 18.8, | |
| "eval_steps_per_second": 0.301, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.4393305439330544, | |
| "grad_norm": 24.016781344152104, | |
| "learning_rate": 3.4440382358952115e-07, | |
| "logits/chosen": 1.11398446559906, | |
| "logits/rejected": 2.043394088745117, | |
| "logps/chosen": -427.09759521484375, | |
| "logps/rejected": -499.5668029785156, | |
| "loss": 0.5271, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.5520648956298828, | |
| "rewards/margins": 0.8630266189575195, | |
| "rewards/rejected": -2.4150915145874023, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4602510460251046, | |
| "grad_norm": 19.623224068729442, | |
| "learning_rate": 3.272542485937368e-07, | |
| "logits/chosen": 0.09422092139720917, | |
| "logits/rejected": 0.9946798086166382, | |
| "logps/chosen": -387.312255859375, | |
| "logps/rejected": -478.73699951171875, | |
| "loss": 0.5347, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.0305876731872559, | |
| "rewards/margins": 1.044886827468872, | |
| "rewards/rejected": -2.075474262237549, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4811715481171548, | |
| "grad_norm": 20.279694061078374, | |
| "learning_rate": 3.096924887558854e-07, | |
| "logits/chosen": 0.5403207540512085, | |
| "logits/rejected": 1.6011781692504883, | |
| "logps/chosen": -414.5859375, | |
| "logps/rejected": -522.0372314453125, | |
| "loss": 0.4963, | |
| "rewards/accuracies": 0.75, | |
| "rewards/chosen": -1.325188398361206, | |
| "rewards/margins": 1.3052270412445068, | |
| "rewards/rejected": -2.630415439605713, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.502092050209205, | |
| "grad_norm": 17.600900176176182, | |
| "learning_rate": 2.9181224366319943e-07, | |
| "logits/chosen": 1.2316529750823975, | |
| "logits/rejected": 2.2336225509643555, | |
| "logps/chosen": -405.16815185546875, | |
| "logps/rejected": -461.0508728027344, | |
| "loss": 0.5208, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.505174994468689, | |
| "rewards/margins": 0.9252998232841492, | |
| "rewards/rejected": -2.4304747581481934, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5230125523012552, | |
| "grad_norm": 21.388657517716165, | |
| "learning_rate": 2.7370891215954565e-07, | |
| "logits/chosen": 0.33056747913360596, | |
| "logits/rejected": 1.398674726486206, | |
| "logps/chosen": -425.12738037109375, | |
| "logps/rejected": -477.9986267089844, | |
| "loss": 0.5202, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.3082549571990967, | |
| "rewards/margins": 1.007277011871338, | |
| "rewards/rejected": -2.3155319690704346, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5230125523012552, | |
| "eval_logits/chosen": 0.6489181518554688, | |
| "eval_logits/rejected": 1.4379549026489258, | |
| "eval_logps/chosen": -376.8782653808594, | |
| "eval_logps/rejected": -495.4537353515625, | |
| "eval_loss": 0.5419028401374817, | |
| "eval_rewards/accuracies": 0.7890625, | |
| "eval_rewards/chosen": -1.142483115196228, | |
| "eval_rewards/margins": 1.185433030128479, | |
| "eval_rewards/rejected": -2.327916383743286, | |
| "eval_runtime": 104.2974, | |
| "eval_samples_per_second": 19.176, | |
| "eval_steps_per_second": 0.307, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5439330543933054, | |
| "grad_norm": 23.92722986686011, | |
| "learning_rate": 2.55479083351317e-07, | |
| "logits/chosen": 1.0843138694763184, | |
| "logits/rejected": 2.0778467655181885, | |
| "logps/chosen": -430.15911865234375, | |
| "logps/rejected": -504.9864807128906, | |
| "loss": 0.5125, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -1.1564133167266846, | |
| "rewards/margins": 1.2437217235565186, | |
| "rewards/rejected": -2.4001352787017822, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.5648535564853556, | |
| "grad_norm": 21.06560694930431, | |
| "learning_rate": 2.3722002126275822e-07, | |
| "logits/chosen": 1.3407487869262695, | |
| "logits/rejected": 2.4112818241119385, | |
| "logps/chosen": -432.30023193359375, | |
| "logps/rejected": -504.80670166015625, | |
| "loss": 0.5096, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -1.3888903856277466, | |
| "rewards/margins": 1.141367793083191, | |
| "rewards/rejected": -2.5302579402923584, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.5857740585774058, | |
| "grad_norm": 21.8560737042057, | |
| "learning_rate": 2.19029145890313e-07, | |
| "logits/chosen": 1.2237210273742676, | |
| "logits/rejected": 2.0341880321502686, | |
| "logps/chosen": -389.9634704589844, | |
| "logps/rejected": -492.298828125, | |
| "loss": 0.5114, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.3260904550552368, | |
| "rewards/margins": 1.0742130279541016, | |
| "rewards/rejected": -2.400303363800049, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.606694560669456, | |
| "grad_norm": 23.993372955897243, | |
| "learning_rate": 2.0100351342479216e-07, | |
| "logits/chosen": 1.5786670446395874, | |
| "logits/rejected": 2.3115456104278564, | |
| "logps/chosen": -447.48382568359375, | |
| "logps/rejected": -546.939697265625, | |
| "loss": 0.513, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -1.5614277124404907, | |
| "rewards/margins": 1.1185188293457031, | |
| "rewards/rejected": -2.6799466609954834, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6276150627615062, | |
| "grad_norm": 22.907667503889392, | |
| "learning_rate": 1.8323929841460178e-07, | |
| "logits/chosen": 2.0128085613250732, | |
| "logits/rejected": 2.6608686447143555, | |
| "logps/chosen": -434.07830810546875, | |
| "logps/rejected": -551.2879028320312, | |
| "loss": 0.5054, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.6475236415863037, | |
| "rewards/margins": 1.0819259881973267, | |
| "rewards/rejected": -2.72944974899292, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6276150627615062, | |
| "eval_logits/chosen": 1.477131962776184, | |
| "eval_logits/rejected": 2.255997657775879, | |
| "eval_logps/chosen": -402.4423522949219, | |
| "eval_logps/rejected": -531.4893798828125, | |
| "eval_loss": 0.5449927449226379, | |
| "eval_rewards/accuracies": 0.77734375, | |
| "eval_rewards/chosen": -1.3981244564056396, | |
| "eval_rewards/margins": 1.290148377418518, | |
| "eval_rewards/rejected": -2.688272714614868, | |
| "eval_runtime": 103.6153, | |
| "eval_samples_per_second": 19.302, | |
| "eval_steps_per_second": 0.309, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6485355648535565, | |
| "grad_norm": 20.97030138169365, | |
| "learning_rate": 1.6583128063291573e-07, | |
| "logits/chosen": 1.2844688892364502, | |
| "logits/rejected": 2.06947660446167, | |
| "logps/chosen": -437.86358642578125, | |
| "logps/rejected": -538.4571533203125, | |
| "loss": 0.5145, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -1.4390218257904053, | |
| "rewards/margins": 1.1747257709503174, | |
| "rewards/rejected": -2.6137475967407227, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6694560669456067, | |
| "grad_norm": 23.361486108617427, | |
| "learning_rate": 1.488723393865766e-07, | |
| "logits/chosen": 1.0614537000656128, | |
| "logits/rejected": 2.0729637145996094, | |
| "logps/chosen": -453.379150390625, | |
| "logps/rejected": -521.8544311523438, | |
| "loss": 0.4923, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.6114938259124756, | |
| "rewards/margins": 1.0873934030532837, | |
| "rewards/rejected": -2.6988871097564697, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.6903765690376569, | |
| "grad_norm": 24.156654265698716, | |
| "learning_rate": 1.3245295796480788e-07, | |
| "logits/chosen": 1.4917339086532593, | |
| "logits/rejected": 2.5346505641937256, | |
| "logps/chosen": -463.26202392578125, | |
| "logps/rejected": -520.6095581054688, | |
| "loss": 0.5059, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.7963454723358154, | |
| "rewards/margins": 0.9912670254707336, | |
| "rewards/rejected": -2.7876124382019043, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7112970711297071, | |
| "grad_norm": 19.422370738594758, | |
| "learning_rate": 1.1666074087171627e-07, | |
| "logits/chosen": 1.7316112518310547, | |
| "logits/rejected": 2.561483144760132, | |
| "logps/chosen": -413.0133361816406, | |
| "logps/rejected": -536.4874877929688, | |
| "loss": 0.5184, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -1.6657421588897705, | |
| "rewards/margins": 1.103548288345337, | |
| "rewards/rejected": -2.7692904472351074, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7322175732217573, | |
| "grad_norm": 19.236548541895125, | |
| "learning_rate": 1.0157994641835734e-07, | |
| "logits/chosen": 1.52297842502594, | |
| "logits/rejected": 2.436295986175537, | |
| "logps/chosen": -450.27239990234375, | |
| "logps/rejected": -541.5868530273438, | |
| "loss": 0.497, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -1.680456519126892, | |
| "rewards/margins": 1.0492502450942993, | |
| "rewards/rejected": -2.7297065258026123, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7322175732217573, | |
| "eval_logits/chosen": 1.3703731298446655, | |
| "eval_logits/rejected": 2.225933074951172, | |
| "eval_logps/chosen": -422.6754455566406, | |
| "eval_logps/rejected": -549.4119873046875, | |
| "eval_loss": 0.5301805138587952, | |
| "eval_rewards/accuracies": 0.7734375, | |
| "eval_rewards/chosen": -1.6004550457000732, | |
| "eval_rewards/margins": 1.2670434713363647, | |
| "eval_rewards/rejected": -2.8674986362457275, | |
| "eval_runtime": 102.8183, | |
| "eval_samples_per_second": 19.452, | |
| "eval_steps_per_second": 0.311, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.7531380753138075, | |
| "grad_norm": 19.689504951232507, | |
| "learning_rate": 8.729103716819111e-08, | |
| "logits/chosen": 1.695031762123108, | |
| "logits/rejected": 2.4951744079589844, | |
| "logps/chosen": -429.60675048828125, | |
| "logps/rejected": -540.1343383789062, | |
| "loss": 0.5127, | |
| "rewards/accuracies": 0.737500011920929, | |
| "rewards/chosen": -1.8274829387664795, | |
| "rewards/margins": 1.1173667907714844, | |
| "rewards/rejected": -2.9448494911193848, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.7740585774058577, | |
| "grad_norm": 23.998324017815545, | |
| "learning_rate": 7.387025063449081e-08, | |
| "logits/chosen": 1.3235762119293213, | |
| "logits/rejected": 2.6002328395843506, | |
| "logps/chosen": -454.5791015625, | |
| "logps/rejected": -572.55078125, | |
| "loss": 0.495, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.667346715927124, | |
| "rewards/margins": 1.3339135646820068, | |
| "rewards/rejected": -3.001260280609131, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.7949790794979079, | |
| "grad_norm": 20.911379608221466, | |
| "learning_rate": 6.138919252022435e-08, | |
| "logits/chosen": 1.435396432876587, | |
| "logits/rejected": 2.2644729614257812, | |
| "logps/chosen": -481.06378173828125, | |
| "logps/rejected": -542.4463500976562, | |
| "loss": 0.5119, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.6459989547729492, | |
| "rewards/margins": 0.9545661211013794, | |
| "rewards/rejected": -2.600564956665039, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8158995815899581, | |
| "grad_norm": 33.66392923940902, | |
| "learning_rate": 4.991445467064689e-08, | |
| "logits/chosen": 1.5889087915420532, | |
| "logits/rejected": 2.5967953205108643, | |
| "logps/chosen": -469.99884033203125, | |
| "logps/rejected": -584.1378784179688, | |
| "loss": 0.5084, | |
| "rewards/accuracies": 0.7437499761581421, | |
| "rewards/chosen": -1.6343872547149658, | |
| "rewards/margins": 1.305241346359253, | |
| "rewards/rejected": -2.9396286010742188, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8368200836820083, | |
| "grad_norm": 21.820601306046125, | |
| "learning_rate": 3.9507259776993954e-08, | |
| "logits/chosen": 1.4544366598129272, | |
| "logits/rejected": 2.2073497772216797, | |
| "logps/chosen": -470.62188720703125, | |
| "logps/rejected": -558.2474365234375, | |
| "loss": 0.5076, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -1.766871452331543, | |
| "rewards/margins": 1.029807209968567, | |
| "rewards/rejected": -2.796678304672241, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8368200836820083, | |
| "eval_logits/chosen": 1.3332302570343018, | |
| "eval_logits/rejected": 2.278512477874756, | |
| "eval_logps/chosen": -423.95953369140625, | |
| "eval_logps/rejected": -558.9130859375, | |
| "eval_loss": 0.5348207950592041, | |
| "eval_rewards/accuracies": 0.7890625, | |
| "eval_rewards/chosen": -1.6132957935333252, | |
| "eval_rewards/margins": 1.3492140769958496, | |
| "eval_rewards/rejected": -2.962510108947754, | |
| "eval_runtime": 105.1303, | |
| "eval_samples_per_second": 19.024, | |
| "eval_steps_per_second": 0.304, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8577405857740585, | |
| "grad_norm": 23.0331791932855, | |
| "learning_rate": 3.022313472693447e-08, | |
| "logits/chosen": 1.0064998865127563, | |
| "logits/rejected": 1.9991016387939453, | |
| "logps/chosen": -479.11846923828125, | |
| "logps/rejected": -583.687255859375, | |
| "loss": 0.4977, | |
| "rewards/accuracies": 0.793749988079071, | |
| "rewards/chosen": -1.6211599111557007, | |
| "rewards/margins": 1.2330563068389893, | |
| "rewards/rejected": -2.8542163372039795, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.8786610878661087, | |
| "grad_norm": 25.09829796115416, | |
| "learning_rate": 2.2111614344599684e-08, | |
| "logits/chosen": 1.3670974969863892, | |
| "logits/rejected": 2.6799685955047607, | |
| "logps/chosen": -465.97674560546875, | |
| "logps/rejected": -562.8436279296875, | |
| "loss": 0.4821, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.6978868246078491, | |
| "rewards/margins": 1.2206120491027832, | |
| "rewards/rejected": -2.9184985160827637, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.899581589958159, | |
| "grad_norm": 21.301017285621448, | |
| "learning_rate": 1.521597710086439e-08, | |
| "logits/chosen": 1.232881784439087, | |
| "logits/rejected": 2.3901188373565674, | |
| "logps/chosen": -466.0107421875, | |
| "logps/rejected": -540.5730590820312, | |
| "loss": 0.5114, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -1.70029616355896, | |
| "rewards/margins": 1.1120824813842773, | |
| "rewards/rejected": -2.8123791217803955, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9205020920502092, | |
| "grad_norm": 17.114528155184686, | |
| "learning_rate": 9.57301420397924e-09, | |
| "logits/chosen": 1.5404767990112305, | |
| "logits/rejected": 2.9349493980407715, | |
| "logps/chosen": -467.8946228027344, | |
| "logps/rejected": -551.1821899414062, | |
| "loss": 0.5065, | |
| "rewards/accuracies": 0.731249988079071, | |
| "rewards/chosen": -1.856208086013794, | |
| "rewards/margins": 1.1952415704727173, | |
| "rewards/rejected": -3.051449775695801, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.9414225941422594, | |
| "grad_norm": 19.50832169815014, | |
| "learning_rate": 5.212833302556258e-09, | |
| "logits/chosen": 1.3590748310089111, | |
| "logits/rejected": 2.3424363136291504, | |
| "logps/chosen": -451.4662170410156, | |
| "logps/rejected": -575.4518432617188, | |
| "loss": 0.5092, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": -1.667201042175293, | |
| "rewards/margins": 1.312021017074585, | |
| "rewards/rejected": -2.979222059249878, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9414225941422594, | |
| "eval_logits/chosen": 1.3857550621032715, | |
| "eval_logits/rejected": 2.3444478511810303, | |
| "eval_logps/chosen": -429.63800048828125, | |
| "eval_logps/rejected": -565.6296997070312, | |
| "eval_loss": 0.5340853333473206, | |
| "eval_rewards/accuracies": 0.78515625, | |
| "eval_rewards/chosen": -1.6700804233551025, | |
| "eval_rewards/margins": 1.3595958948135376, | |
| "eval_rewards/rejected": -3.0296761989593506, | |
| "eval_runtime": 102.8547, | |
| "eval_samples_per_second": 19.445, | |
| "eval_steps_per_second": 0.311, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.9623430962343096, | |
| "grad_norm": 18.50816041515539, | |
| "learning_rate": 2.158697848236607e-09, | |
| "logits/chosen": 1.729901909828186, | |
| "logits/rejected": 2.7589685916900635, | |
| "logps/chosen": -457.56207275390625, | |
| "logps/rejected": -565.1722412109375, | |
| "loss": 0.4978, | |
| "rewards/accuracies": 0.706250011920929, | |
| "rewards/chosen": -1.942845106124878, | |
| "rewards/margins": 1.1519941091537476, | |
| "rewards/rejected": -3.094839334487915, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.9832635983263598, | |
| "grad_norm": 18.70014449283038, | |
| "learning_rate": 4.269029751107489e-10, | |
| "logits/chosen": 1.6632955074310303, | |
| "logits/rejected": 2.187119960784912, | |
| "logps/chosen": -438.1412048339844, | |
| "logps/rejected": -556.7476806640625, | |
| "loss": 0.5241, | |
| "rewards/accuracies": 0.71875, | |
| "rewards/chosen": -1.7501564025878906, | |
| "rewards/margins": 1.1538572311401367, | |
| "rewards/rejected": -2.9040138721466064, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 478, | |
| "total_flos": 0.0, | |
| "train_loss": 0.5431942655451627, | |
| "train_runtime": 12655.0807, | |
| "train_samples_per_second": 4.831, | |
| "train_steps_per_second": 0.038 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 478, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |