| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 385, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 7.90625, | |
| "learning_rate": 1.282051282051282e-07, | |
| "logits/chosen": -2.7358343601226807, | |
| "logits/rejected": -2.7480404376983643, | |
| "logps/chosen": -27.35565757751465, | |
| "logps/rejected": -21.06114387512207, | |
| "loss": 1.0, | |
| "rewards/accuracies": 0.0, | |
| "rewards/chosen": 0.0, | |
| "rewards/margins": 0.0, | |
| "rewards/rejected": 0.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 13.4375, | |
| "learning_rate": 1.282051282051282e-06, | |
| "logits/chosen": -3.009408473968506, | |
| "logits/rejected": -2.9977328777313232, | |
| "logps/chosen": -33.20761489868164, | |
| "logps/rejected": -31.960758209228516, | |
| "loss": 1.003, | |
| "rewards/accuracies": 0.375, | |
| "rewards/chosen": -0.0026669783983379602, | |
| "rewards/margins": -0.002981698838993907, | |
| "rewards/rejected": 0.00031472102273255587, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 9.1875, | |
| "learning_rate": 2.564102564102564e-06, | |
| "logits/chosen": -2.8990566730499268, | |
| "logits/rejected": -2.894094944000244, | |
| "logps/chosen": -32.50642013549805, | |
| "logps/rejected": -28.963293075561523, | |
| "loss": 1.005, | |
| "rewards/accuracies": 0.4000000059604645, | |
| "rewards/chosen": -0.0031326524913311005, | |
| "rewards/margins": -0.00499005988240242, | |
| "rewards/rejected": 0.0018574076239019632, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 8.8125, | |
| "learning_rate": 3.846153846153847e-06, | |
| "logits/chosen": -3.0970427989959717, | |
| "logits/rejected": -3.108668804168701, | |
| "logps/chosen": -32.80826187133789, | |
| "logps/rejected": -30.164159774780273, | |
| "loss": 0.9931, | |
| "rewards/accuracies": 0.550000011920929, | |
| "rewards/chosen": 0.010042307898402214, | |
| "rewards/margins": 0.006853123661130667, | |
| "rewards/rejected": 0.0031891819089651108, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 9.0625, | |
| "learning_rate": 4.999896948438434e-06, | |
| "logits/chosen": -2.8629202842712402, | |
| "logits/rejected": -2.8535947799682617, | |
| "logps/chosen": -31.525354385375977, | |
| "logps/rejected": -32.399391174316406, | |
| "loss": 0.9653, | |
| "rewards/accuracies": 0.612500011920929, | |
| "rewards/chosen": 0.030497059226036072, | |
| "rewards/margins": 0.034690458327531815, | |
| "rewards/rejected": -0.004193395376205444, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 8.1875, | |
| "learning_rate": 4.987541037542187e-06, | |
| "logits/chosen": -2.87955904006958, | |
| "logits/rejected": -2.877352714538574, | |
| "logps/chosen": -29.324417114257812, | |
| "logps/rejected": -30.05527114868164, | |
| "loss": 0.9618, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.04157434031367302, | |
| "rewards/margins": 0.03820687159895897, | |
| "rewards/rejected": 0.0033674687147140503, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 8.0625, | |
| "learning_rate": 4.954691471941119e-06, | |
| "logits/chosen": -2.9058096408843994, | |
| "logits/rejected": -2.90769624710083, | |
| "logps/chosen": -29.792831420898438, | |
| "logps/rejected": -27.967548370361328, | |
| "loss": 0.9651, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.03369411081075668, | |
| "rewards/margins": 0.03488066419959068, | |
| "rewards/rejected": -0.0011865533888339996, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 10.75, | |
| "learning_rate": 4.901618883413549e-06, | |
| "logits/chosen": -2.98918080329895, | |
| "logits/rejected": -2.9952526092529297, | |
| "logps/chosen": -29.09405517578125, | |
| "logps/rejected": -30.891315460205078, | |
| "loss": 0.9814, | |
| "rewards/accuracies": 0.6000000238418579, | |
| "rewards/chosen": 0.018691470846533775, | |
| "rewards/margins": 0.018642673268914223, | |
| "rewards/rejected": 4.879874177277088e-05, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 9.375, | |
| "learning_rate": 4.828760511501322e-06, | |
| "logits/chosen": -2.804455518722534, | |
| "logits/rejected": -2.8193769454956055, | |
| "logps/chosen": -29.27066993713379, | |
| "logps/rejected": -29.863460540771484, | |
| "loss": 0.9541, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.027005871757864952, | |
| "rewards/margins": 0.04592365399003029, | |
| "rewards/rejected": -0.018917780369520187, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 9.5, | |
| "learning_rate": 4.7367166013034295e-06, | |
| "logits/chosen": -2.8940837383270264, | |
| "logits/rejected": -2.8764431476593018, | |
| "logps/chosen": -32.690391540527344, | |
| "logps/rejected": -30.344482421875, | |
| "loss": 0.9531, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": 0.014603940770030022, | |
| "rewards/margins": 0.04688751697540283, | |
| "rewards/rejected": -0.03228358179330826, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 8.4375, | |
| "learning_rate": 4.626245458345211e-06, | |
| "logits/chosen": -2.999018669128418, | |
| "logits/rejected": -2.9999501705169678, | |
| "logps/chosen": -31.824331283569336, | |
| "logps/rejected": -31.003093719482422, | |
| "loss": 0.9581, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.023254599422216415, | |
| "rewards/margins": 0.041946638375520706, | |
| "rewards/rejected": -0.01869204081594944, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "eval_logits/chosen": -2.805298089981079, | |
| "eval_logits/rejected": -2.802727460861206, | |
| "eval_logps/chosen": -31.269874572753906, | |
| "eval_logps/rejected": -34.867759704589844, | |
| "eval_loss": 0.9819382429122925, | |
| "eval_rewards/accuracies": 0.5955149531364441, | |
| "eval_rewards/chosen": 0.0012576259905472398, | |
| "eval_rewards/margins": 0.018218038603663445, | |
| "eval_rewards/rejected": -0.01696041226387024, | |
| "eval_runtime": 113.3967, | |
| "eval_samples_per_second": 3.025, | |
| "eval_steps_per_second": 0.379, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 10.4375, | |
| "learning_rate": 4.498257201263691e-06, | |
| "logits/chosen": -2.9547231197357178, | |
| "logits/rejected": -2.9307026863098145, | |
| "logps/chosen": -31.882299423217773, | |
| "logps/rejected": -31.55330467224121, | |
| "loss": 0.9271, | |
| "rewards/accuracies": 0.675000011920929, | |
| "rewards/chosen": 0.03344918042421341, | |
| "rewards/margins": 0.07289471477270126, | |
| "rewards/rejected": -0.039445530623197556, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 8.9375, | |
| "learning_rate": 4.353806263777678e-06, | |
| "logits/chosen": -3.0366623401641846, | |
| "logits/rejected": -3.0655264854431152, | |
| "logps/chosen": -28.671463012695312, | |
| "logps/rejected": -34.533443450927734, | |
| "loss": 0.9122, | |
| "rewards/accuracies": 0.699999988079071, | |
| "rewards/chosen": 0.049243707209825516, | |
| "rewards/margins": 0.0878329798579216, | |
| "rewards/rejected": -0.03858928382396698, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 8.5625, | |
| "learning_rate": 4.1940827077152755e-06, | |
| "logits/chosen": -2.741682529449463, | |
| "logits/rejected": -2.7387285232543945, | |
| "logps/chosen": -28.681751251220703, | |
| "logps/rejected": -30.781078338623047, | |
| "loss": 0.9069, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.029292941093444824, | |
| "rewards/margins": 0.09402014315128326, | |
| "rewards/rejected": -0.06472718715667725, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 8.6875, | |
| "learning_rate": 4.0204024186666215e-06, | |
| "logits/chosen": -3.0192549228668213, | |
| "logits/rejected": -3.0171239376068115, | |
| "logps/chosen": -27.48164939880371, | |
| "logps/rejected": -32.18421936035156, | |
| "loss": 0.9468, | |
| "rewards/accuracies": 0.5625, | |
| "rewards/chosen": 0.00045731887803412974, | |
| "rewards/margins": 0.0532023087143898, | |
| "rewards/rejected": -0.05274498462677002, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 8.8125, | |
| "learning_rate": 3.834196265035119e-06, | |
| "logits/chosen": -2.8165481090545654, | |
| "logits/rejected": -2.8123271465301514, | |
| "logps/chosen": -27.479320526123047, | |
| "logps/rejected": -32.02801513671875, | |
| "loss": 0.8849, | |
| "rewards/accuracies": 0.625, | |
| "rewards/chosen": 0.034618668258190155, | |
| "rewards/margins": 0.11505589634180069, | |
| "rewards/rejected": -0.08043723553419113, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 10.25, | |
| "learning_rate": 3.636998309800573e-06, | |
| "logits/chosen": -3.137382745742798, | |
| "logits/rejected": -3.1210262775421143, | |
| "logps/chosen": -31.968801498413086, | |
| "logps/rejected": -30.1705265045166, | |
| "loss": 0.8343, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.05229174345731735, | |
| "rewards/margins": 0.17136676609516144, | |
| "rewards/rejected": -0.11907501518726349, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 9.5625, | |
| "learning_rate": 3.4304331721118078e-06, | |
| "logits/chosen": -2.9499218463897705, | |
| "logits/rejected": -2.958042621612549, | |
| "logps/chosen": -29.790802001953125, | |
| "logps/rejected": -32.55553436279297, | |
| "loss": 0.8471, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": 0.010381256230175495, | |
| "rewards/margins": 0.15598610043525696, | |
| "rewards/rejected": -0.14560487866401672, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 10.0625, | |
| "learning_rate": 3.2162026428305436e-06, | |
| "logits/chosen": -2.810059070587158, | |
| "logits/rejected": -2.8075289726257324, | |
| "logps/chosen": -29.805904388427734, | |
| "logps/rejected": -30.83272361755371, | |
| "loss": 0.9106, | |
| "rewards/accuracies": 0.637499988079071, | |
| "rewards/chosen": -0.013605661690235138, | |
| "rewards/margins": 0.09284953773021698, | |
| "rewards/rejected": -0.10645520687103271, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 6.375, | |
| "learning_rate": 2.996071664294641e-06, | |
| "logits/chosen": -2.920551300048828, | |
| "logits/rejected": -2.917001247406006, | |
| "logps/chosen": -30.380672454833984, | |
| "logps/rejected": -29.473743438720703, | |
| "loss": 0.9132, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.019395861774683, | |
| "rewards/margins": 0.09222523123025894, | |
| "rewards/rejected": -0.11162110418081284, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 9.75, | |
| "learning_rate": 2.7718537898066833e-06, | |
| "logits/chosen": -2.9904088973999023, | |
| "logits/rejected": -2.9779367446899414, | |
| "logps/chosen": -33.54608917236328, | |
| "logps/rejected": -31.4923095703125, | |
| "loss": 0.8824, | |
| "rewards/accuracies": 0.574999988079071, | |
| "rewards/chosen": 0.015293078497052193, | |
| "rewards/margins": 0.12111523002386093, | |
| "rewards/rejected": -0.10582216084003448, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "eval_logits/chosen": -2.8315017223358154, | |
| "eval_logits/rejected": -2.8287980556488037, | |
| "eval_logps/chosen": -31.827983856201172, | |
| "eval_logps/rejected": -35.64292526245117, | |
| "eval_loss": 0.9602205753326416, | |
| "eval_rewards/accuracies": 0.6092192530632019, | |
| "eval_rewards/chosen": -0.05455298721790314, | |
| "eval_rewards/margins": 0.039924219250679016, | |
| "eval_rewards/rejected": -0.09447719901800156, | |
| "eval_runtime": 113.1607, | |
| "eval_samples_per_second": 3.031, | |
| "eval_steps_per_second": 0.38, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 12.125, | |
| "learning_rate": 2.5453962426402006e-06, | |
| "logits/chosen": -2.9306328296661377, | |
| "logits/rejected": -2.9321701526641846, | |
| "logps/chosen": -33.18594741821289, | |
| "logps/rejected": -35.648868560791016, | |
| "loss": 0.8442, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.015253487043082714, | |
| "rewards/margins": 0.16348664462566376, | |
| "rewards/rejected": -0.17874012887477875, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 10.375, | |
| "learning_rate": 2.3185646976551794e-06, | |
| "logits/chosen": -2.9161877632141113, | |
| "logits/rejected": -2.9325830936431885, | |
| "logps/chosen": -29.98288345336914, | |
| "logps/rejected": -30.195465087890625, | |
| "loss": 0.818, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.019515613093972206, | |
| "rewards/margins": 0.18623510003089905, | |
| "rewards/rejected": -0.1667194664478302, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 8.9375, | |
| "learning_rate": 2.0932279108998323e-06, | |
| "logits/chosen": -2.9633114337921143, | |
| "logits/rejected": -2.9677231311798096, | |
| "logps/chosen": -31.465539932250977, | |
| "logps/rejected": -33.331993103027344, | |
| "loss": 0.8836, | |
| "rewards/accuracies": 0.6499999761581421, | |
| "rewards/chosen": -0.03136688470840454, | |
| "rewards/margins": 0.1260170042514801, | |
| "rewards/rejected": -0.15738390386104584, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 10.5625, | |
| "learning_rate": 1.8712423238279358e-06, | |
| "logits/chosen": -3.0217947959899902, | |
| "logits/rejected": -3.029099702835083, | |
| "logps/chosen": -31.11104965209961, | |
| "logps/rejected": -31.991535186767578, | |
| "loss": 0.845, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": -0.020612526684999466, | |
| "rewards/margins": 0.15501661598682404, | |
| "rewards/rejected": -0.1756291538476944, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 9.25, | |
| "learning_rate": 1.6544367689701824e-06, | |
| "logits/chosen": -2.8557331562042236, | |
| "logits/rejected": -2.844738721847534, | |
| "logps/chosen": -27.252483367919922, | |
| "logps/rejected": -30.79751968383789, | |
| "loss": 0.9072, | |
| "rewards/accuracies": 0.5874999761581421, | |
| "rewards/chosen": -0.029826903715729713, | |
| "rewards/margins": 0.0967610701918602, | |
| "rewards/rejected": -0.12658795714378357, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 10.875, | |
| "learning_rate": 1.4445974030621963e-06, | |
| "logits/chosen": -2.8332571983337402, | |
| "logits/rejected": -2.854872226715088, | |
| "logps/chosen": -29.75779151916504, | |
| "logps/rejected": -36.265663146972656, | |
| "loss": 0.7843, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.007217098027467728, | |
| "rewards/margins": 0.21637487411499023, | |
| "rewards/rejected": -0.209157794713974, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 11.6875, | |
| "learning_rate": 1.243452991757889e-06, | |
| "logits/chosen": -2.975635051727295, | |
| "logits/rejected": -2.9822394847869873, | |
| "logps/chosen": -31.34609031677246, | |
| "logps/rejected": -32.07633972167969, | |
| "loss": 0.8354, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": -0.05647352337837219, | |
| "rewards/margins": 0.16463439166545868, | |
| "rewards/rejected": -0.22110792994499207, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 10.9375, | |
| "learning_rate": 1.0526606671603523e-06, | |
| "logits/chosen": -2.9911038875579834, | |
| "logits/rejected": -2.9766478538513184, | |
| "logps/chosen": -31.343215942382812, | |
| "logps/rejected": -30.425678253173828, | |
| "loss": 0.8864, | |
| "rewards/accuracies": 0.6625000238418579, | |
| "rewards/chosen": -0.06215282529592514, | |
| "rewards/margins": 0.1172800287604332, | |
| "rewards/rejected": -0.17943285405635834, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 13.5, | |
| "learning_rate": 8.737922755071455e-07, | |
| "logits/chosen": -2.9202213287353516, | |
| "logits/rejected": -2.9021944999694824, | |
| "logps/chosen": -32.332275390625, | |
| "logps/rejected": -32.9135627746582, | |
| "loss": 0.7348, | |
| "rewards/accuracies": 0.7749999761581421, | |
| "rewards/chosen": 0.030402958393096924, | |
| "rewards/margins": 0.26942819356918335, | |
| "rewards/rejected": -0.23902523517608643, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 9.0625, | |
| "learning_rate": 7.08321427484816e-07, | |
| "logits/chosen": -2.9209041595458984, | |
| "logits/rejected": -2.9169058799743652, | |
| "logps/chosen": -32.4577522277832, | |
| "logps/rejected": -29.638641357421875, | |
| "loss": 0.8245, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.02350962720811367, | |
| "rewards/margins": 0.17999625205993652, | |
| "rewards/rejected": -0.20350590348243713, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "eval_logits/chosen": -2.8392350673675537, | |
| "eval_logits/rejected": -2.836524486541748, | |
| "eval_logps/chosen": -32.03562545776367, | |
| "eval_logps/rejected": -35.89268112182617, | |
| "eval_loss": 0.9559255838394165, | |
| "eval_rewards/accuracies": 0.6004983186721802, | |
| "eval_rewards/chosen": -0.07531756907701492, | |
| "eval_rewards/margins": 0.044134579598903656, | |
| "eval_rewards/rejected": -0.11945215612649918, | |
| "eval_runtime": 112.8661, | |
| "eval_samples_per_second": 3.039, | |
| "eval_steps_per_second": 0.381, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 10.1875, | |
| "learning_rate": 5.576113578589035e-07, | |
| "logits/chosen": -2.80441951751709, | |
| "logits/rejected": -2.8214964866638184, | |
| "logps/chosen": -29.70196533203125, | |
| "logps/rejected": -33.18954849243164, | |
| "loss": 0.8066, | |
| "rewards/accuracies": 0.824999988079071, | |
| "rewards/chosen": -0.047254521399736404, | |
| "rewards/margins": 0.19849951565265656, | |
| "rewards/rejected": -0.24575403332710266, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 9.5, | |
| "learning_rate": 4.229036944380913e-07, | |
| "logits/chosen": -3.0470097064971924, | |
| "logits/rejected": -3.033452033996582, | |
| "logps/chosen": -29.982385635375977, | |
| "logps/rejected": -30.24712562561035, | |
| "loss": 0.8017, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": 0.022920172661542892, | |
| "rewards/margins": 0.21575525403022766, | |
| "rewards/rejected": -0.19283509254455566, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 11.875, | |
| "learning_rate": 3.053082288996112e-07, | |
| "logits/chosen": -2.96113657951355, | |
| "logits/rejected": -2.9435229301452637, | |
| "logps/chosen": -28.380762100219727, | |
| "logps/rejected": -32.93931579589844, | |
| "loss": 0.7579, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.0018691420555114746, | |
| "rewards/margins": 0.25289779901504517, | |
| "rewards/rejected": -0.25476694107055664, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 10.5625, | |
| "learning_rate": 2.0579377374915805e-07, | |
| "logits/chosen": -3.182105541229248, | |
| "logits/rejected": -3.1877188682556152, | |
| "logps/chosen": -31.98989486694336, | |
| "logps/rejected": -35.27183532714844, | |
| "loss": 0.7995, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.035068999975919724, | |
| "rewards/margins": 0.22251394391059875, | |
| "rewards/rejected": -0.2575829327106476, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 11.75, | |
| "learning_rate": 1.2518018074041684e-07, | |
| "logits/chosen": -3.057163953781128, | |
| "logits/rejected": -3.0605666637420654, | |
| "logps/chosen": -30.571701049804688, | |
| "logps/rejected": -33.298683166503906, | |
| "loss": 0.8066, | |
| "rewards/accuracies": 0.762499988079071, | |
| "rewards/chosen": 0.023518025875091553, | |
| "rewards/margins": 0.20205526053905487, | |
| "rewards/rejected": -0.1785372495651245, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 11.0, | |
| "learning_rate": 6.41315865106129e-08, | |
| "logits/chosen": -2.8954837322235107, | |
| "logits/rejected": -2.8972554206848145, | |
| "logps/chosen": -28.211322784423828, | |
| "logps/rejected": -31.518001556396484, | |
| "loss": 0.7939, | |
| "rewards/accuracies": 0.7124999761581421, | |
| "rewards/chosen": 0.04384060949087143, | |
| "rewards/margins": 0.21254661679267883, | |
| "rewards/rejected": -0.1687059849500656, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 10.3125, | |
| "learning_rate": 2.3150941078050325e-08, | |
| "logits/chosen": -2.973878860473633, | |
| "logits/rejected": -2.9718644618988037, | |
| "logps/chosen": -30.932153701782227, | |
| "logps/rejected": -33.92184829711914, | |
| "loss": 0.8202, | |
| "rewards/accuracies": 0.6875, | |
| "rewards/chosen": -0.014550352469086647, | |
| "rewards/margins": 0.1848299205303192, | |
| "rewards/rejected": -0.1993802934885025, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 9.8125, | |
| "learning_rate": 2.575864278703266e-09, | |
| "logits/chosen": -2.927494764328003, | |
| "logits/rejected": -2.9113755226135254, | |
| "logps/chosen": -29.455524444580078, | |
| "logps/rejected": -30.044204711914062, | |
| "loss": 0.8384, | |
| "rewards/accuracies": 0.7250000238418579, | |
| "rewards/chosen": -0.05104842036962509, | |
| "rewards/margins": 0.16441020369529724, | |
| "rewards/rejected": -0.21545863151550293, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 385, | |
| "total_flos": 0.0, | |
| "train_loss": 0.8805358874333369, | |
| "train_runtime": 2723.8224, | |
| "train_samples_per_second": 1.13, | |
| "train_steps_per_second": 0.141 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 385, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |