|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.998691442030882, |
|
"eval_steps": 500, |
|
"global_step": 477, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.002093692750588851, |
|
"grad_norm": 4.60470009024502, |
|
"learning_rate": 1.0416666666666666e-08, |
|
"logits/chosen": 5002.53564453125, |
|
"logits/rejected": 4591.72021484375, |
|
"logps/chosen": -265.6396789550781, |
|
"logps/rejected": -206.22401428222656, |
|
"loss": 0.6931, |
|
"rewards/accuracies": 0.0, |
|
"rewards/chosen": 0.0, |
|
"rewards/margins": 0.0, |
|
"rewards/rejected": 0.0, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.02093692750588851, |
|
"grad_norm": 3.9004836147811037, |
|
"learning_rate": 1.0416666666666667e-07, |
|
"logits/chosen": 5977.43212890625, |
|
"logits/rejected": 5015.40087890625, |
|
"logps/chosen": -292.5469970703125, |
|
"logps/rejected": -259.3294677734375, |
|
"loss": 0.6931, |
|
"rewards/accuracies": 0.4618055522441864, |
|
"rewards/chosen": 0.00013807312643621117, |
|
"rewards/margins": 0.00019646011060103774, |
|
"rewards/rejected": -5.838695869897492e-05, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.04187385501177702, |
|
"grad_norm": 4.409000344379127, |
|
"learning_rate": 2.0833333333333333e-07, |
|
"logits/chosen": 5877.6181640625, |
|
"logits/rejected": 4855.9072265625, |
|
"logps/chosen": -290.42059326171875, |
|
"logps/rejected": -249.79830932617188, |
|
"loss": 0.6926, |
|
"rewards/accuracies": 0.6343749761581421, |
|
"rewards/chosen": 0.0017289689276367426, |
|
"rewards/margins": 0.002378009492531419, |
|
"rewards/rejected": -0.000649040739517659, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.06281078251766553, |
|
"grad_norm": 4.1115595198104895, |
|
"learning_rate": 3.1249999999999997e-07, |
|
"logits/chosen": 5644.3134765625, |
|
"logits/rejected": 4922.59326171875, |
|
"logps/chosen": -270.7469787597656, |
|
"logps/rejected": -238.15133666992188, |
|
"loss": 0.6907, |
|
"rewards/accuracies": 0.581250011920929, |
|
"rewards/chosen": 0.014904869720339775, |
|
"rewards/margins": 0.008396080695092678, |
|
"rewards/rejected": 0.00650878855958581, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.08374771002355404, |
|
"grad_norm": 3.9812692815534487, |
|
"learning_rate": 4.1666666666666667e-07, |
|
"logits/chosen": 5738.85693359375, |
|
"logits/rejected": 4991.9970703125, |
|
"logps/chosen": -269.1136169433594, |
|
"logps/rejected": -256.45672607421875, |
|
"loss": 0.6855, |
|
"rewards/accuracies": 0.6781250238418579, |
|
"rewards/chosen": 0.04843459650874138, |
|
"rewards/margins": 0.0284621212631464, |
|
"rewards/rejected": 0.01997247338294983, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.10468463752944256, |
|
"grad_norm": 4.46544140589733, |
|
"learning_rate": 4.999731868769026e-07, |
|
"logits/chosen": 6019.5712890625, |
|
"logits/rejected": 5225.3515625, |
|
"logps/chosen": -276.187744140625, |
|
"logps/rejected": -271.6847839355469, |
|
"loss": 0.6789, |
|
"rewards/accuracies": 0.6812499761581421, |
|
"rewards/chosen": 0.02429984137415886, |
|
"rewards/margins": 0.0659550279378891, |
|
"rewards/rejected": -0.04165518283843994, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.12562156503533106, |
|
"grad_norm": 4.993454825378816, |
|
"learning_rate": 4.990353313429303e-07, |
|
"logits/chosen": 5790.12646484375, |
|
"logits/rejected": 4928.04296875, |
|
"logps/chosen": -284.0537109375, |
|
"logps/rejected": -269.0484313964844, |
|
"loss": 0.6716, |
|
"rewards/accuracies": 0.6781250238418579, |
|
"rewards/chosen": -0.05631242319941521, |
|
"rewards/margins": 0.11938270181417465, |
|
"rewards/rejected": -0.17569510638713837, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.14655849254121958, |
|
"grad_norm": 6.738341247222451, |
|
"learning_rate": 4.967625656594781e-07, |
|
"logits/chosen": 5591.57421875, |
|
"logits/rejected": 5146.23779296875, |
|
"logps/chosen": -288.3265380859375, |
|
"logps/rejected": -293.3169860839844, |
|
"loss": 0.664, |
|
"rewards/accuracies": 0.659375011920929, |
|
"rewards/chosen": -0.1001863032579422, |
|
"rewards/margins": 0.16495101153850555, |
|
"rewards/rejected": -0.26513734459877014, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.16749542004710807, |
|
"grad_norm": 6.195725545958302, |
|
"learning_rate": 4.93167072587771e-07, |
|
"logits/chosen": 5745.04443359375, |
|
"logits/rejected": 4929.4033203125, |
|
"logps/chosen": -285.25628662109375, |
|
"logps/rejected": -284.0039367675781, |
|
"loss": 0.6578, |
|
"rewards/accuracies": 0.6781250238418579, |
|
"rewards/chosen": -0.09169374406337738, |
|
"rewards/margins": 0.2176208794116974, |
|
"rewards/rejected": -0.30931463837623596, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.1884323475529966, |
|
"grad_norm": 8.253469662871487, |
|
"learning_rate": 4.882681251368548e-07, |
|
"logits/chosen": 5881.85693359375, |
|
"logits/rejected": 4956.69482421875, |
|
"logps/chosen": -295.1579895019531, |
|
"logps/rejected": -297.10235595703125, |
|
"loss": 0.654, |
|
"rewards/accuracies": 0.684374988079071, |
|
"rewards/chosen": -0.1745041012763977, |
|
"rewards/margins": 0.24678310751914978, |
|
"rewards/rejected": -0.4212872087955475, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.2093692750588851, |
|
"grad_norm": 9.862843851151778, |
|
"learning_rate": 4.820919832540181e-07, |
|
"logits/chosen": 5971.64794921875, |
|
"logits/rejected": 5451.3154296875, |
|
"logps/chosen": -310.86712646484375, |
|
"logps/rejected": -344.1552734375, |
|
"loss": 0.6478, |
|
"rewards/accuracies": 0.6875, |
|
"rewards/chosen": -0.3099839687347412, |
|
"rewards/margins": 0.38641446828842163, |
|
"rewards/rejected": -0.6963984370231628, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.23030620256477363, |
|
"grad_norm": 7.664167361758326, |
|
"learning_rate": 4.7467175306295647e-07, |
|
"logits/chosen": 6156.85791015625, |
|
"logits/rejected": 4974.04296875, |
|
"logps/chosen": -305.1463317871094, |
|
"logps/rejected": -332.91363525390625, |
|
"loss": 0.6379, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": -0.3052375018596649, |
|
"rewards/margins": 0.4351824223995209, |
|
"rewards/rejected": -0.7404199838638306, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.2512431300706621, |
|
"grad_norm": 8.556770009207627, |
|
"learning_rate": 4.6604720940421207e-07, |
|
"logits/chosen": 5761.1591796875, |
|
"logits/rejected": 5308.05224609375, |
|
"logps/chosen": -308.85296630859375, |
|
"logps/rejected": -346.0533447265625, |
|
"loss": 0.6438, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.41673964262008667, |
|
"rewards/margins": 0.37503576278686523, |
|
"rewards/rejected": -0.7917754054069519, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.2721800575765506, |
|
"grad_norm": 11.15595897330077, |
|
"learning_rate": 4.5626458262912735e-07, |
|
"logits/chosen": 5819.66357421875, |
|
"logits/rejected": 5165.65771484375, |
|
"logps/chosen": -305.6595458984375, |
|
"logps/rejected": -344.2333679199219, |
|
"loss": 0.6412, |
|
"rewards/accuracies": 0.690625011920929, |
|
"rewards/chosen": -0.37142214179039, |
|
"rewards/margins": 0.38082534074783325, |
|
"rewards/rejected": -0.7522474527359009, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.29311698508243916, |
|
"grad_norm": 8.207443336589154, |
|
"learning_rate": 4.453763107901675e-07, |
|
"logits/chosen": 5615.36279296875, |
|
"logits/rejected": 4947.9091796875, |
|
"logps/chosen": -302.700927734375, |
|
"logps/rejected": -326.2046203613281, |
|
"loss": 0.6425, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": -0.30796128511428833, |
|
"rewards/margins": 0.4427562654018402, |
|
"rewards/rejected": -0.7507175207138062, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.31405391258832765, |
|
"grad_norm": 11.288023963328941, |
|
"learning_rate": 4.3344075855595097e-07, |
|
"logits/chosen": 6199.8349609375, |
|
"logits/rejected": 5057.89990234375, |
|
"logps/chosen": -339.68829345703125, |
|
"logps/rejected": -355.90643310546875, |
|
"loss": 0.636, |
|
"rewards/accuracies": 0.7437499761581421, |
|
"rewards/chosen": -0.5057331323623657, |
|
"rewards/margins": 0.5384237766265869, |
|
"rewards/rejected": -1.044156789779663, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.33499084009421615, |
|
"grad_norm": 10.16845225583156, |
|
"learning_rate": 4.2052190435769554e-07, |
|
"logits/chosen": 5847.75634765625, |
|
"logits/rejected": 4841.6728515625, |
|
"logps/chosen": -347.44244384765625, |
|
"logps/rejected": -371.6456604003906, |
|
"loss": 0.6265, |
|
"rewards/accuracies": 0.7406250238418579, |
|
"rewards/chosen": -0.6284831762313843, |
|
"rewards/margins": 0.5241909027099609, |
|
"rewards/rejected": -1.1526740789413452, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.3559277676001047, |
|
"grad_norm": 12.422675446136871, |
|
"learning_rate": 4.0668899744407567e-07, |
|
"logits/chosen": 6190.0078125, |
|
"logits/rejected": 5228.7177734375, |
|
"logps/chosen": -333.79241943359375, |
|
"logps/rejected": -341.86968994140625, |
|
"loss": 0.6269, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -0.5147669911384583, |
|
"rewards/margins": 0.4974791407585144, |
|
"rewards/rejected": -1.012246012687683, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.3768646951059932, |
|
"grad_norm": 8.193752614141884, |
|
"learning_rate": 3.920161866827889e-07, |
|
"logits/chosen": 5827.63427734375, |
|
"logits/rejected": 4925.3046875, |
|
"logps/chosen": -342.14068603515625, |
|
"logps/rejected": -365.3731689453125, |
|
"loss": 0.6314, |
|
"rewards/accuracies": 0.7437499761581421, |
|
"rewards/chosen": -0.548454225063324, |
|
"rewards/margins": 0.6054696440696716, |
|
"rewards/rejected": -1.1539238691329956, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.39780162261188173, |
|
"grad_norm": 11.037520790820944, |
|
"learning_rate": 3.765821230985757e-07, |
|
"logits/chosen": 5962.33056640625, |
|
"logits/rejected": 4930.44091796875, |
|
"logps/chosen": -310.32427978515625, |
|
"logps/rejected": -358.37774658203125, |
|
"loss": 0.6203, |
|
"rewards/accuracies": 0.7593749761581421, |
|
"rewards/chosen": -0.40572792291641235, |
|
"rewards/margins": 0.6992164850234985, |
|
"rewards/rejected": -1.1049444675445557, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.4187385501177702, |
|
"grad_norm": 9.527716723410062, |
|
"learning_rate": 3.604695382782159e-07, |
|
"logits/chosen": 5604.3310546875, |
|
"logits/rejected": 4814.98828125, |
|
"logps/chosen": -303.8848876953125, |
|
"logps/rejected": -348.39678955078125, |
|
"loss": 0.6349, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -0.495340496301651, |
|
"rewards/margins": 0.5600203275680542, |
|
"rewards/rejected": -1.0553607940673828, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.4396754776236587, |
|
"grad_norm": 12.016875634198213, |
|
"learning_rate": 3.4376480090239047e-07, |
|
"logits/chosen": 6233.3984375, |
|
"logits/rejected": 5250.013671875, |
|
"logps/chosen": -342.68585205078125, |
|
"logps/rejected": -369.70904541015625, |
|
"loss": 0.63, |
|
"rewards/accuracies": 0.7124999761581421, |
|
"rewards/chosen": -0.5591845512390137, |
|
"rewards/margins": 0.5208701491355896, |
|
"rewards/rejected": -1.0800548791885376, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.46061240512954726, |
|
"grad_norm": 12.10521639246573, |
|
"learning_rate": 3.265574537815398e-07, |
|
"logits/chosen": 5601.2353515625, |
|
"logits/rejected": 5141.3740234375, |
|
"logps/chosen": -329.7219543457031, |
|
"logps/rejected": -389.0228271484375, |
|
"loss": 0.6191, |
|
"rewards/accuracies": 0.778124988079071, |
|
"rewards/chosen": -0.5860620737075806, |
|
"rewards/margins": 0.7007312178611755, |
|
"rewards/rejected": -1.2867933511734009, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.48154933263543576, |
|
"grad_norm": 10.132071032109883, |
|
"learning_rate": 3.0893973387735683e-07, |
|
"logits/chosen": 5776.4091796875, |
|
"logits/rejected": 4840.046875, |
|
"logps/chosen": -335.4309997558594, |
|
"logps/rejected": -385.7275695800781, |
|
"loss": 0.619, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.6716222763061523, |
|
"rewards/margins": 0.7313363552093506, |
|
"rewards/rejected": -1.402958631515503, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.5024862601413242, |
|
"grad_norm": 11.646693213650721, |
|
"learning_rate": 2.910060778827554e-07, |
|
"logits/chosen": 5956.8095703125, |
|
"logits/rejected": 4968.13232421875, |
|
"logps/chosen": -334.1136779785156, |
|
"logps/rejected": -376.6734619140625, |
|
"loss": 0.6266, |
|
"rewards/accuracies": 0.7281249761581421, |
|
"rewards/chosen": -0.6853675246238708, |
|
"rewards/margins": 0.5835575461387634, |
|
"rewards/rejected": -1.2689251899719238, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.5234231876472127, |
|
"grad_norm": 9.693640002493433, |
|
"learning_rate": 2.7285261601056697e-07, |
|
"logits/chosen": 6025.3310546875, |
|
"logits/rejected": 5018.666015625, |
|
"logps/chosen": -362.75250244140625, |
|
"logps/rejected": -400.90155029296875, |
|
"loss": 0.6285, |
|
"rewards/accuracies": 0.7250000238418579, |
|
"rewards/chosen": -0.7680386304855347, |
|
"rewards/margins": 0.6202248334884644, |
|
"rewards/rejected": -1.3882635831832886, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.5443601151531012, |
|
"grad_norm": 12.110581184936667, |
|
"learning_rate": 2.5457665670441937e-07, |
|
"logits/chosen": 5841.02197265625, |
|
"logits/rejected": 5221.95703125, |
|
"logps/chosen": -350.9548034667969, |
|
"logps/rejected": -391.574951171875, |
|
"loss": 0.6227, |
|
"rewards/accuracies": 0.7875000238418579, |
|
"rewards/chosen": -0.6488569974899292, |
|
"rewards/margins": 0.6560698747634888, |
|
"rewards/rejected": -1.304926872253418, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.5652970426589898, |
|
"grad_norm": 11.095733736590391, |
|
"learning_rate": 2.3627616503391812e-07, |
|
"logits/chosen": 5896.1123046875, |
|
"logits/rejected": 5345.4794921875, |
|
"logps/chosen": -338.3013000488281, |
|
"logps/rejected": -395.18365478515625, |
|
"loss": 0.6222, |
|
"rewards/accuracies": 0.784375011920929, |
|
"rewards/chosen": -0.6368743777275085, |
|
"rewards/margins": 0.7109971046447754, |
|
"rewards/rejected": -1.3478715419769287, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.5862339701648783, |
|
"grad_norm": 9.899171422824232, |
|
"learning_rate": 2.1804923757009882e-07, |
|
"logits/chosen": 5956.20751953125, |
|
"logits/rejected": 4884.17724609375, |
|
"logps/chosen": -336.65655517578125, |
|
"logps/rejected": -378.20257568359375, |
|
"loss": 0.6209, |
|
"rewards/accuracies": 0.737500011920929, |
|
"rewards/chosen": -0.5593088269233704, |
|
"rewards/margins": 0.7548757791519165, |
|
"rewards/rejected": -1.314184546470642, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.6071708976707668, |
|
"grad_norm": 14.00097312046503, |
|
"learning_rate": 1.9999357655598891e-07, |
|
"logits/chosen": 6179.2138671875, |
|
"logits/rejected": 5698.1767578125, |
|
"logps/chosen": -375.6763000488281, |
|
"logps/rejected": -440.81805419921875, |
|
"loss": 0.6268, |
|
"rewards/accuracies": 0.715624988079071, |
|
"rewards/chosen": -0.8424223065376282, |
|
"rewards/margins": 0.5865424275398254, |
|
"rewards/rejected": -1.4289648532867432, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.6281078251766553, |
|
"grad_norm": 14.203557312456628, |
|
"learning_rate": 1.8220596619089573e-07, |
|
"logits/chosen": 5657.166015625, |
|
"logits/rejected": 5264.8466796875, |
|
"logps/chosen": -350.72784423828125, |
|
"logps/rejected": -411.1917419433594, |
|
"loss": 0.6231, |
|
"rewards/accuracies": 0.721875011920929, |
|
"rewards/chosen": -0.7937782406806946, |
|
"rewards/margins": 0.6089503765106201, |
|
"rewards/rejected": -1.40272855758667, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.6490447526825438, |
|
"grad_norm": 10.770674638912404, |
|
"learning_rate": 1.647817538357072e-07, |
|
"logits/chosen": 6337.3076171875, |
|
"logits/rejected": 5510.4892578125, |
|
"logps/chosen": -360.8718566894531, |
|
"logps/rejected": -397.1591796875, |
|
"loss": 0.6192, |
|
"rewards/accuracies": 0.765625, |
|
"rewards/chosen": -0.7004226446151733, |
|
"rewards/margins": 0.5756019353866577, |
|
"rewards/rejected": -1.2760246992111206, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.6699816801884323, |
|
"grad_norm": 12.586206352705796, |
|
"learning_rate": 1.478143389201113e-07, |
|
"logits/chosen": 5900.0009765625, |
|
"logits/rejected": 5053.62744140625, |
|
"logps/chosen": -338.50836181640625, |
|
"logps/rejected": -400.8262023925781, |
|
"loss": 0.6206, |
|
"rewards/accuracies": 0.7437499761581421, |
|
"rewards/chosen": -0.666448712348938, |
|
"rewards/margins": 0.7436891198158264, |
|
"rewards/rejected": -1.4101378917694092, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.6909186076943209, |
|
"grad_norm": 11.467937206667434, |
|
"learning_rate": 1.3139467229135998e-07, |
|
"logits/chosen": 5687.28564453125, |
|
"logits/rejected": 5192.236328125, |
|
"logps/chosen": -344.19000244140625, |
|
"logps/rejected": -401.15484619140625, |
|
"loss": 0.6266, |
|
"rewards/accuracies": 0.768750011920929, |
|
"rewards/chosen": -0.6740574240684509, |
|
"rewards/margins": 0.6941967010498047, |
|
"rewards/rejected": -1.3682540655136108, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.7118555352002094, |
|
"grad_norm": 12.06333657205332, |
|
"learning_rate": 1.1561076868822755e-07, |
|
"logits/chosen": 5517.1494140625, |
|
"logits/rejected": 4742.1181640625, |
|
"logps/chosen": -323.6866760253906, |
|
"logps/rejected": -374.42169189453125, |
|
"loss": 0.6239, |
|
"rewards/accuracies": 0.7281249761581421, |
|
"rewards/chosen": -0.6611974239349365, |
|
"rewards/margins": 0.6709928512573242, |
|
"rewards/rejected": -1.3321901559829712, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.7327924627060979, |
|
"grad_norm": 11.532722063890231, |
|
"learning_rate": 1.0054723495346482e-07, |
|
"logits/chosen": 6202.3583984375, |
|
"logits/rejected": 4977.18115234375, |
|
"logps/chosen": -362.2864990234375, |
|
"logps/rejected": -393.2933349609375, |
|
"loss": 0.6251, |
|
"rewards/accuracies": 0.768750011920929, |
|
"rewards/chosen": -0.6137198209762573, |
|
"rewards/margins": 0.635828971862793, |
|
"rewards/rejected": -1.2495489120483398, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.7537293902119864, |
|
"grad_norm": 12.827074983628805, |
|
"learning_rate": 8.628481651367875e-08, |
|
"logits/chosen": 5883.3125, |
|
"logits/rejected": 5070.5986328125, |
|
"logps/chosen": -341.68011474609375, |
|
"logps/rejected": -390.2152099609375, |
|
"loss": 0.6263, |
|
"rewards/accuracies": 0.737500011920929, |
|
"rewards/chosen": -0.6395655870437622, |
|
"rewards/margins": 0.6800642013549805, |
|
"rewards/rejected": -1.3196299076080322, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.7746663177178749, |
|
"grad_norm": 9.251136847302337, |
|
"learning_rate": 7.289996455765748e-08, |
|
"logits/chosen": 5781.0732421875, |
|
"logits/rejected": 4854.42333984375, |
|
"logps/chosen": -352.7303161621094, |
|
"logps/rejected": -394.6329345703125, |
|
"loss": 0.6134, |
|
"rewards/accuracies": 0.778124988079071, |
|
"rewards/chosen": -0.639372706413269, |
|
"rewards/margins": 0.7373119592666626, |
|
"rewards/rejected": -1.376684546470642, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.7956032452237635, |
|
"grad_norm": 12.098049848889806, |
|
"learning_rate": 6.046442623320145e-08, |
|
"logits/chosen": 6060.8408203125, |
|
"logits/rejected": 5173.55126953125, |
|
"logps/chosen": -345.76519775390625, |
|
"logps/rejected": -371.1272888183594, |
|
"loss": 0.6248, |
|
"rewards/accuracies": 0.753125011920929, |
|
"rewards/chosen": -0.6106373071670532, |
|
"rewards/margins": 0.5920594334602356, |
|
"rewards/rejected": -1.202696681022644, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.816540172729652, |
|
"grad_norm": 9.994672177952491, |
|
"learning_rate": 4.904486005914027e-08, |
|
"logits/chosen": 6291.39501953125, |
|
"logits/rejected": 5494.92431640625, |
|
"logps/chosen": -351.2883605957031, |
|
"logps/rejected": -405.6934814453125, |
|
"loss": 0.6161, |
|
"rewards/accuracies": 0.7562500238418579, |
|
"rewards/chosen": -0.5953892469406128, |
|
"rewards/margins": 0.7137786746025085, |
|
"rewards/rejected": -1.3091679811477661, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.8374771002355405, |
|
"grad_norm": 10.018943299981588, |
|
"learning_rate": 3.8702478614051345e-08, |
|
"logits/chosen": 5870.2626953125, |
|
"logits/rejected": 5131.064453125, |
|
"logps/chosen": -352.17132568359375, |
|
"logps/rejected": -379.8227233886719, |
|
"loss": 0.6228, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -0.6126725673675537, |
|
"rewards/margins": 0.626976490020752, |
|
"rewards/rejected": -1.2396490573883057, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.8584140277414289, |
|
"grad_norm": 12.50787678673054, |
|
"learning_rate": 2.9492720416985e-08, |
|
"logits/chosen": 5846.2509765625, |
|
"logits/rejected": 5063.88818359375, |
|
"logps/chosen": -348.4322814941406, |
|
"logps/rejected": -408.62237548828125, |
|
"loss": 0.6161, |
|
"rewards/accuracies": 0.71875, |
|
"rewards/chosen": -0.6656492948532104, |
|
"rewards/margins": 0.6094722151756287, |
|
"rewards/rejected": -1.2751215696334839, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.8793509552473174, |
|
"grad_norm": 13.503737918548683, |
|
"learning_rate": 2.1464952759020856e-08, |
|
"logits/chosen": 6236.7607421875, |
|
"logits/rejected": 5383.3154296875, |
|
"logps/chosen": -355.6968994140625, |
|
"logps/rejected": -402.67681884765625, |
|
"loss": 0.6183, |
|
"rewards/accuracies": 0.746874988079071, |
|
"rewards/chosen": -0.6371773481369019, |
|
"rewards/margins": 0.6647071242332458, |
|
"rewards/rejected": -1.3018845319747925, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.9002878827532059, |
|
"grad_norm": 12.19770327814421, |
|
"learning_rate": 1.4662207078575684e-08, |
|
"logits/chosen": 6165.9990234375, |
|
"logits/rejected": 4889.5029296875, |
|
"logps/chosen": -349.86297607421875, |
|
"logps/rejected": -390.2151184082031, |
|
"loss": 0.6092, |
|
"rewards/accuracies": 0.768750011920929, |
|
"rewards/chosen": -0.6226025819778442, |
|
"rewards/margins": 0.7049604654312134, |
|
"rewards/rejected": -1.3275630474090576, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.9212248102590945, |
|
"grad_norm": 11.802179491960299, |
|
"learning_rate": 9.12094829893642e-09, |
|
"logits/chosen": 5992.05029296875, |
|
"logits/rejected": 5618.6328125, |
|
"logps/chosen": -338.279541015625, |
|
"logps/rejected": -399.9471130371094, |
|
"loss": 0.6193, |
|
"rewards/accuracies": 0.7437499761581421, |
|
"rewards/chosen": -0.5973601937294006, |
|
"rewards/margins": 0.6416517496109009, |
|
"rewards/rejected": -1.2390120029449463, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.942161737764983, |
|
"grad_norm": 15.288231020941236, |
|
"learning_rate": 4.8708793644441086e-09, |
|
"logits/chosen": 6121.3330078125, |
|
"logits/rejected": 5176.08349609375, |
|
"logps/chosen": -350.21331787109375, |
|
"logps/rejected": -387.7196044921875, |
|
"loss": 0.6177, |
|
"rewards/accuracies": 0.7906249761581421, |
|
"rewards/chosen": -0.598780632019043, |
|
"rewards/margins": 0.6639107465744019, |
|
"rewards/rejected": -1.2626912593841553, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.9630986652708715, |
|
"grad_norm": 14.093283682723904, |
|
"learning_rate": 1.9347820230782295e-09, |
|
"logits/chosen": 6296.310546875, |
|
"logits/rejected": 5237.2060546875, |
|
"logps/chosen": -378.281982421875, |
|
"logps/rejected": -422.67242431640625, |
|
"loss": 0.6236, |
|
"rewards/accuracies": 0.753125011920929, |
|
"rewards/chosen": -0.652829647064209, |
|
"rewards/margins": 0.7167678475379944, |
|
"rewards/rejected": -1.3695974349975586, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.98403559277676, |
|
"grad_norm": 9.706197753513703, |
|
"learning_rate": 3.2839470889836627e-10, |
|
"logits/chosen": 5874.50830078125, |
|
"logits/rejected": 5139.2998046875, |
|
"logps/chosen": -342.42303466796875, |
|
"logps/rejected": -391.7164001464844, |
|
"loss": 0.6113, |
|
"rewards/accuracies": 0.7749999761581421, |
|
"rewards/chosen": -0.6158385872840881, |
|
"rewards/margins": 0.7232969999313354, |
|
"rewards/rejected": -1.3391354084014893, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.998691442030882, |
|
"step": 477, |
|
"total_flos": 0.0, |
|
"train_loss": 0.6349887008187156, |
|
"train_runtime": 13807.6715, |
|
"train_samples_per_second": 4.428, |
|
"train_steps_per_second": 0.035 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 477, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 1000, |
|
"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": 4, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|