|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.0, |
|
"eval_steps": 100, |
|
"global_step": 1000, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 5.0000000000000004e-08, |
|
"logits/chosen": -3.1678528785705566, |
|
"logits/rejected": -3.1461400985717773, |
|
"logps/chosen": -344.33587646484375, |
|
"logps/rejected": -334.830810546875, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.0, |
|
"rewards/chosen": 0.0, |
|
"rewards/margins": 0.0, |
|
"rewards/rejected": 0.0, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 5.000000000000001e-07, |
|
"logits/chosen": -2.8056228160858154, |
|
"logits/rejected": -2.711918830871582, |
|
"logps/chosen": -137.3001708984375, |
|
"logps/rejected": -114.56926727294922, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.6111111044883728, |
|
"rewards/chosen": 0.0003753364726435393, |
|
"rewards/margins": 0.001190218492411077, |
|
"rewards/rejected": -0.0008148818160407245, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.0000000000000002e-06, |
|
"logits/chosen": -2.664208173751831, |
|
"logits/rejected": -2.3068690299987793, |
|
"logps/chosen": -95.76632690429688, |
|
"logps/rejected": -91.30467224121094, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.4000000059604645, |
|
"rewards/chosen": -7.150838064262643e-05, |
|
"rewards/margins": 0.00038482382660731673, |
|
"rewards/rejected": -0.0004563321708701551, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.5e-06, |
|
"logits/chosen": -2.8235249519348145, |
|
"logits/rejected": -2.742885112762451, |
|
"logps/chosen": -103.0609359741211, |
|
"logps/rejected": -119.29620361328125, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": -0.00019755656830966473, |
|
"rewards/margins": 0.00018337149231228977, |
|
"rewards/rejected": -0.00038092798786237836, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 2.0000000000000003e-06, |
|
"logits/chosen": -2.812087059020996, |
|
"logits/rejected": -2.3899402618408203, |
|
"logps/chosen": -56.24201583862305, |
|
"logps/rejected": -52.55266189575195, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.4000000059604645, |
|
"rewards/chosen": -0.002384146675467491, |
|
"rewards/margins": 0.0011411935556679964, |
|
"rewards/rejected": -0.0035253397654742002, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 2.5e-06, |
|
"logits/chosen": -2.919544219970703, |
|
"logits/rejected": -2.68827486038208, |
|
"logps/chosen": -85.77859497070312, |
|
"logps/rejected": -82.93376159667969, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": -0.002311713993549347, |
|
"rewards/margins": 0.0031323134899139404, |
|
"rewards/rejected": -0.005444027483463287, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 3e-06, |
|
"logits/chosen": -2.7442586421966553, |
|
"logits/rejected": -2.5409207344055176, |
|
"logps/chosen": -135.08984375, |
|
"logps/rejected": -137.5052032470703, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": -0.0012713891919702291, |
|
"rewards/margins": 0.005557497497648001, |
|
"rewards/rejected": -0.006828887853771448, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 3.5e-06, |
|
"logits/chosen": -2.680795669555664, |
|
"logits/rejected": -2.3159432411193848, |
|
"logps/chosen": -86.95841979980469, |
|
"logps/rejected": -71.47272491455078, |
|
"loss": 0.0012, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": 0.0019758909475058317, |
|
"rewards/margins": 0.005556103773415089, |
|
"rewards/rejected": -0.0035802137572318316, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 4.000000000000001e-06, |
|
"logits/chosen": -2.789736270904541, |
|
"logits/rejected": -2.5091891288757324, |
|
"logps/chosen": -54.42887496948242, |
|
"logps/rejected": -45.48096466064453, |
|
"loss": 0.0009, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.011168469674885273, |
|
"rewards/margins": 0.00967374723404646, |
|
"rewards/rejected": 0.0014947212766855955, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 4.5e-06, |
|
"logits/chosen": -2.686554431915283, |
|
"logits/rejected": -2.3966121673583984, |
|
"logps/chosen": -88.09907531738281, |
|
"logps/rejected": -87.56222534179688, |
|
"loss": 0.0008, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.016689006239175797, |
|
"rewards/margins": 0.0082827378064394, |
|
"rewards/rejected": 0.008406268432736397, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 5e-06, |
|
"logits/chosen": -2.821899652481079, |
|
"logits/rejected": -2.5310721397399902, |
|
"logps/chosen": -64.77155303955078, |
|
"logps/rejected": -73.11659240722656, |
|
"loss": 0.0006, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.018898095935583115, |
|
"rewards/margins": 0.017503971233963966, |
|
"rewards/rejected": 0.001394127495586872, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"eval_logits/chosen": -2.343594551086426, |
|
"eval_logits/rejected": -2.15059494972229, |
|
"eval_logps/chosen": -304.50653076171875, |
|
"eval_logps/rejected": -278.0669250488281, |
|
"eval_loss": 0.003129809396341443, |
|
"eval_rewards/accuracies": 0.4244999885559082, |
|
"eval_rewards/chosen": -0.054112162441015244, |
|
"eval_rewards/margins": -0.007367968093603849, |
|
"eval_rewards/rejected": -0.046744197607040405, |
|
"eval_runtime": 1422.4177, |
|
"eval_samples_per_second": 1.406, |
|
"eval_steps_per_second": 0.703, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 4.99847706754774e-06, |
|
"logits/chosen": -2.8732924461364746, |
|
"logits/rejected": -2.670675754547119, |
|
"logps/chosen": -80.2041244506836, |
|
"logps/rejected": -76.78239440917969, |
|
"loss": 0.0007, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.0098586305975914, |
|
"rewards/margins": 0.01424575038254261, |
|
"rewards/rejected": -0.004387119319289923, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 4.993910125649561e-06, |
|
"logits/chosen": -2.841202735900879, |
|
"logits/rejected": -2.513766050338745, |
|
"logps/chosen": -60.289215087890625, |
|
"logps/rejected": -46.964500427246094, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.013731713406741619, |
|
"rewards/margins": 0.013367941603064537, |
|
"rewards/rejected": 0.00036377101787365973, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 4.986304738420684e-06, |
|
"logits/chosen": -2.712897777557373, |
|
"logits/rejected": -2.391780138015747, |
|
"logps/chosen": -134.18972778320312, |
|
"logps/rejected": -118.10213470458984, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.0306332316249609, |
|
"rewards/margins": 0.015287762507796288, |
|
"rewards/rejected": 0.015345467254519463, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 4.975670171853926e-06, |
|
"logits/chosen": -2.9027531147003174, |
|
"logits/rejected": -2.519896984100342, |
|
"logps/chosen": -57.02116012573242, |
|
"logps/rejected": -53.072364807128906, |
|
"loss": 0.0009, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": -0.0006130338879302144, |
|
"rewards/margins": 0.008738933131098747, |
|
"rewards/rejected": -0.00935196690261364, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 4.962019382530521e-06, |
|
"logits/chosen": -2.8638014793395996, |
|
"logits/rejected": -2.5246753692626953, |
|
"logps/chosen": -61.296531677246094, |
|
"logps/rejected": -76.2929916381836, |
|
"loss": 0.0014, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.025341859087347984, |
|
"rewards/margins": 0.015296203084290028, |
|
"rewards/rejected": -0.04063806310296059, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 4.9453690018345144e-06, |
|
"logits/chosen": -2.8262181282043457, |
|
"logits/rejected": -2.3856072425842285, |
|
"logps/chosen": -141.7677459716797, |
|
"logps/rejected": -134.89096069335938, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.005657578818500042, |
|
"rewards/margins": 0.017082063481211662, |
|
"rewards/rejected": -0.011424483731389046, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 4.925739315689991e-06, |
|
"logits/chosen": -2.871880531311035, |
|
"logits/rejected": -2.5630600452423096, |
|
"logps/chosen": -61.668121337890625, |
|
"logps/rejected": -58.292213439941406, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.016033068299293518, |
|
"rewards/margins": 0.01451338641345501, |
|
"rewards/rejected": -0.030546456575393677, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.903154239845798e-06, |
|
"logits/chosen": -2.893052339553833, |
|
"logits/rejected": -2.682218074798584, |
|
"logps/chosen": -92.70951080322266, |
|
"logps/rejected": -104.5350570678711, |
|
"loss": 0.0014, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": -0.004353048745542765, |
|
"rewards/margins": 0.005879141390323639, |
|
"rewards/rejected": -0.010232190601527691, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 4.8776412907378845e-06, |
|
"logits/chosen": -2.6648871898651123, |
|
"logits/rejected": -2.4408602714538574, |
|
"logps/chosen": -68.06556701660156, |
|
"logps/rejected": -83.47774505615234, |
|
"loss": 0.0016, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": -0.023349497467279434, |
|
"rewards/margins": 0.0038781266193836927, |
|
"rewards/rejected": -0.027227621525526047, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 4.849231551964771e-06, |
|
"logits/chosen": -2.9131691455841064, |
|
"logits/rejected": -2.6140384674072266, |
|
"logps/chosen": -118.0514907836914, |
|
"logps/rejected": -105.72004699707031, |
|
"loss": 0.0025, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": -0.01497165858745575, |
|
"rewards/margins": 0.009273095987737179, |
|
"rewards/rejected": -0.024244757369160652, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"eval_logits/chosen": -2.3611958026885986, |
|
"eval_logits/rejected": -2.168394088745117, |
|
"eval_logps/chosen": -300.2420349121094, |
|
"eval_logps/rejected": -274.46185302734375, |
|
"eval_loss": 0.0032995175570249557, |
|
"eval_rewards/accuracies": 0.4909999966621399, |
|
"eval_rewards/chosen": -0.011466986499726772, |
|
"eval_rewards/margins": -0.0007737927371636033, |
|
"eval_rewards/rejected": -0.010693194344639778, |
|
"eval_runtime": 1422.8869, |
|
"eval_samples_per_second": 1.406, |
|
"eval_steps_per_second": 0.703, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 4.817959636416969e-06, |
|
"logits/chosen": -2.7499923706054688, |
|
"logits/rejected": -2.4605581760406494, |
|
"logps/chosen": -60.11127471923828, |
|
"logps/rejected": -57.330078125, |
|
"loss": 0.0008, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 2.1754764020442963e-05, |
|
"rewards/margins": 0.01557920128107071, |
|
"rewards/rejected": -0.015557448379695415, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.783863644106502e-06, |
|
"logits/chosen": -2.8885045051574707, |
|
"logits/rejected": -2.744602680206299, |
|
"logps/chosen": -133.10980224609375, |
|
"logps/rejected": -116.32179260253906, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.012220713309943676, |
|
"rewards/margins": 0.013802575878798962, |
|
"rewards/rejected": -0.0015818632673472166, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.746985115747918e-06, |
|
"logits/chosen": -2.6919429302215576, |
|
"logits/rejected": -2.4415671825408936, |
|
"logps/chosen": -57.36275100708008, |
|
"logps/rejected": -43.19419479370117, |
|
"loss": 0.0008, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.01155207958072424, |
|
"rewards/margins": 0.017826344817876816, |
|
"rewards/rejected": -0.0062742652371525764, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 4.707368982147318e-06, |
|
"logits/chosen": -2.7176296710968018, |
|
"logits/rejected": -2.6274712085723877, |
|
"logps/chosen": -59.256858825683594, |
|
"logps/rejected": -60.37900924682617, |
|
"loss": 0.0019, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.004980463068932295, |
|
"rewards/margins": 0.006475840695202351, |
|
"rewards/rejected": -0.0014953784411773086, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.665063509461098e-06, |
|
"logits/chosen": -2.8012423515319824, |
|
"logits/rejected": -2.525057554244995, |
|
"logps/chosen": -98.50082397460938, |
|
"logps/rejected": -90.67588806152344, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.008941109292209148, |
|
"rewards/margins": 0.012530329637229443, |
|
"rewards/rejected": -0.021471437066793442, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 4.620120240391065e-06, |
|
"logits/chosen": -2.762904644012451, |
|
"logits/rejected": -2.579153537750244, |
|
"logps/chosen": -79.9581069946289, |
|
"logps/rejected": -84.54290771484375, |
|
"loss": 0.0016, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": -0.008370987139642239, |
|
"rewards/margins": 0.0011044790735468268, |
|
"rewards/rejected": -0.009475464932620525, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.572593931387604e-06, |
|
"logits/chosen": -2.697049140930176, |
|
"logits/rejected": -2.420090675354004, |
|
"logps/chosen": -68.5470199584961, |
|
"logps/rejected": -59.51050567626953, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": -0.007652081549167633, |
|
"rewards/margins": 0.009131882339715958, |
|
"rewards/rejected": -0.01678396575152874, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.522542485937369e-06, |
|
"logits/chosen": -2.8917226791381836, |
|
"logits/rejected": -2.56544828414917, |
|
"logps/chosen": -133.83926391601562, |
|
"logps/rejected": -106.41746520996094, |
|
"loss": 0.0014, |
|
"rewards/accuracies": 0.4000000059604645, |
|
"rewards/chosen": 0.0028765772003680468, |
|
"rewards/margins": 0.0040840087458491325, |
|
"rewards/rejected": -0.001207432011142373, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 4.470026884016805e-06, |
|
"logits/chosen": -2.8042643070220947, |
|
"logits/rejected": -2.6527392864227295, |
|
"logps/chosen": -119.4009017944336, |
|
"logps/rejected": -110.30342102050781, |
|
"loss": 0.0015, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.002782848197966814, |
|
"rewards/margins": 0.009942631237208843, |
|
"rewards/rejected": -0.007159784436225891, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 4.415111107797445e-06, |
|
"logits/chosen": -2.963005781173706, |
|
"logits/rejected": -2.809359073638916, |
|
"logps/chosen": -124.54817199707031, |
|
"logps/rejected": -123.4760513305664, |
|
"loss": 0.0009, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.008486522361636162, |
|
"rewards/margins": 0.007400536444038153, |
|
"rewards/rejected": 0.0010859851026907563, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"eval_logits/chosen": -2.335963249206543, |
|
"eval_logits/rejected": -2.142676830291748, |
|
"eval_logps/chosen": -301.29595947265625, |
|
"eval_logps/rejected": -275.5566711425781, |
|
"eval_loss": 0.0030309471767395735, |
|
"eval_rewards/accuracies": 0.4934999942779541, |
|
"eval_rewards/chosen": -0.02200666442513466, |
|
"eval_rewards/margins": -0.0003654572064988315, |
|
"eval_rewards/rejected": -0.021641207858920097, |
|
"eval_runtime": 1423.8728, |
|
"eval_samples_per_second": 1.405, |
|
"eval_steps_per_second": 0.702, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 4.357862063693486e-06, |
|
"logits/chosen": -2.8513238430023193, |
|
"logits/rejected": -2.672865390777588, |
|
"logps/chosen": -66.75782775878906, |
|
"logps/rejected": -73.05040740966797, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.007263772189617157, |
|
"rewards/margins": 0.007496596314013004, |
|
"rewards/rejected": -0.014760365709662437, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 4.2983495008466285e-06, |
|
"logits/chosen": -2.7180657386779785, |
|
"logits/rejected": -2.3951573371887207, |
|
"logps/chosen": -88.17804718017578, |
|
"logps/rejected": -78.4397964477539, |
|
"loss": 0.0007, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.009861504659056664, |
|
"rewards/margins": 0.019821077585220337, |
|
"rewards/rejected": -0.009959569200873375, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 4.236645926147493e-06, |
|
"logits/chosen": -2.687565326690674, |
|
"logits/rejected": -2.4989490509033203, |
|
"logps/chosen": -130.00393676757812, |
|
"logps/rejected": -130.13922119140625, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.5, |
|
"rewards/chosen": 0.04140586778521538, |
|
"rewards/margins": 0.013412979431450367, |
|
"rewards/rejected": 0.027992893010377884, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 4.172826515897146e-06, |
|
"logits/chosen": -2.6931281089782715, |
|
"logits/rejected": -2.3428401947021484, |
|
"logps/chosen": -79.95687103271484, |
|
"logps/rejected": -78.25990295410156, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.010445808991789818, |
|
"rewards/margins": 0.013384808786213398, |
|
"rewards/rejected": -0.0029389983974397182, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 4.106969024216348e-06, |
|
"logits/chosen": -2.71099853515625, |
|
"logits/rejected": -2.4201858043670654, |
|
"logps/chosen": -51.87015914916992, |
|
"logps/rejected": -49.29432678222656, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": -0.0006995106232352555, |
|
"rewards/margins": 0.008277077227830887, |
|
"rewards/rejected": -0.008976588025689125, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 4.039153688314146e-06, |
|
"logits/chosen": -2.80985689163208, |
|
"logits/rejected": -2.5829150676727295, |
|
"logps/chosen": -73.7370834350586, |
|
"logps/rejected": -69.81620788574219, |
|
"loss": 0.0008, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.014964630827307701, |
|
"rewards/margins": 0.014514106325805187, |
|
"rewards/rejected": 0.0004505239485297352, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.969463130731183e-06, |
|
"logits/chosen": -2.7932417392730713, |
|
"logits/rejected": -2.5239012241363525, |
|
"logps/chosen": -114.29008483886719, |
|
"logps/rejected": -102.12156677246094, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.004044980742037296, |
|
"rewards/margins": 0.010396437719464302, |
|
"rewards/rejected": -0.006351456046104431, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.897982258676867e-06, |
|
"logits/chosen": -2.7236623764038086, |
|
"logits/rejected": -2.5601420402526855, |
|
"logps/chosen": -84.38848876953125, |
|
"logps/rejected": -85.55775451660156, |
|
"loss": 0.0008, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.012120635248720646, |
|
"rewards/margins": 0.010280453599989414, |
|
"rewards/rejected": 0.0018401825800538063, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 3.824798160583012e-06, |
|
"logits/chosen": -2.8765292167663574, |
|
"logits/rejected": -2.7375664710998535, |
|
"logps/chosen": -107.2117919921875, |
|
"logps/rejected": -103.0725326538086, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": 0.021095167845487595, |
|
"rewards/margins": 0.009734300896525383, |
|
"rewards/rejected": 0.011360866948962212, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 3.7500000000000005e-06, |
|
"logits/chosen": -2.8528428077697754, |
|
"logits/rejected": -2.539654493331909, |
|
"logps/chosen": -86.22474670410156, |
|
"logps/rejected": -74.02642822265625, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.02159132994711399, |
|
"rewards/margins": 0.008513709530234337, |
|
"rewards/rejected": 0.013077618554234505, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"eval_logits/chosen": -2.340451955795288, |
|
"eval_logits/rejected": -2.1462416648864746, |
|
"eval_logps/chosen": -300.6561279296875, |
|
"eval_logps/rejected": -274.8155822753906, |
|
"eval_loss": 0.0034076529555022717, |
|
"eval_rewards/accuracies": 0.4934999942779541, |
|
"eval_rewards/chosen": -0.015608005225658417, |
|
"eval_rewards/margins": -0.0013772795209661126, |
|
"eval_rewards/rejected": -0.014230725355446339, |
|
"eval_runtime": 1422.7868, |
|
"eval_samples_per_second": 1.406, |
|
"eval_steps_per_second": 0.703, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 3.6736789069647273e-06, |
|
"logits/chosen": -2.8410146236419678, |
|
"logits/rejected": -2.368011713027954, |
|
"logps/chosen": -64.5174789428711, |
|
"logps/rejected": -56.3655891418457, |
|
"loss": 0.0015, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": 0.02277212217450142, |
|
"rewards/margins": 0.015335053205490112, |
|
"rewards/rejected": 0.007437069900333881, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 3.595927866972694e-06, |
|
"logits/chosen": -2.773725986480713, |
|
"logits/rejected": -2.6576743125915527, |
|
"logps/chosen": -95.10247802734375, |
|
"logps/rejected": -95.42603302001953, |
|
"loss": 0.0007, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.0311850905418396, |
|
"rewards/margins": 0.010761396028101444, |
|
"rewards/rejected": 0.02042369544506073, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 3.516841607689501e-06, |
|
"logits/chosen": -2.824734926223755, |
|
"logits/rejected": -2.556206226348877, |
|
"logps/chosen": -108.12715911865234, |
|
"logps/rejected": -96.86244201660156, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.032226406037807465, |
|
"rewards/margins": 0.0070249298587441444, |
|
"rewards/rejected": 0.025201475247740746, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 3.436516483539781e-06, |
|
"logits/chosen": -2.926262617111206, |
|
"logits/rejected": -2.665631055831909, |
|
"logps/chosen": -61.947792053222656, |
|
"logps/rejected": -50.202911376953125, |
|
"loss": 0.0006, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 0.0329199843108654, |
|
"rewards/margins": 0.02404285967350006, |
|
"rewards/rejected": 0.00887712650001049, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 3.3550503583141726e-06, |
|
"logits/chosen": -2.8114187717437744, |
|
"logits/rejected": -2.5821549892425537, |
|
"logps/chosen": -56.29111862182617, |
|
"logps/rejected": -47.3681526184082, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": 0.0194512028247118, |
|
"rewards/margins": 0.006624855101108551, |
|
"rewards/rejected": 0.012826347723603249, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 3.272542485937369e-06, |
|
"logits/chosen": -2.826735019683838, |
|
"logits/rejected": -2.5145363807678223, |
|
"logps/chosen": -141.8760986328125, |
|
"logps/rejected": -135.92344665527344, |
|
"loss": 0.0016, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": 0.02360866777598858, |
|
"rewards/margins": 0.004182119388133287, |
|
"rewards/rejected": 0.019426550716161728, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 3.189093389542498e-06, |
|
"logits/chosen": -2.9071967601776123, |
|
"logits/rejected": -2.701315402984619, |
|
"logps/chosen": -52.77317428588867, |
|
"logps/rejected": -43.23090362548828, |
|
"loss": 0.0008, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.028543706983327866, |
|
"rewards/margins": 0.014000201597809792, |
|
"rewards/rejected": 0.014543506316840649, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 3.1048047389991693e-06, |
|
"logits/chosen": -2.944225788116455, |
|
"logits/rejected": -2.7375664710998535, |
|
"logps/chosen": -123.64131927490234, |
|
"logps/rejected": -121.98396301269531, |
|
"loss": 0.0007, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.025555819272994995, |
|
"rewards/margins": 0.01265828125178814, |
|
"rewards/rejected": 0.012897538021206856, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 3.019779227044398e-06, |
|
"logits/chosen": -2.7981390953063965, |
|
"logits/rejected": -2.5953376293182373, |
|
"logps/chosen": -100.98303985595703, |
|
"logps/rejected": -113.0623550415039, |
|
"loss": 0.0009, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.021701429039239883, |
|
"rewards/margins": 0.013312505558133125, |
|
"rewards/rejected": 0.008388923481106758, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 2.9341204441673267e-06, |
|
"logits/chosen": -2.7481579780578613, |
|
"logits/rejected": -2.5245938301086426, |
|
"logps/chosen": -215.2183074951172, |
|
"logps/rejected": -199.92349243164062, |
|
"loss": 0.0011, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.024366769939661026, |
|
"rewards/margins": 0.014933845028281212, |
|
"rewards/rejected": 0.00943292398005724, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_logits/chosen": -2.3392491340637207, |
|
"eval_logits/rejected": -2.1454310417175293, |
|
"eval_logps/chosen": -304.7457275390625, |
|
"eval_logps/rejected": -278.41650390625, |
|
"eval_loss": 0.0036536348052322865, |
|
"eval_rewards/accuracies": 0.4519999921321869, |
|
"eval_rewards/chosen": -0.05650419369339943, |
|
"eval_rewards/margins": -0.0062644826248288155, |
|
"eval_rewards/rejected": -0.05023970827460289, |
|
"eval_runtime": 1422.9482, |
|
"eval_samples_per_second": 1.406, |
|
"eval_steps_per_second": 0.703, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 2.847932752400164e-06, |
|
"logits/chosen": -2.8486416339874268, |
|
"logits/rejected": -2.635902166366577, |
|
"logps/chosen": -69.38902282714844, |
|
"logps/rejected": -46.49626159667969, |
|
"loss": 0.0112, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 0.062225647270679474, |
|
"rewards/margins": 0.07529176026582718, |
|
"rewards/rejected": -0.013066110201179981, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 2.761321158169134e-06, |
|
"logits/chosen": -2.7715601921081543, |
|
"logits/rejected": -2.531752347946167, |
|
"logps/chosen": -59.361083984375, |
|
"logps/rejected": -49.393150329589844, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.018055101856589317, |
|
"rewards/margins": 0.018152302131056786, |
|
"rewards/rejected": -9.720241359900683e-05, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 2.6743911843603134e-06, |
|
"logits/chosen": -2.88091778755188, |
|
"logits/rejected": -2.714815139770508, |
|
"logps/chosen": -75.8203353881836, |
|
"logps/rejected": -91.77738952636719, |
|
"loss": 0.0053, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": 0.03629200905561447, |
|
"rewards/margins": 0.05072981119155884, |
|
"rewards/rejected": -0.014437800273299217, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 2.587248741756253e-06, |
|
"logits/chosen": -2.7633297443389893, |
|
"logits/rejected": -2.4665675163269043, |
|
"logps/chosen": -100.9280776977539, |
|
"logps/rejected": -104.80192565917969, |
|
"loss": 0.0037, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.0258990116417408, |
|
"rewards/margins": 0.028545299544930458, |
|
"rewards/rejected": -0.002646292094141245, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 2.5e-06, |
|
"logits/chosen": -2.8614132404327393, |
|
"logits/rejected": -2.539499282836914, |
|
"logps/chosen": -58.6580810546875, |
|
"logps/rejected": -53.3350715637207, |
|
"loss": 0.0207, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.06129790097475052, |
|
"rewards/margins": 0.07703909277915955, |
|
"rewards/rejected": -0.01574121043086052, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 2.4127512582437486e-06, |
|
"logits/chosen": -2.688095808029175, |
|
"logits/rejected": -2.4665043354034424, |
|
"logps/chosen": -87.18086242675781, |
|
"logps/rejected": -87.13175964355469, |
|
"loss": 0.0036, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": -0.012053154408931732, |
|
"rewards/margins": 0.03777941316366196, |
|
"rewards/rejected": -0.04983257129788399, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 2.325608815639687e-06, |
|
"logits/chosen": -2.9195265769958496, |
|
"logits/rejected": -2.6637749671936035, |
|
"logps/chosen": -130.01551818847656, |
|
"logps/rejected": -122.54390716552734, |
|
"loss": 0.0058, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.012607279233634472, |
|
"rewards/margins": 0.04042843356728554, |
|
"rewards/rejected": -0.02782115340232849, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 2.238678841830867e-06, |
|
"logits/chosen": -2.6946258544921875, |
|
"logits/rejected": -2.5128772258758545, |
|
"logps/chosen": -143.2740478515625, |
|
"logps/rejected": -139.3197479248047, |
|
"loss": 0.0026, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.0010408941889181733, |
|
"rewards/margins": 0.011470312252640724, |
|
"rewards/rejected": -0.010429417714476585, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 2.1520672475998374e-06, |
|
"logits/chosen": -2.934413433074951, |
|
"logits/rejected": -2.6382033824920654, |
|
"logps/chosen": -97.65046691894531, |
|
"logps/rejected": -84.37725830078125, |
|
"loss": 0.0031, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.02459452673792839, |
|
"rewards/margins": 0.02938224747776985, |
|
"rewards/rejected": -0.004787721671164036, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 2.0658795558326745e-06, |
|
"logits/chosen": -2.9206016063690186, |
|
"logits/rejected": -2.740886926651001, |
|
"logps/chosen": -59.428001403808594, |
|
"logps/rejected": -76.50398254394531, |
|
"loss": 0.0116, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.056895893067121506, |
|
"rewards/margins": 0.06610508263111115, |
|
"rewards/rejected": -0.00920918770134449, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"eval_logits/chosen": -2.3448727130889893, |
|
"eval_logits/rejected": -2.152658700942993, |
|
"eval_logps/chosen": -301.92657470703125, |
|
"eval_logps/rejected": -275.6791076660156, |
|
"eval_loss": 0.004949736408889294, |
|
"eval_rewards/accuracies": 0.44350001215934753, |
|
"eval_rewards/chosen": -0.028312114998698235, |
|
"eval_rewards/margins": -0.005446531809866428, |
|
"eval_rewards/rejected": -0.02286558412015438, |
|
"eval_runtime": 1423.4392, |
|
"eval_samples_per_second": 1.405, |
|
"eval_steps_per_second": 0.703, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 1.9802207729556023e-06, |
|
"logits/chosen": -2.9324729442596436, |
|
"logits/rejected": -2.5813498497009277, |
|
"logps/chosen": -204.2996063232422, |
|
"logps/rejected": -187.82083129882812, |
|
"loss": 0.0123, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": -0.0035219527781009674, |
|
"rewards/margins": -0.00877875555306673, |
|
"rewards/rejected": 0.005256800912320614, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 1.895195261000831e-06, |
|
"logits/chosen": -2.8246846199035645, |
|
"logits/rejected": -2.683941602706909, |
|
"logps/chosen": -59.7250862121582, |
|
"logps/rejected": -53.9716682434082, |
|
"loss": 0.0048, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.055084872990846634, |
|
"rewards/margins": 0.05010204762220383, |
|
"rewards/rejected": 0.004982819315046072, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 1.8109066104575023e-06, |
|
"logits/chosen": -2.9231362342834473, |
|
"logits/rejected": -2.695604085922241, |
|
"logps/chosen": -56.9963264465332, |
|
"logps/rejected": -63.59523391723633, |
|
"loss": 0.0045, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.04905426502227783, |
|
"rewards/margins": 0.04801047965884209, |
|
"rewards/rejected": 0.0010437825694680214, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 1.7274575140626318e-06, |
|
"logits/chosen": -2.905579090118408, |
|
"logits/rejected": -2.642456531524658, |
|
"logps/chosen": -98.58207702636719, |
|
"logps/rejected": -93.85881805419922, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.03151048347353935, |
|
"rewards/margins": 0.029562344774603844, |
|
"rewards/rejected": 0.0019481380004435778, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 1.6449496416858285e-06, |
|
"logits/chosen": -3.0052525997161865, |
|
"logits/rejected": -2.7315216064453125, |
|
"logps/chosen": -89.30766296386719, |
|
"logps/rejected": -104.63911437988281, |
|
"loss": 0.0338, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.0381547287106514, |
|
"rewards/margins": 0.0724136009812355, |
|
"rewards/rejected": -0.034258872270584106, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 1.56348351646022e-06, |
|
"logits/chosen": -2.6560475826263428, |
|
"logits/rejected": -2.5114564895629883, |
|
"logps/chosen": -131.77236938476562, |
|
"logps/rejected": -127.79585266113281, |
|
"loss": 0.0026, |
|
"rewards/accuracies": 0.550000011920929, |
|
"rewards/chosen": 0.020144177600741386, |
|
"rewards/margins": 0.023866238072514534, |
|
"rewards/rejected": -0.003722056746482849, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 1.4831583923105e-06, |
|
"logits/chosen": -2.8219382762908936, |
|
"logits/rejected": -2.446211814880371, |
|
"logps/chosen": -68.78748321533203, |
|
"logps/rejected": -53.5217170715332, |
|
"loss": 0.0027, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.04556558281183243, |
|
"rewards/margins": 0.04576994851231575, |
|
"rewards/rejected": -0.00020437035709619522, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 1.4040721330273063e-06, |
|
"logits/chosen": -2.910677433013916, |
|
"logits/rejected": -2.7675867080688477, |
|
"logps/chosen": -59.946563720703125, |
|
"logps/rejected": -60.795616149902344, |
|
"loss": 0.0106, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.05829508230090141, |
|
"rewards/margins": 0.04981726408004761, |
|
"rewards/rejected": 0.008477816358208656, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 1.3263210930352737e-06, |
|
"logits/chosen": -2.8532004356384277, |
|
"logits/rejected": -2.6016685962677, |
|
"logps/chosen": -196.2278289794922, |
|
"logps/rejected": -204.3356170654297, |
|
"loss": 0.0076, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.008091023191809654, |
|
"rewards/margins": 0.023959284648299217, |
|
"rewards/rejected": -0.015868261456489563, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 1.2500000000000007e-06, |
|
"logits/chosen": -2.841736316680908, |
|
"logits/rejected": -2.5827274322509766, |
|
"logps/chosen": -52.04328155517578, |
|
"logps/rejected": -62.17645263671875, |
|
"loss": 0.015, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.026109540835022926, |
|
"rewards/margins": 0.010281763039529324, |
|
"rewards/rejected": 0.015827778726816177, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"eval_logits/chosen": -2.3585643768310547, |
|
"eval_logits/rejected": -2.164954662322998, |
|
"eval_logps/chosen": -301.7041320800781, |
|
"eval_logps/rejected": -275.2170104980469, |
|
"eval_loss": 0.006464843638241291, |
|
"eval_rewards/accuracies": 0.4449999928474426, |
|
"eval_rewards/chosen": -0.026087837293744087, |
|
"eval_rewards/margins": -0.007843218743801117, |
|
"eval_rewards/rejected": -0.01824462041258812, |
|
"eval_runtime": 1422.5043, |
|
"eval_samples_per_second": 1.406, |
|
"eval_steps_per_second": 0.703, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 1.1752018394169882e-06, |
|
"logits/chosen": -2.8012290000915527, |
|
"logits/rejected": -2.348876476287842, |
|
"logps/chosen": -50.55641555786133, |
|
"logps/rejected": -46.36492156982422, |
|
"loss": 0.0033, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 0.0446331761777401, |
|
"rewards/margins": 0.04935276508331299, |
|
"rewards/rejected": -0.004719586111605167, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 1.1020177413231334e-06, |
|
"logits/chosen": -2.766397476196289, |
|
"logits/rejected": -2.423826217651367, |
|
"logps/chosen": -111.70279693603516, |
|
"logps/rejected": -129.45628356933594, |
|
"loss": 0.0049, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": 0.04479371756315231, |
|
"rewards/margins": 0.06533690541982651, |
|
"rewards/rejected": -0.020543191581964493, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 1.0305368692688175e-06, |
|
"logits/chosen": -2.9610862731933594, |
|
"logits/rejected": -2.7894320487976074, |
|
"logps/chosen": -86.23368835449219, |
|
"logps/rejected": -86.16008758544922, |
|
"loss": 0.0027, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 0.04561948776245117, |
|
"rewards/margins": 0.04706469178199768, |
|
"rewards/rejected": -0.0014452062314376235, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 9.608463116858544e-07, |
|
"logits/chosen": -2.8900036811828613, |
|
"logits/rejected": -2.4443981647491455, |
|
"logps/chosen": -70.54168701171875, |
|
"logps/rejected": -50.89817428588867, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": 0.031780801713466644, |
|
"rewards/margins": 0.038311202079057693, |
|
"rewards/rejected": -0.006530401296913624, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 8.930309757836517e-07, |
|
"logits/chosen": -2.9060871601104736, |
|
"logits/rejected": -2.524108648300171, |
|
"logps/chosen": -61.922813415527344, |
|
"logps/rejected": -65.15422821044922, |
|
"loss": 0.0056, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.0495627298951149, |
|
"rewards/margins": 0.04960542917251587, |
|
"rewards/rejected": -4.269964847480878e-05, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 8.271734841028553e-07, |
|
"logits/chosen": -2.644831657409668, |
|
"logits/rejected": -2.305298328399658, |
|
"logps/chosen": -67.29566955566406, |
|
"logps/rejected": -63.685523986816406, |
|
"loss": 0.0035, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.04994013160467148, |
|
"rewards/margins": 0.052011966705322266, |
|
"rewards/rejected": -0.002071837428957224, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 7.633540738525066e-07, |
|
"logits/chosen": -2.9390599727630615, |
|
"logits/rejected": -2.688645839691162, |
|
"logps/chosen": -65.12007904052734, |
|
"logps/rejected": -62.56877517700195, |
|
"loss": 0.1119, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.05300932377576828, |
|
"rewards/margins": 0.12731197476387024, |
|
"rewards/rejected": -0.07430265843868256, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 7.016504991533727e-07, |
|
"logits/chosen": -2.625112295150757, |
|
"logits/rejected": -2.447942018508911, |
|
"logps/chosen": -58.0603141784668, |
|
"logps/rejected": -50.18201446533203, |
|
"loss": 0.0028, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.009573986753821373, |
|
"rewards/margins": 0.0028710602782666683, |
|
"rewards/rejected": 0.006702926941215992, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 6.421379363065142e-07, |
|
"logits/chosen": -2.982449769973755, |
|
"logits/rejected": -2.7615859508514404, |
|
"logps/chosen": -51.4494743347168, |
|
"logps/rejected": -57.383644104003906, |
|
"loss": 0.0061, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.05977553874254227, |
|
"rewards/margins": 0.05702256038784981, |
|
"rewards/rejected": 0.002752984408289194, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 5.848888922025553e-07, |
|
"logits/chosen": -2.7822391986846924, |
|
"logits/rejected": -2.5350170135498047, |
|
"logps/chosen": -43.52353286743164, |
|
"logps/rejected": -54.28175735473633, |
|
"loss": 0.0009, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": 0.03572973236441612, |
|
"rewards/margins": 0.035435039550065994, |
|
"rewards/rejected": 0.0002946966851595789, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"eval_logits/chosen": -2.3689210414886475, |
|
"eval_logits/rejected": -2.174628496170044, |
|
"eval_logps/chosen": -298.3010559082031, |
|
"eval_logps/rejected": -272.15399169921875, |
|
"eval_loss": 0.006914817728102207, |
|
"eval_rewards/accuracies": 0.47200000286102295, |
|
"eval_rewards/chosen": 0.007942860946059227, |
|
"eval_rewards/margins": -0.0044423178769648075, |
|
"eval_rewards/rejected": 0.012385179288685322, |
|
"eval_runtime": 1422.6658, |
|
"eval_samples_per_second": 1.406, |
|
"eval_steps_per_second": 0.703, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 5.299731159831953e-07, |
|
"logits/chosen": -2.868394136428833, |
|
"logits/rejected": -2.6725270748138428, |
|
"logps/chosen": -113.2739486694336, |
|
"logps/rejected": -118.2945327758789, |
|
"loss": 0.0086, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.050359614193439484, |
|
"rewards/margins": 0.05471315234899521, |
|
"rewards/rejected": -0.004353542346507311, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 4.774575140626317e-07, |
|
"logits/chosen": -2.7905325889587402, |
|
"logits/rejected": -2.5390892028808594, |
|
"logps/chosen": -61.987098693847656, |
|
"logps/rejected": -72.47862243652344, |
|
"loss": 0.0051, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.035242609679698944, |
|
"rewards/margins": 0.04513431712985039, |
|
"rewards/rejected": -0.009891709312796593, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 4.27406068612396e-07, |
|
"logits/chosen": -2.744027614593506, |
|
"logits/rejected": -2.451847553253174, |
|
"logps/chosen": -94.64505767822266, |
|
"logps/rejected": -86.02244567871094, |
|
"loss": 0.0009, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.022839754819869995, |
|
"rewards/margins": 0.03421180695295334, |
|
"rewards/rejected": -0.011372053064405918, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 3.798797596089351e-07, |
|
"logits/chosen": -2.88804292678833, |
|
"logits/rejected": -2.348177194595337, |
|
"logps/chosen": -62.28667068481445, |
|
"logps/rejected": -47.92333221435547, |
|
"loss": 0.0024, |
|
"rewards/accuracies": 0.8999999761581421, |
|
"rewards/chosen": 0.031227191910147667, |
|
"rewards/margins": 0.039407771080732346, |
|
"rewards/rejected": -0.008180581033229828, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 3.3493649053890325e-07, |
|
"logits/chosen": -2.8707077503204346, |
|
"logits/rejected": -2.6572961807250977, |
|
"logps/chosen": -97.13041687011719, |
|
"logps/rejected": -102.47623443603516, |
|
"loss": 0.0026, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.015258235856890678, |
|
"rewards/margins": 0.032782141119241714, |
|
"rewards/rejected": -0.017523903399705887, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 2.9263101785268253e-07, |
|
"logits/chosen": -2.6006197929382324, |
|
"logits/rejected": -2.57403826713562, |
|
"logps/chosen": -182.5235595703125, |
|
"logps/rejected": -166.66905212402344, |
|
"loss": 0.001, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.0023535334039479494, |
|
"rewards/margins": 0.01678496040403843, |
|
"rewards/rejected": -0.0144314244389534, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 2.53014884252083e-07, |
|
"logits/chosen": -2.9139671325683594, |
|
"logits/rejected": -2.8268768787384033, |
|
"logps/chosen": -132.52719116210938, |
|
"logps/rejected": -130.03152465820312, |
|
"loss": 0.0014, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.038164813071489334, |
|
"rewards/margins": 0.03188065439462662, |
|
"rewards/rejected": 0.006284159608185291, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 2.1613635589349756e-07, |
|
"logits/chosen": -2.772710084915161, |
|
"logits/rejected": -2.3980655670166016, |
|
"logps/chosen": -47.05355453491211, |
|
"logps/rejected": -44.157875061035156, |
|
"loss": 0.0034, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.03233030438423157, |
|
"rewards/margins": 0.03725982457399368, |
|
"rewards/rejected": -0.00492951599881053, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 1.8204036358303173e-07, |
|
"logits/chosen": -2.9192633628845215, |
|
"logits/rejected": -2.627864360809326, |
|
"logps/chosen": -70.42695617675781, |
|
"logps/rejected": -99.09819030761719, |
|
"loss": 0.1069, |
|
"rewards/accuracies": 0.8500000238418579, |
|
"rewards/chosen": 0.06590764224529266, |
|
"rewards/margins": 0.1522078961133957, |
|
"rewards/rejected": -0.08630025386810303, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 1.507684480352292e-07, |
|
"logits/chosen": -2.7338740825653076, |
|
"logits/rejected": -2.4327640533447266, |
|
"logps/chosen": -48.297508239746094, |
|
"logps/rejected": -40.352996826171875, |
|
"loss": 0.0109, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.037285320460796356, |
|
"rewards/margins": 0.0314871221780777, |
|
"rewards/rejected": 0.005798191763460636, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"eval_logits/chosen": -2.368154764175415, |
|
"eval_logits/rejected": -2.1738698482513428, |
|
"eval_logps/chosen": -298.8583068847656, |
|
"eval_logps/rejected": -272.5880432128906, |
|
"eval_loss": 0.006868657190352678, |
|
"eval_rewards/accuracies": 0.4569999873638153, |
|
"eval_rewards/chosen": 0.0023701086174696684, |
|
"eval_rewards/margins": -0.005675030872225761, |
|
"eval_rewards/rejected": 0.008045138791203499, |
|
"eval_runtime": 1423.7285, |
|
"eval_samples_per_second": 1.405, |
|
"eval_steps_per_second": 0.702, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 1.223587092621162e-07, |
|
"logits/chosen": -3.062129497528076, |
|
"logits/rejected": -2.726337432861328, |
|
"logps/chosen": -103.9426040649414, |
|
"logps/rejected": -109.22408294677734, |
|
"loss": 0.0132, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.01760241389274597, |
|
"rewards/margins": 0.04251215234398842, |
|
"rewards/rejected": -0.024909736588597298, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 9.684576015420277e-08, |
|
"logits/chosen": -2.8683624267578125, |
|
"logits/rejected": -2.5635323524475098, |
|
"logps/chosen": -132.88455200195312, |
|
"logps/rejected": -131.37045288085938, |
|
"loss": 0.0063, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.03422384709119797, |
|
"rewards/margins": 0.05052335187792778, |
|
"rewards/rejected": -0.016299499198794365, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 7.426068431000883e-08, |
|
"logits/chosen": -2.763186454772949, |
|
"logits/rejected": -2.4577767848968506, |
|
"logps/chosen": -62.349082946777344, |
|
"logps/rejected": -48.44455337524414, |
|
"loss": 0.0027, |
|
"rewards/accuracies": 0.949999988079071, |
|
"rewards/chosen": 0.03767557442188263, |
|
"rewards/margins": 0.04414733499288559, |
|
"rewards/rejected": -0.006471761967986822, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 5.463099816548578e-08, |
|
"logits/chosen": -2.749911308288574, |
|
"logits/rejected": -2.5365560054779053, |
|
"logps/chosen": -109.21867370605469, |
|
"logps/rejected": -97.2970962524414, |
|
"loss": 0.0013, |
|
"rewards/accuracies": 0.75, |
|
"rewards/chosen": 0.03286752849817276, |
|
"rewards/margins": 0.033335305750370026, |
|
"rewards/rejected": -0.00046777556417509913, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 3.798061746947995e-08, |
|
"logits/chosen": -2.9467415809631348, |
|
"logits/rejected": -2.6370301246643066, |
|
"logps/chosen": -142.62974548339844, |
|
"logps/rejected": -140.1845703125, |
|
"loss": 0.0015, |
|
"rewards/accuracies": 0.800000011920929, |
|
"rewards/chosen": 0.03325565904378891, |
|
"rewards/margins": 0.027630716562271118, |
|
"rewards/rejected": 0.005624941550195217, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 2.4329828146074096e-08, |
|
"logits/chosen": -2.7907421588897705, |
|
"logits/rejected": -2.5240225791931152, |
|
"logps/chosen": -112.5775146484375, |
|
"logps/rejected": -115.86857604980469, |
|
"loss": 0.0156, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.012037028558552265, |
|
"rewards/margins": -0.006716550327837467, |
|
"rewards/rejected": 0.01875358074903488, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 1.3695261579316776e-08, |
|
"logits/chosen": -3.0265414714813232, |
|
"logits/rejected": -2.6483359336853027, |
|
"logps/chosen": -85.69535064697266, |
|
"logps/rejected": -82.9826889038086, |
|
"loss": 0.0296, |
|
"rewards/accuracies": 0.6499999761581421, |
|
"rewards/chosen": 0.013691084459424019, |
|
"rewards/margins": 0.002223056508228183, |
|
"rewards/rejected": 0.011468032374978065, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 6.089874350439507e-09, |
|
"logits/chosen": -2.6236181259155273, |
|
"logits/rejected": -2.498073101043701, |
|
"logps/chosen": -81.69920349121094, |
|
"logps/rejected": -90.76789855957031, |
|
"loss": 0.0062, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.018706561997532845, |
|
"rewards/margins": 0.04268919676542282, |
|
"rewards/rejected": -0.023982631042599678, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 1.5229324522605949e-09, |
|
"logits/chosen": -2.952378034591675, |
|
"logits/rejected": -2.7012767791748047, |
|
"logps/chosen": -114.75297546386719, |
|
"logps/rejected": -98.20231628417969, |
|
"loss": 0.0073, |
|
"rewards/accuracies": 0.6000000238418579, |
|
"rewards/chosen": 0.020036840811371803, |
|
"rewards/margins": 0.031693797558546066, |
|
"rewards/rejected": -0.011656956747174263, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 0.0, |
|
"logits/chosen": -2.9289968013763428, |
|
"logits/rejected": -2.6982641220092773, |
|
"logps/chosen": -100.9759750366211, |
|
"logps/rejected": -94.3797607421875, |
|
"loss": 0.0015, |
|
"rewards/accuracies": 0.699999988079071, |
|
"rewards/chosen": 0.02024517022073269, |
|
"rewards/margins": 0.02688099816441536, |
|
"rewards/rejected": -0.006635826081037521, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_logits/chosen": -2.3691599369049072, |
|
"eval_logits/rejected": -2.1749157905578613, |
|
"eval_logps/chosen": -298.8498229980469, |
|
"eval_logps/rejected": -272.58660888671875, |
|
"eval_loss": 0.006900906562805176, |
|
"eval_rewards/accuracies": 0.4595000147819519, |
|
"eval_rewards/chosen": 0.002455136040225625, |
|
"eval_rewards/margins": -0.005604185629636049, |
|
"eval_rewards/rejected": 0.008059320971369743, |
|
"eval_runtime": 1422.3751, |
|
"eval_samples_per_second": 1.406, |
|
"eval_steps_per_second": 0.703, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"step": 1000, |
|
"total_flos": 0.0, |
|
"train_loss": 0.005910432943725027, |
|
"train_runtime": 19053.6119, |
|
"train_samples_per_second": 0.105, |
|
"train_steps_per_second": 0.052 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 1000, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 2, |
|
"save_steps": 100, |
|
"total_flos": 0.0, |
|
"train_batch_size": 1, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|