| { |
| "best_metric": 72.2228513192164, |
| "best_model_checkpoint": "experiments/all_data_any_res_adain_finetuning_blip_lr5e-05_bs128_ip512_op256_ep12/checkpoint-40190", |
| "epoch": 7.0, |
| "eval_steps": 500, |
| "global_step": 56266, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.062204528489674046, |
| "grad_norm": 0.12522129714488983, |
| "learning_rate": 2.5917478747667426e-06, |
| "loss": 15.4626, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.12440905697934809, |
| "grad_norm": 0.09409886598587036, |
| "learning_rate": 5.183495749533485e-06, |
| "loss": 0.0492, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.18661358546902215, |
| "grad_norm": 0.052524592727422714, |
| "learning_rate": 7.775243624300229e-06, |
| "loss": 0.0449, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.24881811395869619, |
| "grad_norm": 0.18614421784877777, |
| "learning_rate": 1.036699149906697e-05, |
| "loss": 0.0431, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.31102264244837025, |
| "grad_norm": 0.0732133612036705, |
| "learning_rate": 1.2958739373833714e-05, |
| "loss": 0.0421, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.3732271709380443, |
| "grad_norm": 0.05523424595594406, |
| "learning_rate": 1.5550487248600457e-05, |
| "loss": 0.0411, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.4354316994277183, |
| "grad_norm": 0.04952659457921982, |
| "learning_rate": 1.8142235123367197e-05, |
| "loss": 0.0404, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.49763622791739237, |
| "grad_norm": 0.05076267197728157, |
| "learning_rate": 2.073398299813394e-05, |
| "loss": 0.0396, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.5598407564070664, |
| "grad_norm": 0.04251039773225784, |
| "learning_rate": 2.3325730872900688e-05, |
| "loss": 0.0385, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.6220452848967405, |
| "grad_norm": 0.045916568487882614, |
| "learning_rate": 2.5917478747667428e-05, |
| "loss": 0.0382, |
| "step": 5000 |
| }, |
| { |
| "epoch": 0.6842498133864146, |
| "grad_norm": 0.0338948592543602, |
| "learning_rate": 2.850922662243417e-05, |
| "loss": 0.0376, |
| "step": 5500 |
| }, |
| { |
| "epoch": 0.7464543418760886, |
| "grad_norm": 0.03767935559153557, |
| "learning_rate": 3.1100974497200915e-05, |
| "loss": 0.0374, |
| "step": 6000 |
| }, |
| { |
| "epoch": 0.8086588703657627, |
| "grad_norm": 0.03969201073050499, |
| "learning_rate": 3.369272237196766e-05, |
| "loss": 0.0372, |
| "step": 6500 |
| }, |
| { |
| "epoch": 0.8708633988554366, |
| "grad_norm": 0.037294551730155945, |
| "learning_rate": 3.6284470246734395e-05, |
| "loss": 0.0368, |
| "step": 7000 |
| }, |
| { |
| "epoch": 0.9330679273451107, |
| "grad_norm": 0.04354941099882126, |
| "learning_rate": 3.887621812150114e-05, |
| "loss": 0.0364, |
| "step": 7500 |
| }, |
| { |
| "epoch": 0.9952724558347847, |
| "grad_norm": 0.039071835577487946, |
| "learning_rate": 4.146796599626788e-05, |
| "loss": 0.0363, |
| "step": 8000 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_val_general_accuracy": 58.761243310941595, |
| "eval_val_general_action_correct": 5161, |
| "eval_val_general_loss": 0.03570643812417984, |
| "eval_val_general_runtime": 381.2285, |
| "eval_val_general_samples_per_second": 23.039, |
| "eval_val_general_steps_per_second": 0.092, |
| "eval_val_general_text_acc": 57.78207901628145, |
| "eval_val_general_text_correct": 5075, |
| "eval_val_general_type_acc": 83.9234885574405, |
| "eval_val_general_type_correct": 7371, |
| "step": 8038 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_test_general_accuracy": 60.98969539123542, |
| "eval_test_general_action_correct": 5386, |
| "eval_test_general_loss": 0.03328056260943413, |
| "eval_test_general_runtime": 358.8188, |
| "eval_test_general_samples_per_second": 24.611, |
| "eval_test_general_steps_per_second": 0.098, |
| "eval_test_general_text_acc": 60.33291812931718, |
| "eval_test_general_text_correct": 5328, |
| "eval_test_general_type_acc": 84.04484203374476, |
| "eval_test_general_type_correct": 7422, |
| "step": 8038 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_val_single_accuracy": 73.20165813216289, |
| "eval_val_single_action_correct": 6004, |
| "eval_val_single_loss": 0.0292816162109375, |
| "eval_val_single_overall_val_accuracy": 26.3925802886209, |
| "eval_val_single_runtime": 331.8893, |
| "eval_val_single_samples_per_second": 24.713, |
| "eval_val_single_steps_per_second": 0.099, |
| "eval_val_single_text_acc": 72.70178005364545, |
| "eval_val_single_text_correct": 5963, |
| "eval_val_single_type_acc": 87.3811265544989, |
| "eval_val_single_type_correct": 7167, |
| "step": 8038 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_test_single_accuracy": 73.34963325183375, |
| "eval_test_single_action_correct": 6000, |
| "eval_test_single_loss": 0.02950906753540039, |
| "eval_test_single_overall_test_accuracy": 26.867865728613832, |
| "eval_test_single_runtime": 320.8764, |
| "eval_test_single_samples_per_second": 25.493, |
| "eval_test_single_steps_per_second": 0.1, |
| "eval_test_single_text_acc": 72.86063569682152, |
| "eval_test_single_text_correct": 5960, |
| "eval_test_single_type_acc": 87.70171149144255, |
| "eval_test_single_type_correct": 7174, |
| "step": 8038 |
| }, |
| { |
| "epoch": 1.057476984324459, |
| "grad_norm": 0.04436585307121277, |
| "learning_rate": 4.405971387103463e-05, |
| "loss": 0.036, |
| "step": 8500 |
| }, |
| { |
| "epoch": 1.1196815128141329, |
| "grad_norm": 0.036587439477443695, |
| "learning_rate": 4.6651461745801375e-05, |
| "loss": 0.0356, |
| "step": 9000 |
| }, |
| { |
| "epoch": 1.181886041303807, |
| "grad_norm": 0.036080844700336456, |
| "learning_rate": 4.924320962056811e-05, |
| "loss": 0.0356, |
| "step": 9500 |
| }, |
| { |
| "epoch": 1.244090569793481, |
| "grad_norm": 0.031715475022792816, |
| "learning_rate": 4.979610643935031e-05, |
| "loss": 0.0354, |
| "step": 10000 |
| }, |
| { |
| "epoch": 1.306295098283155, |
| "grad_norm": 0.033429063856601715, |
| "learning_rate": 4.950812118419537e-05, |
| "loss": 0.0354, |
| "step": 10500 |
| }, |
| { |
| "epoch": 1.368499626772829, |
| "grad_norm": 0.03252234682440758, |
| "learning_rate": 4.922013592904044e-05, |
| "loss": 0.035, |
| "step": 11000 |
| }, |
| { |
| "epoch": 1.430704155262503, |
| "grad_norm": 0.030717169865965843, |
| "learning_rate": 4.89321506738855e-05, |
| "loss": 0.0349, |
| "step": 11500 |
| }, |
| { |
| "epoch": 1.4929086837521772, |
| "grad_norm": 0.02983550727367401, |
| "learning_rate": 4.864416541873056e-05, |
| "loss": 0.0346, |
| "step": 12000 |
| }, |
| { |
| "epoch": 1.5551132122418512, |
| "grad_norm": 0.03211957961320877, |
| "learning_rate": 4.835618016357562e-05, |
| "loss": 0.0349, |
| "step": 12500 |
| }, |
| { |
| "epoch": 1.6173177407315253, |
| "grad_norm": 0.03446466848254204, |
| "learning_rate": 4.806819490842069e-05, |
| "loss": 0.0344, |
| "step": 13000 |
| }, |
| { |
| "epoch": 1.6795222692211993, |
| "grad_norm": 0.029267357662320137, |
| "learning_rate": 4.7780209653265756e-05, |
| "loss": 0.0346, |
| "step": 13500 |
| }, |
| { |
| "epoch": 1.7417267977108732, |
| "grad_norm": 0.02974924072623253, |
| "learning_rate": 4.7492224398110816e-05, |
| "loss": 0.0342, |
| "step": 14000 |
| }, |
| { |
| "epoch": 1.8039313262005474, |
| "grad_norm": 0.027496496215462685, |
| "learning_rate": 4.720423914295588e-05, |
| "loss": 0.0341, |
| "step": 14500 |
| }, |
| { |
| "epoch": 1.8661358546902216, |
| "grad_norm": 0.03153369575738907, |
| "learning_rate": 4.691625388780095e-05, |
| "loss": 0.0341, |
| "step": 15000 |
| }, |
| { |
| "epoch": 1.9283403831798955, |
| "grad_norm": 0.024351611733436584, |
| "learning_rate": 4.662826863264601e-05, |
| "loss": 0.0343, |
| "step": 15500 |
| }, |
| { |
| "epoch": 1.9905449116695695, |
| "grad_norm": 0.030573569238185883, |
| "learning_rate": 4.6340283377491076e-05, |
| "loss": 0.034, |
| "step": 16000 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_val_general_accuracy": 65.2282819082318, |
| "eval_val_general_action_correct": 5729, |
| "eval_val_general_loss": 0.03412911668419838, |
| "eval_val_general_runtime": 376.6551, |
| "eval_val_general_samples_per_second": 23.318, |
| "eval_val_general_steps_per_second": 0.093, |
| "eval_val_general_text_acc": 64.26050324490492, |
| "eval_val_general_text_correct": 5644, |
| "eval_val_general_type_acc": 86.05260161675965, |
| "eval_val_general_type_correct": 7558, |
| "step": 16076 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_test_general_accuracy": 66.52700713395991, |
| "eval_test_general_action_correct": 5875, |
| "eval_test_general_loss": 0.031758010387420654, |
| "eval_test_general_runtime": 351.859, |
| "eval_test_general_samples_per_second": 25.098, |
| "eval_test_general_steps_per_second": 0.099, |
| "eval_test_general_text_acc": 65.85890612614654, |
| "eval_test_general_text_correct": 5816, |
| "eval_test_general_type_acc": 85.89061261465292, |
| "eval_test_general_type_correct": 7585, |
| "step": 16076 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_val_single_accuracy": 80.48037064130699, |
| "eval_val_single_action_correct": 6601, |
| "eval_val_single_loss": 0.02729242481291294, |
| "eval_val_single_overall_val_accuracy": 29.141730509907756, |
| "eval_val_single_runtime": 333.3923, |
| "eval_val_single_samples_per_second": 24.602, |
| "eval_val_single_steps_per_second": 0.099, |
| "eval_val_single_text_acc": 80.04145330407218, |
| "eval_val_single_text_correct": 6565, |
| "eval_val_single_type_acc": 90.27066569129481, |
| "eval_val_single_type_correct": 7404, |
| "step": 16076 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_test_single_accuracy": 81.38141809290954, |
| "eval_test_single_action_correct": 6657, |
| "eval_test_single_loss": 0.027449607849121094, |
| "eval_test_single_overall_test_accuracy": 29.581685045373888, |
| "eval_test_single_runtime": 333.3531, |
| "eval_test_single_samples_per_second": 24.539, |
| "eval_test_single_steps_per_second": 0.096, |
| "eval_test_single_text_acc": 81.01466992665036, |
| "eval_test_single_text_correct": 6627, |
| "eval_test_single_type_acc": 90.79462102689486, |
| "eval_test_single_type_correct": 7427, |
| "step": 16076 |
| }, |
| { |
| "epoch": 2.0527494401592437, |
| "grad_norm": 0.027034949511289597, |
| "learning_rate": 4.605229812233614e-05, |
| "loss": 0.0336, |
| "step": 16500 |
| }, |
| { |
| "epoch": 2.114953968648918, |
| "grad_norm": 0.026598777621984482, |
| "learning_rate": 4.57643128671812e-05, |
| "loss": 0.0335, |
| "step": 17000 |
| }, |
| { |
| "epoch": 2.1771584971385916, |
| "grad_norm": 0.0320325531065464, |
| "learning_rate": 4.547632761202626e-05, |
| "loss": 0.0335, |
| "step": 17500 |
| }, |
| { |
| "epoch": 2.2393630256282657, |
| "grad_norm": 0.02739659883081913, |
| "learning_rate": 4.5188342356871336e-05, |
| "loss": 0.0333, |
| "step": 18000 |
| }, |
| { |
| "epoch": 2.30156755411794, |
| "grad_norm": 0.029264872893691063, |
| "learning_rate": 4.4900357101716396e-05, |
| "loss": 0.0333, |
| "step": 18500 |
| }, |
| { |
| "epoch": 2.363772082607614, |
| "grad_norm": 0.03222252428531647, |
| "learning_rate": 4.4612371846561455e-05, |
| "loss": 0.0334, |
| "step": 19000 |
| }, |
| { |
| "epoch": 2.425976611097288, |
| "grad_norm": 0.03381313756108284, |
| "learning_rate": 4.432438659140652e-05, |
| "loss": 0.0333, |
| "step": 19500 |
| }, |
| { |
| "epoch": 2.488181139586962, |
| "grad_norm": 0.028583131730556488, |
| "learning_rate": 4.403640133625159e-05, |
| "loss": 0.0333, |
| "step": 20000 |
| }, |
| { |
| "epoch": 2.550385668076636, |
| "grad_norm": 0.02700190059840679, |
| "learning_rate": 4.374841608109665e-05, |
| "loss": 0.0334, |
| "step": 20500 |
| }, |
| { |
| "epoch": 2.61259019656631, |
| "grad_norm": 0.026916461065411568, |
| "learning_rate": 4.346043082594171e-05, |
| "loss": 0.0332, |
| "step": 21000 |
| }, |
| { |
| "epoch": 2.674794725055984, |
| "grad_norm": 0.028127728030085564, |
| "learning_rate": 4.317244557078678e-05, |
| "loss": 0.0331, |
| "step": 21500 |
| }, |
| { |
| "epoch": 2.736999253545658, |
| "grad_norm": 0.02740449644625187, |
| "learning_rate": 4.288446031563184e-05, |
| "loss": 0.0333, |
| "step": 22000 |
| }, |
| { |
| "epoch": 2.799203782035332, |
| "grad_norm": 0.027998166158795357, |
| "learning_rate": 4.25964750604769e-05, |
| "loss": 0.0329, |
| "step": 22500 |
| }, |
| { |
| "epoch": 2.861408310525006, |
| "grad_norm": 0.03141350299119949, |
| "learning_rate": 4.230848980532197e-05, |
| "loss": 0.033, |
| "step": 23000 |
| }, |
| { |
| "epoch": 2.9236128390146803, |
| "grad_norm": 0.02853882871568203, |
| "learning_rate": 4.2020504550167035e-05, |
| "loss": 0.0329, |
| "step": 23500 |
| }, |
| { |
| "epoch": 2.9858173675043544, |
| "grad_norm": 0.027763281017541885, |
| "learning_rate": 4.1732519295012095e-05, |
| "loss": 0.0329, |
| "step": 24000 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_val_general_accuracy": 68.01776158487988, |
| "eval_val_general_action_correct": 5974, |
| "eval_val_general_loss": 0.033568356186151505, |
| "eval_val_general_runtime": 381.1985, |
| "eval_val_general_samples_per_second": 23.04, |
| "eval_val_general_steps_per_second": 0.092, |
| "eval_val_general_text_acc": 67.09552544688603, |
| "eval_val_general_text_correct": 5893, |
| "eval_val_general_type_acc": 86.69019697142207, |
| "eval_val_general_type_correct": 7614, |
| "step": 24114 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_test_general_accuracy": 69.84486468123656, |
| "eval_test_general_action_correct": 6168, |
| "eval_test_general_loss": 0.031217793002724648, |
| "eval_test_general_runtime": 355.8903, |
| "eval_test_general_samples_per_second": 24.814, |
| "eval_test_general_steps_per_second": 0.098, |
| "eval_test_general_text_acc": 69.2107349111086, |
| "eval_test_general_text_correct": 6112, |
| "eval_test_general_type_acc": 87.2834333597554, |
| "eval_test_general_type_correct": 7708, |
| "step": 24114 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_val_single_accuracy": 83.57717629846378, |
| "eval_val_single_action_correct": 6855, |
| "eval_val_single_loss": 0.02634221874177456, |
| "eval_val_single_overall_val_accuracy": 30.318987576668736, |
| "eval_val_single_runtime": 330.8711, |
| "eval_val_single_samples_per_second": 24.789, |
| "eval_val_single_steps_per_second": 0.1, |
| "eval_val_single_text_acc": 83.15045110948549, |
| "eval_val_single_text_correct": 6820, |
| "eval_val_single_type_acc": 92.4530602292124, |
| "eval_val_single_type_correct": 7583, |
| "step": 24114 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_test_single_accuracy": 84.21760391198045, |
| "eval_test_single_action_correct": 6889, |
| "eval_test_single_loss": 0.026479482650756836, |
| "eval_test_single_overall_test_accuracy": 30.812493718643402, |
| "eval_test_single_runtime": 322.2695, |
| "eval_test_single_samples_per_second": 25.382, |
| "eval_test_single_steps_per_second": 0.099, |
| "eval_test_single_text_acc": 83.80195599022005, |
| "eval_test_single_text_correct": 6855, |
| "eval_test_single_type_acc": 92.61613691931541, |
| "eval_test_single_type_correct": 7576, |
| "step": 24114 |
| }, |
| { |
| "epoch": 3.048021895994028, |
| "grad_norm": 0.028695425018668175, |
| "learning_rate": 4.144453403985716e-05, |
| "loss": 0.0325, |
| "step": 24500 |
| }, |
| { |
| "epoch": 3.1102264244837023, |
| "grad_norm": 0.02727818675339222, |
| "learning_rate": 4.115654878470223e-05, |
| "loss": 0.0322, |
| "step": 25000 |
| }, |
| { |
| "epoch": 3.1724309529733765, |
| "grad_norm": 0.03355926647782326, |
| "learning_rate": 4.086856352954729e-05, |
| "loss": 0.0326, |
| "step": 25500 |
| }, |
| { |
| "epoch": 3.2346354814630507, |
| "grad_norm": 0.029240937903523445, |
| "learning_rate": 4.0580578274392355e-05, |
| "loss": 0.0325, |
| "step": 26000 |
| }, |
| { |
| "epoch": 3.2968400099527244, |
| "grad_norm": 0.031840600073337555, |
| "learning_rate": 4.029259301923742e-05, |
| "loss": 0.0322, |
| "step": 26500 |
| }, |
| { |
| "epoch": 3.3590445384423986, |
| "grad_norm": 0.030460665002465248, |
| "learning_rate": 4.000460776408248e-05, |
| "loss": 0.0322, |
| "step": 27000 |
| }, |
| { |
| "epoch": 3.4212490669320728, |
| "grad_norm": 0.028305282816290855, |
| "learning_rate": 3.971662250892754e-05, |
| "loss": 0.0323, |
| "step": 27500 |
| }, |
| { |
| "epoch": 3.4834535954217465, |
| "grad_norm": 0.03181800991296768, |
| "learning_rate": 3.942863725377261e-05, |
| "loss": 0.0323, |
| "step": 28000 |
| }, |
| { |
| "epoch": 3.5456581239114207, |
| "grad_norm": 0.026895670220255852, |
| "learning_rate": 3.9140651998617674e-05, |
| "loss": 0.0322, |
| "step": 28500 |
| }, |
| { |
| "epoch": 3.607862652401095, |
| "grad_norm": 0.025613972917199135, |
| "learning_rate": 3.8852666743462734e-05, |
| "loss": 0.0323, |
| "step": 29000 |
| }, |
| { |
| "epoch": 3.670067180890769, |
| "grad_norm": 0.03223800286650658, |
| "learning_rate": 3.85646814883078e-05, |
| "loss": 0.0323, |
| "step": 29500 |
| }, |
| { |
| "epoch": 3.732271709380443, |
| "grad_norm": 0.028018802404403687, |
| "learning_rate": 3.827669623315287e-05, |
| "loss": 0.0322, |
| "step": 30000 |
| }, |
| { |
| "epoch": 3.794476237870117, |
| "grad_norm": 0.029745567589998245, |
| "learning_rate": 3.798871097799793e-05, |
| "loss": 0.0321, |
| "step": 30500 |
| }, |
| { |
| "epoch": 3.856680766359791, |
| "grad_norm": 0.02972617745399475, |
| "learning_rate": 3.7700725722842994e-05, |
| "loss": 0.0321, |
| "step": 31000 |
| }, |
| { |
| "epoch": 3.9188852948494652, |
| "grad_norm": 0.027803702279925346, |
| "learning_rate": 3.7412740467688054e-05, |
| "loss": 0.0323, |
| "step": 31500 |
| }, |
| { |
| "epoch": 3.981089823339139, |
| "grad_norm": 0.029762666672468185, |
| "learning_rate": 3.712475521253312e-05, |
| "loss": 0.0324, |
| "step": 32000 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_val_general_accuracy": 69.29295229420471, |
| "eval_val_general_action_correct": 6086, |
| "eval_val_general_loss": 0.033421870321035385, |
| "eval_val_general_runtime": 353.2757, |
| "eval_val_general_samples_per_second": 24.862, |
| "eval_val_general_steps_per_second": 0.099, |
| "eval_val_general_text_acc": 68.39348741887737, |
| "eval_val_general_text_correct": 6007, |
| "eval_val_general_type_acc": 87.12285096208585, |
| "eval_val_general_type_correct": 7652, |
| "step": 32152 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_test_general_accuracy": 71.30562790170988, |
| "eval_test_general_action_correct": 6297, |
| "eval_test_general_loss": 0.030959919095039368, |
| "eval_test_general_runtime": 353.0533, |
| "eval_test_general_samples_per_second": 25.013, |
| "eval_test_general_steps_per_second": 0.099, |
| "eval_test_general_text_acc": 70.66017438568679, |
| "eval_test_general_text_correct": 6240, |
| "eval_test_general_type_acc": 87.81564941682709, |
| "eval_test_general_type_correct": 7755, |
| "step": 32152 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_val_single_accuracy": 85.67422579858571, |
| "eval_val_single_action_correct": 7027, |
| "eval_val_single_loss": 0.02561488375067711, |
| "eval_val_single_overall_val_accuracy": 30.993435618558085, |
| "eval_val_single_runtime": 337.1825, |
| "eval_val_single_samples_per_second": 24.325, |
| "eval_val_single_steps_per_second": 0.098, |
| "eval_val_single_text_acc": 85.2962692026335, |
| "eval_val_single_text_correct": 6996, |
| "eval_val_single_type_acc": 93.70885149963424, |
| "eval_val_single_type_correct": 7686, |
| "step": 32152 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_test_single_accuracy": 86.44254278728606, |
| "eval_test_single_action_correct": 7071, |
| "eval_test_single_loss": 0.025772809982299805, |
| "eval_test_single_overall_test_accuracy": 31.54963413779919, |
| "eval_test_single_runtime": 325.0512, |
| "eval_test_single_samples_per_second": 25.165, |
| "eval_test_single_steps_per_second": 0.098, |
| "eval_test_single_text_acc": 86.06356968215158, |
| "eval_test_single_text_correct": 7040, |
| "eval_test_single_type_acc": 93.86308068459658, |
| "eval_test_single_type_correct": 7678, |
| "step": 32152 |
| }, |
| { |
| "epoch": 4.043294351828813, |
| "grad_norm": 0.031451448798179626, |
| "learning_rate": 3.683676995737819e-05, |
| "loss": 0.0315, |
| "step": 32500 |
| }, |
| { |
| "epoch": 4.105498880318487, |
| "grad_norm": 0.030802294611930847, |
| "learning_rate": 3.654878470222325e-05, |
| "loss": 0.0313, |
| "step": 33000 |
| }, |
| { |
| "epoch": 4.167703408808161, |
| "grad_norm": 0.027965443208813667, |
| "learning_rate": 3.6260799447068314e-05, |
| "loss": 0.0314, |
| "step": 33500 |
| }, |
| { |
| "epoch": 4.229907937297836, |
| "grad_norm": 0.02930346131324768, |
| "learning_rate": 3.5972814191913373e-05, |
| "loss": 0.0313, |
| "step": 34000 |
| }, |
| { |
| "epoch": 4.292112465787509, |
| "grad_norm": 0.03407023474574089, |
| "learning_rate": 3.568482893675844e-05, |
| "loss": 0.0314, |
| "step": 34500 |
| }, |
| { |
| "epoch": 4.354316994277183, |
| "grad_norm": 0.030333781614899635, |
| "learning_rate": 3.53968436816035e-05, |
| "loss": 0.0314, |
| "step": 35000 |
| }, |
| { |
| "epoch": 4.416521522766858, |
| "grad_norm": 0.03493532910943031, |
| "learning_rate": 3.510885842644857e-05, |
| "loss": 0.0316, |
| "step": 35500 |
| }, |
| { |
| "epoch": 4.4787260512565314, |
| "grad_norm": 0.03431040793657303, |
| "learning_rate": 3.482087317129363e-05, |
| "loss": 0.0312, |
| "step": 36000 |
| }, |
| { |
| "epoch": 4.540930579746205, |
| "grad_norm": 0.028342559933662415, |
| "learning_rate": 3.453288791613869e-05, |
| "loss": 0.0311, |
| "step": 36500 |
| }, |
| { |
| "epoch": 4.60313510823588, |
| "grad_norm": 0.03341212868690491, |
| "learning_rate": 3.424490266098376e-05, |
| "loss": 0.0313, |
| "step": 37000 |
| }, |
| { |
| "epoch": 4.6653396367255535, |
| "grad_norm": 0.03309859707951546, |
| "learning_rate": 3.3956917405828826e-05, |
| "loss": 0.0315, |
| "step": 37500 |
| }, |
| { |
| "epoch": 4.727544165215228, |
| "grad_norm": 0.0328318327665329, |
| "learning_rate": 3.3668932150673886e-05, |
| "loss": 0.0316, |
| "step": 38000 |
| }, |
| { |
| "epoch": 4.789748693704902, |
| "grad_norm": 0.029687820002436638, |
| "learning_rate": 3.338094689551895e-05, |
| "loss": 0.0315, |
| "step": 38500 |
| }, |
| { |
| "epoch": 4.851953222194576, |
| "grad_norm": 0.037615273147821426, |
| "learning_rate": 3.309296164036402e-05, |
| "loss": 0.0313, |
| "step": 39000 |
| }, |
| { |
| "epoch": 4.91415775068425, |
| "grad_norm": 0.03171864524483681, |
| "learning_rate": 3.280497638520908e-05, |
| "loss": 0.0315, |
| "step": 39500 |
| }, |
| { |
| "epoch": 4.976362279173924, |
| "grad_norm": 0.03344856947660446, |
| "learning_rate": 3.251699113005414e-05, |
| "loss": 0.0314, |
| "step": 40000 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_val_general_accuracy": 70.19241716953205, |
| "eval_val_general_action_correct": 6165, |
| "eval_val_general_loss": 0.033489227294921875, |
| "eval_val_general_runtime": 357.9361, |
| "eval_val_general_samples_per_second": 24.538, |
| "eval_val_general_steps_per_second": 0.098, |
| "eval_val_general_text_acc": 69.28156666287146, |
| "eval_val_general_text_correct": 6085, |
| "eval_val_general_type_acc": 87.15700785608563, |
| "eval_val_general_type_correct": 7655, |
| "step": 40190 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_test_general_accuracy": 72.2228513192164, |
| "eval_test_general_action_correct": 6378, |
| "eval_test_general_loss": 0.031051145866513252, |
| "eval_test_general_runtime": 357.8649, |
| "eval_test_general_samples_per_second": 24.677, |
| "eval_test_general_steps_per_second": 0.098, |
| "eval_test_general_text_acc": 71.56607405729815, |
| "eval_test_general_text_correct": 6320, |
| "eval_test_general_type_acc": 87.96285811346394, |
| "eval_test_general_type_correct": 7768, |
| "step": 40190 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_val_single_accuracy": 87.4420872957815, |
| "eval_val_single_action_correct": 7172, |
| "eval_val_single_loss": 0.02487783692777157, |
| "eval_val_single_overall_val_accuracy": 31.526900893062713, |
| "eval_val_single_runtime": 336.6433, |
| "eval_val_single_samples_per_second": 24.364, |
| "eval_val_single_steps_per_second": 0.098, |
| "eval_val_single_text_acc": 87.0641306998293, |
| "eval_val_single_text_correct": 7141, |
| "eval_val_single_type_acc": 94.26969031943429, |
| "eval_val_single_type_correct": 7732, |
| "step": 40190 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_test_single_accuracy": 88.53300733496332, |
| "eval_test_single_action_correct": 7242, |
| "eval_test_single_loss": 0.02492070198059082, |
| "eval_test_single_overall_test_accuracy": 32.15117173083594, |
| "eval_test_single_runtime": 333.0898, |
| "eval_test_single_samples_per_second": 24.558, |
| "eval_test_single_steps_per_second": 0.096, |
| "eval_test_single_text_acc": 88.17848410757946, |
| "eval_test_single_text_correct": 7213, |
| "eval_test_single_type_acc": 94.63325183374083, |
| "eval_test_single_type_correct": 7741, |
| "step": 40190 |
| }, |
| { |
| "epoch": 5.038566807663598, |
| "grad_norm": 0.03410176560282707, |
| "learning_rate": 3.2229005874899206e-05, |
| "loss": 0.0308, |
| "step": 40500 |
| }, |
| { |
| "epoch": 5.100771336153272, |
| "grad_norm": 0.03393740952014923, |
| "learning_rate": 3.194102061974427e-05, |
| "loss": 0.0303, |
| "step": 41000 |
| }, |
| { |
| "epoch": 5.162975864642946, |
| "grad_norm": 0.03690695762634277, |
| "learning_rate": 3.165303536458933e-05, |
| "loss": 0.0304, |
| "step": 41500 |
| }, |
| { |
| "epoch": 5.22518039313262, |
| "grad_norm": 0.031830355525016785, |
| "learning_rate": 3.13650501094344e-05, |
| "loss": 0.0303, |
| "step": 42000 |
| }, |
| { |
| "epoch": 5.287384921622294, |
| "grad_norm": 0.04112740978598595, |
| "learning_rate": 3.1077064854279466e-05, |
| "loss": 0.0302, |
| "step": 42500 |
| }, |
| { |
| "epoch": 5.349589450111968, |
| "grad_norm": 0.03433283418416977, |
| "learning_rate": 3.0789079599124526e-05, |
| "loss": 0.0304, |
| "step": 43000 |
| }, |
| { |
| "epoch": 5.411793978601642, |
| "grad_norm": 0.032931845635175705, |
| "learning_rate": 3.050109434396959e-05, |
| "loss": 0.0306, |
| "step": 43500 |
| }, |
| { |
| "epoch": 5.473998507091316, |
| "grad_norm": 0.035872191190719604, |
| "learning_rate": 3.0213109088814656e-05, |
| "loss": 0.0306, |
| "step": 44000 |
| }, |
| { |
| "epoch": 5.53620303558099, |
| "grad_norm": 0.03302635997533798, |
| "learning_rate": 2.992512383365972e-05, |
| "loss": 0.0305, |
| "step": 44500 |
| }, |
| { |
| "epoch": 5.598407564070664, |
| "grad_norm": 0.031577784568071365, |
| "learning_rate": 2.9637138578504782e-05, |
| "loss": 0.0306, |
| "step": 45000 |
| }, |
| { |
| "epoch": 5.6606120925603385, |
| "grad_norm": 0.03194281458854675, |
| "learning_rate": 2.934915332334985e-05, |
| "loss": 0.0305, |
| "step": 45500 |
| }, |
| { |
| "epoch": 5.722816621050012, |
| "grad_norm": 0.03472689166665077, |
| "learning_rate": 2.9061168068194912e-05, |
| "loss": 0.0305, |
| "step": 46000 |
| }, |
| { |
| "epoch": 5.785021149539687, |
| "grad_norm": 0.031870368868112564, |
| "learning_rate": 2.8773182813039972e-05, |
| "loss": 0.0306, |
| "step": 46500 |
| }, |
| { |
| "epoch": 5.8472256780293606, |
| "grad_norm": 0.031068740412592888, |
| "learning_rate": 2.8485197557885035e-05, |
| "loss": 0.0305, |
| "step": 47000 |
| }, |
| { |
| "epoch": 5.909430206519034, |
| "grad_norm": 0.03671320155262947, |
| "learning_rate": 2.81972123027301e-05, |
| "loss": 0.0304, |
| "step": 47500 |
| }, |
| { |
| "epoch": 5.971634735008709, |
| "grad_norm": 0.03440012410283089, |
| "learning_rate": 2.7909227047575165e-05, |
| "loss": 0.0305, |
| "step": 48000 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_val_general_accuracy": 69.61174997153591, |
| "eval_val_general_action_correct": 6114, |
| "eval_val_general_loss": 0.03399135172367096, |
| "eval_val_general_runtime": 357.1857, |
| "eval_val_general_samples_per_second": 24.589, |
| "eval_val_general_steps_per_second": 0.098, |
| "eval_val_general_text_acc": 68.70089946487533, |
| "eval_val_general_text_correct": 6034, |
| "eval_val_general_type_acc": 87.15700785608563, |
| "eval_val_general_type_correct": 7655, |
| "step": 48228 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_test_general_accuracy": 71.33959913939532, |
| "eval_test_general_action_correct": 6300, |
| "eval_test_general_loss": 0.03140591084957123, |
| "eval_test_general_runtime": 356.2452, |
| "eval_test_general_samples_per_second": 24.789, |
| "eval_test_general_steps_per_second": 0.098, |
| "eval_test_general_text_acc": 70.67149813158193, |
| "eval_test_general_text_correct": 6241, |
| "eval_test_general_type_acc": 87.44196580228738, |
| "eval_test_general_type_correct": 7722, |
| "step": 48228 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_val_single_accuracy": 89.01487442087296, |
| "eval_val_single_action_correct": 7301, |
| "eval_val_single_loss": 0.024073051288723946, |
| "eval_val_single_overall_val_accuracy": 31.725324878481775, |
| "eval_val_single_runtime": 337.3699, |
| "eval_val_single_samples_per_second": 24.312, |
| "eval_val_single_steps_per_second": 0.098, |
| "eval_val_single_text_acc": 88.61253352840771, |
| "eval_val_single_text_correct": 7268, |
| "eval_val_single_type_acc": 94.78176054620823, |
| "eval_val_single_type_correct": 7774, |
| "step": 48228 |
| }, |
| { |
| "epoch": 6.0, |
| "eval_test_single_accuracy": 89.73105134474328, |
| "eval_test_single_action_correct": 7340, |
| "eval_test_single_loss": 0.02414107322692871, |
| "eval_test_single_overall_test_accuracy": 32.214130096827716, |
| "eval_test_single_runtime": 329.628, |
| "eval_test_single_samples_per_second": 24.816, |
| "eval_test_single_steps_per_second": 0.097, |
| "eval_test_single_text_acc": 89.3398533007335, |
| "eval_test_single_text_correct": 7308, |
| "eval_test_single_type_acc": 94.97555012224939, |
| "eval_test_single_type_correct": 7769, |
| "step": 48228 |
| }, |
| { |
| "epoch": 6.033839263498383, |
| "grad_norm": 0.03462919965386391, |
| "learning_rate": 2.7621241792420228e-05, |
| "loss": 0.0297, |
| "step": 48500 |
| }, |
| { |
| "epoch": 6.096043791988056, |
| "grad_norm": 0.0364200696349144, |
| "learning_rate": 2.7333256537265295e-05, |
| "loss": 0.0293, |
| "step": 49000 |
| }, |
| { |
| "epoch": 6.158248320477731, |
| "grad_norm": 0.04009445756673813, |
| "learning_rate": 2.7045271282110358e-05, |
| "loss": 0.0292, |
| "step": 49500 |
| }, |
| { |
| "epoch": 6.220452848967405, |
| "grad_norm": 0.04062647372484207, |
| "learning_rate": 2.6757286026955418e-05, |
| "loss": 0.0292, |
| "step": 50000 |
| }, |
| { |
| "epoch": 6.282657377457079, |
| "grad_norm": 0.03505484387278557, |
| "learning_rate": 2.6469300771800488e-05, |
| "loss": 0.0294, |
| "step": 50500 |
| }, |
| { |
| "epoch": 6.344861905946753, |
| "grad_norm": 0.03484097495675087, |
| "learning_rate": 2.6181315516645548e-05, |
| "loss": 0.0293, |
| "step": 51000 |
| }, |
| { |
| "epoch": 6.407066434436427, |
| "grad_norm": 0.03451770171523094, |
| "learning_rate": 2.589333026149061e-05, |
| "loss": 0.0293, |
| "step": 51500 |
| }, |
| { |
| "epoch": 6.469270962926101, |
| "grad_norm": 0.04329177364706993, |
| "learning_rate": 2.5605345006335678e-05, |
| "loss": 0.0293, |
| "step": 52000 |
| }, |
| { |
| "epoch": 6.531475491415775, |
| "grad_norm": 0.03902963548898697, |
| "learning_rate": 2.531735975118074e-05, |
| "loss": 0.0293, |
| "step": 52500 |
| }, |
| { |
| "epoch": 6.593680019905449, |
| "grad_norm": 0.0379670187830925, |
| "learning_rate": 2.5029374496025804e-05, |
| "loss": 0.0293, |
| "step": 53000 |
| }, |
| { |
| "epoch": 6.6558845483951234, |
| "grad_norm": 0.0427132174372673, |
| "learning_rate": 2.4741389240870868e-05, |
| "loss": 0.0295, |
| "step": 53500 |
| }, |
| { |
| "epoch": 6.718089076884797, |
| "grad_norm": 0.038951486349105835, |
| "learning_rate": 2.4453403985715934e-05, |
| "loss": 0.0294, |
| "step": 54000 |
| }, |
| { |
| "epoch": 6.780293605374471, |
| "grad_norm": 0.03940974548459053, |
| "learning_rate": 2.4165418730560997e-05, |
| "loss": 0.0294, |
| "step": 54500 |
| }, |
| { |
| "epoch": 6.8424981338641455, |
| "grad_norm": 0.04146928712725639, |
| "learning_rate": 2.387743347540606e-05, |
| "loss": 0.0294, |
| "step": 55000 |
| }, |
| { |
| "epoch": 6.904702662353819, |
| "grad_norm": 0.03553101047873497, |
| "learning_rate": 2.3589448220251124e-05, |
| "loss": 0.0294, |
| "step": 55500 |
| }, |
| { |
| "epoch": 6.966907190843493, |
| "grad_norm": 0.03867221623659134, |
| "learning_rate": 2.3301462965096187e-05, |
| "loss": 0.0293, |
| "step": 56000 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_val_general_accuracy": 69.66867812820222, |
| "eval_val_general_action_correct": 6119, |
| "eval_val_general_loss": 0.034982189536094666, |
| "eval_val_general_runtime": 356.027, |
| "eval_val_general_samples_per_second": 24.669, |
| "eval_val_general_steps_per_second": 0.098, |
| "eval_val_general_text_acc": 68.74644199020837, |
| "eval_val_general_text_correct": 6038, |
| "eval_val_general_type_acc": 86.90652396675395, |
| "eval_val_general_type_correct": 7633, |
| "step": 56266 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_test_general_accuracy": 70.62620314800135, |
| "eval_test_general_action_correct": 6237, |
| "eval_test_general_loss": 0.03225135803222656, |
| "eval_test_general_runtime": 355.1867, |
| "eval_test_general_samples_per_second": 24.863, |
| "eval_test_general_steps_per_second": 0.099, |
| "eval_test_general_text_acc": 69.9920733778734, |
| "eval_test_general_text_correct": 6181, |
| "eval_test_general_type_acc": 87.30608085154569, |
| "eval_test_general_type_correct": 7710, |
| "step": 56266 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_val_single_accuracy": 89.92928554011218, |
| "eval_val_single_action_correct": 7376, |
| "eval_val_single_loss": 0.023164229467511177, |
| "eval_val_single_overall_val_accuracy": 31.919592733662874, |
| "eval_val_single_runtime": 335.5525, |
| "eval_val_single_samples_per_second": 24.443, |
| "eval_val_single_steps_per_second": 0.098, |
| "eval_val_single_text_acc": 89.55132894415996, |
| "eval_val_single_text_correct": 7345, |
| "eval_val_single_type_acc": 95.08656425262132, |
| "eval_val_single_type_correct": 7799, |
| "step": 56266 |
| }, |
| { |
| "epoch": 7.0, |
| "eval_test_single_accuracy": 91.05134474327629, |
| "eval_test_single_action_correct": 7448, |
| "eval_test_single_loss": 0.02320098876953125, |
| "eval_test_single_overall_test_accuracy": 32.335509578255525, |
| "eval_test_single_runtime": 329.0431, |
| "eval_test_single_samples_per_second": 24.86, |
| "eval_test_single_steps_per_second": 0.097, |
| "eval_test_single_text_acc": 90.6601466992665, |
| "eval_test_single_text_correct": 7416, |
| "eval_test_single_type_acc": 95.26894865525672, |
| "eval_test_single_type_correct": 7793, |
| "step": 56266 |
| } |
| ], |
| "logging_steps": 500, |
| "max_steps": 96456, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 12, |
| "save_steps": 500, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
