| { | |
| "best_metric": 1.9086253643035889, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.06705783738474434, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00033528918692372173, | |
| "grad_norm": 2.9861958026885986, | |
| "learning_rate": 5e-06, | |
| "loss": 2.5227, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00033528918692372173, | |
| "eval_loss": 4.411320209503174, | |
| "eval_runtime": 414.8908, | |
| "eval_samples_per_second": 12.107, | |
| "eval_steps_per_second": 6.055, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0006705783738474435, | |
| "grad_norm": 4.46765661239624, | |
| "learning_rate": 1e-05, | |
| "loss": 3.4584, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001005867560771165, | |
| "grad_norm": 4.501293182373047, | |
| "learning_rate": 1.5e-05, | |
| "loss": 3.2332, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.001341156747694887, | |
| "grad_norm": 5.019760608673096, | |
| "learning_rate": 2e-05, | |
| "loss": 3.4905, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0016764459346186086, | |
| "grad_norm": 3.9887285232543945, | |
| "learning_rate": 2.5e-05, | |
| "loss": 3.4113, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00201173512154233, | |
| "grad_norm": 4.365457534790039, | |
| "learning_rate": 3e-05, | |
| "loss": 3.9976, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.002347024308466052, | |
| "grad_norm": 3.3956708908081055, | |
| "learning_rate": 3.5e-05, | |
| "loss": 2.872, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.002682313495389774, | |
| "grad_norm": 4.005932807922363, | |
| "learning_rate": 4e-05, | |
| "loss": 3.1288, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0030176026823134955, | |
| "grad_norm": 4.477086544036865, | |
| "learning_rate": 4.5e-05, | |
| "loss": 2.5967, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.003352891869237217, | |
| "grad_norm": 4.308046340942383, | |
| "learning_rate": 5e-05, | |
| "loss": 3.07, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0036881810561609388, | |
| "grad_norm": 4.498213291168213, | |
| "learning_rate": 5.500000000000001e-05, | |
| "loss": 2.3732, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00402347024308466, | |
| "grad_norm": 4.328564167022705, | |
| "learning_rate": 6e-05, | |
| "loss": 2.3216, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0043587594300083824, | |
| "grad_norm": 3.040907144546509, | |
| "learning_rate": 6.500000000000001e-05, | |
| "loss": 2.3894, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.004694048616932104, | |
| "grad_norm": 3.7709429264068604, | |
| "learning_rate": 7e-05, | |
| "loss": 2.3643, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.005029337803855826, | |
| "grad_norm": 3.9415462017059326, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 2.0563, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.005364626990779548, | |
| "grad_norm": 3.5004279613494873, | |
| "learning_rate": 8e-05, | |
| "loss": 2.4909, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.005699916177703269, | |
| "grad_norm": 3.334517478942871, | |
| "learning_rate": 8.5e-05, | |
| "loss": 2.0743, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.006035205364626991, | |
| "grad_norm": 6.377008438110352, | |
| "learning_rate": 9e-05, | |
| "loss": 2.6997, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.006370494551550712, | |
| "grad_norm": 2.8685460090637207, | |
| "learning_rate": 9.5e-05, | |
| "loss": 2.2964, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.006705783738474434, | |
| "grad_norm": 3.3562538623809814, | |
| "learning_rate": 0.0001, | |
| "loss": 2.421, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.007041072925398156, | |
| "grad_norm": 3.677227735519409, | |
| "learning_rate": 9.999238475781957e-05, | |
| "loss": 2.5033, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0073763621123218775, | |
| "grad_norm": 2.6636979579925537, | |
| "learning_rate": 9.99695413509548e-05, | |
| "loss": 2.1439, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0077116512992456, | |
| "grad_norm": 3.073500633239746, | |
| "learning_rate": 9.99314767377287e-05, | |
| "loss": 2.3846, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.00804694048616932, | |
| "grad_norm": 2.4505372047424316, | |
| "learning_rate": 9.987820251299122e-05, | |
| "loss": 2.1418, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.008382229673093043, | |
| "grad_norm": 2.179154634475708, | |
| "learning_rate": 9.980973490458728e-05, | |
| "loss": 2.1383, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.008717518860016765, | |
| "grad_norm": 2.208550453186035, | |
| "learning_rate": 9.972609476841367e-05, | |
| "loss": 1.8978, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.009052808046940487, | |
| "grad_norm": 2.594154119491577, | |
| "learning_rate": 9.962730758206611e-05, | |
| "loss": 2.1316, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.009388097233864207, | |
| "grad_norm": 2.4568934440612793, | |
| "learning_rate": 9.951340343707852e-05, | |
| "loss": 2.1858, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.00972338642078793, | |
| "grad_norm": 2.208120107650757, | |
| "learning_rate": 9.938441702975689e-05, | |
| "loss": 2.1615, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.010058675607711651, | |
| "grad_norm": 2.639620542526245, | |
| "learning_rate": 9.924038765061042e-05, | |
| "loss": 2.0046, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.010393964794635373, | |
| "grad_norm": 2.475559711456299, | |
| "learning_rate": 9.908135917238321e-05, | |
| "loss": 2.1072, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.010729253981559096, | |
| "grad_norm": 2.317560911178589, | |
| "learning_rate": 9.890738003669029e-05, | |
| "loss": 2.0917, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.011064543168482816, | |
| "grad_norm": 2.204341173171997, | |
| "learning_rate": 9.871850323926177e-05, | |
| "loss": 2.2847, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.011399832355406538, | |
| "grad_norm": 2.481131076812744, | |
| "learning_rate": 9.851478631379982e-05, | |
| "loss": 2.3459, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01173512154233026, | |
| "grad_norm": 2.8748056888580322, | |
| "learning_rate": 9.829629131445342e-05, | |
| "loss": 2.1818, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.012070410729253982, | |
| "grad_norm": 2.2782208919525146, | |
| "learning_rate": 9.806308479691595e-05, | |
| "loss": 2.4348, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.012405699916177704, | |
| "grad_norm": 2.5899646282196045, | |
| "learning_rate": 9.781523779815179e-05, | |
| "loss": 2.2178, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.012740989103101424, | |
| "grad_norm": 3.060225486755371, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 2.2754, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.013076278290025146, | |
| "grad_norm": 1.975296139717102, | |
| "learning_rate": 9.727592877996585e-05, | |
| "loss": 1.9798, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.013411567476948869, | |
| "grad_norm": 2.1507437229156494, | |
| "learning_rate": 9.698463103929542e-05, | |
| "loss": 2.2097, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.01374685666387259, | |
| "grad_norm": 2.612276554107666, | |
| "learning_rate": 9.667902132486009e-05, | |
| "loss": 2.0923, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.014082145850796313, | |
| "grad_norm": 2.005558490753174, | |
| "learning_rate": 9.635919272833938e-05, | |
| "loss": 1.9008, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.014417435037720033, | |
| "grad_norm": 2.4271011352539062, | |
| "learning_rate": 9.602524267262203e-05, | |
| "loss": 2.287, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.014752724224643755, | |
| "grad_norm": 2.588599681854248, | |
| "learning_rate": 9.567727288213005e-05, | |
| "loss": 2.5046, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.015088013411567477, | |
| "grad_norm": 2.2825281620025635, | |
| "learning_rate": 9.53153893518325e-05, | |
| "loss": 2.2567, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0154233025984912, | |
| "grad_norm": 2.020557403564453, | |
| "learning_rate": 9.493970231495835e-05, | |
| "loss": 2.2521, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.01575859178541492, | |
| "grad_norm": 1.8200678825378418, | |
| "learning_rate": 9.45503262094184e-05, | |
| "loss": 2.0787, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.01609388097233864, | |
| "grad_norm": 1.7692738771438599, | |
| "learning_rate": 9.414737964294636e-05, | |
| "loss": 1.9956, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.016429170159262364, | |
| "grad_norm": 2.9966800212860107, | |
| "learning_rate": 9.373098535696979e-05, | |
| "loss": 2.0788, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.016764459346186086, | |
| "grad_norm": 2.042726755142212, | |
| "learning_rate": 9.330127018922194e-05, | |
| "loss": 2.1631, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.016764459346186086, | |
| "eval_loss": 2.065981149673462, | |
| "eval_runtime": 417.1736, | |
| "eval_samples_per_second": 12.041, | |
| "eval_steps_per_second": 6.021, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.017099748533109808, | |
| "grad_norm": 2.6870553493499756, | |
| "learning_rate": 9.285836503510562e-05, | |
| "loss": 1.3603, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.01743503772003353, | |
| "grad_norm": 2.5689351558685303, | |
| "learning_rate": 9.24024048078213e-05, | |
| "loss": 1.9551, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.017770326906957252, | |
| "grad_norm": 1.6339844465255737, | |
| "learning_rate": 9.193352839727121e-05, | |
| "loss": 1.6703, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.018105616093880974, | |
| "grad_norm": 1.4754221439361572, | |
| "learning_rate": 9.145187862775209e-05, | |
| "loss": 1.6919, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.018440905280804692, | |
| "grad_norm": 1.7828068733215332, | |
| "learning_rate": 9.09576022144496e-05, | |
| "loss": 1.6638, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.018776194467728415, | |
| "grad_norm": 1.8266091346740723, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 2.0036, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.019111483654652137, | |
| "grad_norm": 2.0084939002990723, | |
| "learning_rate": 8.993177550236464e-05, | |
| "loss": 1.9848, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.01944677284157586, | |
| "grad_norm": 1.6527378559112549, | |
| "learning_rate": 8.940053768033609e-05, | |
| "loss": 1.6421, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.01978206202849958, | |
| "grad_norm": 1.4973303079605103, | |
| "learning_rate": 8.885729807284856e-05, | |
| "loss": 1.6786, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.020117351215423303, | |
| "grad_norm": 2.012096643447876, | |
| "learning_rate": 8.83022221559489e-05, | |
| "loss": 2.1586, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.020452640402347025, | |
| "grad_norm": 1.5897817611694336, | |
| "learning_rate": 8.773547901113862e-05, | |
| "loss": 1.6621, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.020787929589270747, | |
| "grad_norm": 1.691221833229065, | |
| "learning_rate": 8.715724127386972e-05, | |
| "loss": 1.7684, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.02112321877619447, | |
| "grad_norm": 1.5985442399978638, | |
| "learning_rate": 8.656768508095853e-05, | |
| "loss": 1.7924, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.02145850796311819, | |
| "grad_norm": 1.5729451179504395, | |
| "learning_rate": 8.596699001693255e-05, | |
| "loss": 1.5607, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.02179379715004191, | |
| "grad_norm": 1.6570827960968018, | |
| "learning_rate": 8.535533905932738e-05, | |
| "loss": 1.851, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.02212908633696563, | |
| "grad_norm": 2.0177438259124756, | |
| "learning_rate": 8.473291852294987e-05, | |
| "loss": 2.0311, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.022464375523889354, | |
| "grad_norm": 1.829530954360962, | |
| "learning_rate": 8.409991800312493e-05, | |
| "loss": 1.9164, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.022799664710813076, | |
| "grad_norm": 1.6682900190353394, | |
| "learning_rate": 8.345653031794292e-05, | |
| "loss": 1.8306, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.023134953897736798, | |
| "grad_norm": 2.0108556747436523, | |
| "learning_rate": 8.280295144952536e-05, | |
| "loss": 2.0025, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.02347024308466052, | |
| "grad_norm": 1.7402783632278442, | |
| "learning_rate": 8.213938048432697e-05, | |
| "loss": 1.5311, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.023805532271584242, | |
| "grad_norm": 1.6198512315750122, | |
| "learning_rate": 8.146601955249188e-05, | |
| "loss": 1.9323, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.024140821458507964, | |
| "grad_norm": 2.033475399017334, | |
| "learning_rate": 8.07830737662829e-05, | |
| "loss": 2.0555, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.024476110645431686, | |
| "grad_norm": 1.9241336584091187, | |
| "learning_rate": 8.009075115760243e-05, | |
| "loss": 2.259, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.024811399832355408, | |
| "grad_norm": 1.6944104433059692, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.7978, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.025146689019279127, | |
| "grad_norm": 1.7152888774871826, | |
| "learning_rate": 7.86788218175523e-05, | |
| "loss": 1.9311, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02548197820620285, | |
| "grad_norm": 1.7262462377548218, | |
| "learning_rate": 7.795964517353735e-05, | |
| "loss": 1.9446, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.02581726739312657, | |
| "grad_norm": 1.6206432580947876, | |
| "learning_rate": 7.723195175075136e-05, | |
| "loss": 1.8652, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.026152556580050293, | |
| "grad_norm": 1.870849370956421, | |
| "learning_rate": 7.649596321166024e-05, | |
| "loss": 1.9929, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.026487845766974015, | |
| "grad_norm": 1.8466707468032837, | |
| "learning_rate": 7.575190374550272e-05, | |
| "loss": 2.0701, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.026823134953897737, | |
| "grad_norm": 1.694617509841919, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 2.2557, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.02715842414082146, | |
| "grad_norm": 1.8418231010437012, | |
| "learning_rate": 7.424048101231686e-05, | |
| "loss": 2.0265, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.02749371332774518, | |
| "grad_norm": 1.8243086338043213, | |
| "learning_rate": 7.347357813929454e-05, | |
| "loss": 2.1743, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.027829002514668903, | |
| "grad_norm": 1.9101988077163696, | |
| "learning_rate": 7.269952498697734e-05, | |
| "loss": 2.0917, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.028164291701592625, | |
| "grad_norm": 1.7680463790893555, | |
| "learning_rate": 7.191855733945387e-05, | |
| "loss": 2.1972, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.028499580888516344, | |
| "grad_norm": 1.8039413690567017, | |
| "learning_rate": 7.113091308703498e-05, | |
| "loss": 2.2545, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.028834870075440066, | |
| "grad_norm": 2.0174150466918945, | |
| "learning_rate": 7.033683215379002e-05, | |
| "loss": 2.0376, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.029170159262363788, | |
| "grad_norm": 1.7926223278045654, | |
| "learning_rate": 6.953655642446368e-05, | |
| "loss": 1.8398, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.02950544844928751, | |
| "grad_norm": 2.14631986618042, | |
| "learning_rate": 6.873032967079561e-05, | |
| "loss": 2.0343, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.029840737636211232, | |
| "grad_norm": 1.9503663778305054, | |
| "learning_rate": 6.7918397477265e-05, | |
| "loss": 2.1005, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.030176026823134954, | |
| "grad_norm": 1.8664933443069458, | |
| "learning_rate": 6.710100716628344e-05, | |
| "loss": 1.9907, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.030511316010058676, | |
| "grad_norm": 2.2707552909851074, | |
| "learning_rate": 6.627840772285784e-05, | |
| "loss": 2.1686, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0308466051969824, | |
| "grad_norm": 1.9594255685806274, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 2.1432, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.03118189438390612, | |
| "grad_norm": 2.176732301712036, | |
| "learning_rate": 6.461858523613684e-05, | |
| "loss": 2.3652, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.03151718357082984, | |
| "grad_norm": 1.833224892616272, | |
| "learning_rate": 6.378186779084995e-05, | |
| "loss": 2.0414, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.03185247275775356, | |
| "grad_norm": 1.9593424797058105, | |
| "learning_rate": 6.294095225512603e-05, | |
| "loss": 2.3346, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.03218776194467728, | |
| "grad_norm": 1.7518645524978638, | |
| "learning_rate": 6.209609477998338e-05, | |
| "loss": 2.0419, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.032523051131601005, | |
| "grad_norm": 1.6147215366363525, | |
| "learning_rate": 6.124755271719325e-05, | |
| "loss": 1.9152, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.03285834031852473, | |
| "grad_norm": 1.6326552629470825, | |
| "learning_rate": 6.0395584540887963e-05, | |
| "loss": 2.1208, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.03319362950544845, | |
| "grad_norm": 1.6675904989242554, | |
| "learning_rate": 5.9540449768827246e-05, | |
| "loss": 2.0033, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.03352891869237217, | |
| "grad_norm": 2.3749098777770996, | |
| "learning_rate": 5.868240888334653e-05, | |
| "loss": 2.2195, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03352891869237217, | |
| "eval_loss": 1.9729357957839966, | |
| "eval_runtime": 417.3141, | |
| "eval_samples_per_second": 12.036, | |
| "eval_steps_per_second": 6.019, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.03386420787929589, | |
| "grad_norm": 2.3393783569335938, | |
| "learning_rate": 5.782172325201155e-05, | |
| "loss": 1.6591, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.034199497066219615, | |
| "grad_norm": 2.494509220123291, | |
| "learning_rate": 5.695865504800327e-05, | |
| "loss": 2.076, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.03453478625314334, | |
| "grad_norm": 1.6689271926879883, | |
| "learning_rate": 5.6093467170257374e-05, | |
| "loss": 1.5192, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.03487007544006706, | |
| "grad_norm": 1.61103093624115, | |
| "learning_rate": 5.522642316338268e-05, | |
| "loss": 1.7262, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.03520536462699078, | |
| "grad_norm": 1.6886314153671265, | |
| "learning_rate": 5.435778713738292e-05, | |
| "loss": 1.5618, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.035540653813914504, | |
| "grad_norm": 1.6932485103607178, | |
| "learning_rate": 5.348782368720626e-05, | |
| "loss": 1.5998, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.035875943000838226, | |
| "grad_norm": 1.794480800628662, | |
| "learning_rate": 5.26167978121472e-05, | |
| "loss": 1.8081, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.03621123218776195, | |
| "grad_norm": 1.8547143936157227, | |
| "learning_rate": 5.174497483512506e-05, | |
| "loss": 1.9437, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.03654652137468566, | |
| "grad_norm": 1.7825583219528198, | |
| "learning_rate": 5.0872620321864185e-05, | |
| "loss": 1.708, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.036881810561609385, | |
| "grad_norm": 1.591256856918335, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6669, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.03721709974853311, | |
| "grad_norm": 1.4155341386795044, | |
| "learning_rate": 4.912737967813583e-05, | |
| "loss": 1.5597, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.03755238893545683, | |
| "grad_norm": 1.4923256635665894, | |
| "learning_rate": 4.825502516487497e-05, | |
| "loss": 1.405, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.03788767812238055, | |
| "grad_norm": 1.5631953477859497, | |
| "learning_rate": 4.738320218785281e-05, | |
| "loss": 1.6186, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.03822296730930427, | |
| "grad_norm": 1.584811806678772, | |
| "learning_rate": 4.6512176312793736e-05, | |
| "loss": 1.9145, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.038558256496227995, | |
| "grad_norm": 1.780116081237793, | |
| "learning_rate": 4.564221286261709e-05, | |
| "loss": 1.8281, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.03889354568315172, | |
| "grad_norm": 1.6140252351760864, | |
| "learning_rate": 4.477357683661734e-05, | |
| "loss": 1.7809, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.03922883487007544, | |
| "grad_norm": 1.4927319288253784, | |
| "learning_rate": 4.390653282974264e-05, | |
| "loss": 1.728, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.03956412405699916, | |
| "grad_norm": 1.2532942295074463, | |
| "learning_rate": 4.3041344951996746e-05, | |
| "loss": 1.5454, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.039899413243922884, | |
| "grad_norm": 1.5712227821350098, | |
| "learning_rate": 4.2178276747988446e-05, | |
| "loss": 1.9015, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.040234702430846606, | |
| "grad_norm": 1.3712291717529297, | |
| "learning_rate": 4.131759111665349e-05, | |
| "loss": 1.6745, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.04056999161777033, | |
| "grad_norm": 1.6665000915527344, | |
| "learning_rate": 4.045955023117276e-05, | |
| "loss": 1.8049, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.04090528080469405, | |
| "grad_norm": 1.5372314453125, | |
| "learning_rate": 3.960441545911204e-05, | |
| "loss": 1.9282, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.04124056999161777, | |
| "grad_norm": 1.3043067455291748, | |
| "learning_rate": 3.875244728280676e-05, | |
| "loss": 1.5475, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.041575859178541494, | |
| "grad_norm": 1.587207555770874, | |
| "learning_rate": 3.790390522001662e-05, | |
| "loss": 1.7552, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.041911148365465216, | |
| "grad_norm": 1.3804585933685303, | |
| "learning_rate": 3.705904774487396e-05, | |
| "loss": 1.835, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.04224643755238894, | |
| "grad_norm": 1.54267156124115, | |
| "learning_rate": 3.6218132209150045e-05, | |
| "loss": 1.8165, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.04258172673931266, | |
| "grad_norm": 1.9306060075759888, | |
| "learning_rate": 3.5381414763863166e-05, | |
| "loss": 2.1788, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.04291701592623638, | |
| "grad_norm": 1.8999401330947876, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 2.0431, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.0432523051131601, | |
| "grad_norm": 1.869875431060791, | |
| "learning_rate": 3.372159227714218e-05, | |
| "loss": 1.9761, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.04358759430008382, | |
| "grad_norm": 1.7410246133804321, | |
| "learning_rate": 3.289899283371657e-05, | |
| "loss": 1.8724, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.04392288348700754, | |
| "grad_norm": 1.7089636325836182, | |
| "learning_rate": 3.2081602522734986e-05, | |
| "loss": 1.8295, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.04425817267393126, | |
| "grad_norm": 1.5494524240493774, | |
| "learning_rate": 3.12696703292044e-05, | |
| "loss": 1.7068, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.044593461860854985, | |
| "grad_norm": 1.7785358428955078, | |
| "learning_rate": 3.046344357553632e-05, | |
| "loss": 1.7901, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.04492875104777871, | |
| "grad_norm": 2.0692455768585205, | |
| "learning_rate": 2.9663167846209998e-05, | |
| "loss": 2.2282, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.04526404023470243, | |
| "grad_norm": 1.858677625656128, | |
| "learning_rate": 2.886908691296504e-05, | |
| "loss": 1.9704, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.04559932942162615, | |
| "grad_norm": 1.7712692022323608, | |
| "learning_rate": 2.8081442660546125e-05, | |
| "loss": 1.9881, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.045934618608549874, | |
| "grad_norm": 1.7921137809753418, | |
| "learning_rate": 2.7300475013022663e-05, | |
| "loss": 2.2442, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.046269907795473596, | |
| "grad_norm": 1.7283982038497925, | |
| "learning_rate": 2.6526421860705473e-05, | |
| "loss": 2.0147, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.04660519698239732, | |
| "grad_norm": 1.8532720804214478, | |
| "learning_rate": 2.575951898768315e-05, | |
| "loss": 2.0814, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.04694048616932104, | |
| "grad_norm": 1.800716757774353, | |
| "learning_rate": 2.500000000000001e-05, | |
| "loss": 2.0, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.04727577535624476, | |
| "grad_norm": 1.8511884212493896, | |
| "learning_rate": 2.4248096254497288e-05, | |
| "loss": 1.8164, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.047611064543168484, | |
| "grad_norm": 1.819697618484497, | |
| "learning_rate": 2.350403678833976e-05, | |
| "loss": 2.0316, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.047946353730092206, | |
| "grad_norm": 1.7285267114639282, | |
| "learning_rate": 2.2768048249248648e-05, | |
| "loss": 2.0531, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.04828164291701593, | |
| "grad_norm": 1.8958516120910645, | |
| "learning_rate": 2.2040354826462668e-05, | |
| "loss": 2.2323, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.04861693210393965, | |
| "grad_norm": 1.7046324014663696, | |
| "learning_rate": 2.132117818244771e-05, | |
| "loss": 2.0309, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.04895222129086337, | |
| "grad_norm": 1.733049988746643, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 2.0362, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.049287510477787094, | |
| "grad_norm": 2.089198350906372, | |
| "learning_rate": 1.9909248842397584e-05, | |
| "loss": 2.0955, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.049622799664710816, | |
| "grad_norm": 1.4590038061141968, | |
| "learning_rate": 1.9216926233717085e-05, | |
| "loss": 1.8357, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.04995808885163453, | |
| "grad_norm": 1.496126413345337, | |
| "learning_rate": 1.8533980447508137e-05, | |
| "loss": 1.9012, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.050293378038558254, | |
| "grad_norm": 1.7759606838226318, | |
| "learning_rate": 1.7860619515673033e-05, | |
| "loss": 1.9648, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.050293378038558254, | |
| "eval_loss": 1.918830394744873, | |
| "eval_runtime": 417.5811, | |
| "eval_samples_per_second": 12.029, | |
| "eval_steps_per_second": 6.016, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.050628667225481976, | |
| "grad_norm": 1.7234727144241333, | |
| "learning_rate": 1.7197048550474643e-05, | |
| "loss": 1.6154, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.0509639564124057, | |
| "grad_norm": 1.6809676885604858, | |
| "learning_rate": 1.6543469682057106e-05, | |
| "loss": 1.8106, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.05129924559932942, | |
| "grad_norm": 1.6216368675231934, | |
| "learning_rate": 1.5900081996875083e-05, | |
| "loss": 1.927, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.05163453478625314, | |
| "grad_norm": 1.702203392982483, | |
| "learning_rate": 1.526708147705013e-05, | |
| "loss": 2.0449, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.051969823973176864, | |
| "grad_norm": 1.4775599241256714, | |
| "learning_rate": 1.4644660940672627e-05, | |
| "loss": 1.5648, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.052305113160100586, | |
| "grad_norm": 1.254103660583496, | |
| "learning_rate": 1.4033009983067452e-05, | |
| "loss": 1.2817, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.05264040234702431, | |
| "grad_norm": 1.5809673070907593, | |
| "learning_rate": 1.3432314919041478e-05, | |
| "loss": 1.8523, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.05297569153394803, | |
| "grad_norm": 1.5104883909225464, | |
| "learning_rate": 1.2842758726130283e-05, | |
| "loss": 1.6423, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.05331098072087175, | |
| "grad_norm": 1.4574472904205322, | |
| "learning_rate": 1.22645209888614e-05, | |
| "loss": 1.7987, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.053646269907795474, | |
| "grad_norm": 1.3530458211898804, | |
| "learning_rate": 1.1697777844051105e-05, | |
| "loss": 1.6336, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.053981559094719196, | |
| "grad_norm": 1.515069842338562, | |
| "learning_rate": 1.1142701927151456e-05, | |
| "loss": 1.6613, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.05431684828164292, | |
| "grad_norm": 1.5327993631362915, | |
| "learning_rate": 1.0599462319663905e-05, | |
| "loss": 1.8434, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.05465213746856664, | |
| "grad_norm": 1.9559135437011719, | |
| "learning_rate": 1.006822449763537e-05, | |
| "loss": 1.7568, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.05498742665549036, | |
| "grad_norm": 1.5940124988555908, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.7559, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.055322715842414084, | |
| "grad_norm": 1.5147747993469238, | |
| "learning_rate": 9.042397785550405e-06, | |
| "loss": 1.804, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.055658005029337806, | |
| "grad_norm": 1.5718660354614258, | |
| "learning_rate": 8.548121372247918e-06, | |
| "loss": 1.8784, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.05599329421626153, | |
| "grad_norm": 1.716925859451294, | |
| "learning_rate": 8.066471602728803e-06, | |
| "loss": 1.9153, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.05632858340318525, | |
| "grad_norm": 1.8211606740951538, | |
| "learning_rate": 7.597595192178702e-06, | |
| "loss": 1.8741, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.056663872590108966, | |
| "grad_norm": 1.437688946723938, | |
| "learning_rate": 7.1416349648943894e-06, | |
| "loss": 1.6253, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.05699916177703269, | |
| "grad_norm": 1.4065502882003784, | |
| "learning_rate": 6.698729810778065e-06, | |
| "loss": 1.6419, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.05733445096395641, | |
| "grad_norm": 1.6610018014907837, | |
| "learning_rate": 6.269014643030213e-06, | |
| "loss": 1.853, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.05766974015088013, | |
| "grad_norm": 1.736626386642456, | |
| "learning_rate": 5.852620357053651e-06, | |
| "loss": 1.9798, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.058005029337803854, | |
| "grad_norm": 1.7204649448394775, | |
| "learning_rate": 5.449673790581611e-06, | |
| "loss": 2.0083, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.058340318524727576, | |
| "grad_norm": 1.7506190538406372, | |
| "learning_rate": 5.060297685041659e-06, | |
| "loss": 2.0595, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.0586756077116513, | |
| "grad_norm": 1.5946072340011597, | |
| "learning_rate": 4.684610648167503e-06, | |
| "loss": 2.0093, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.05901089689857502, | |
| "grad_norm": 1.824426531791687, | |
| "learning_rate": 4.322727117869951e-06, | |
| "loss": 1.8961, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.05934618608549874, | |
| "grad_norm": 1.5746324062347412, | |
| "learning_rate": 3.974757327377981e-06, | |
| "loss": 1.7856, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.059681475272422464, | |
| "grad_norm": 1.908447027206421, | |
| "learning_rate": 3.6408072716606346e-06, | |
| "loss": 1.9998, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.060016764459346186, | |
| "grad_norm": 1.581841230392456, | |
| "learning_rate": 3.3209786751399187e-06, | |
| "loss": 1.8297, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.06035205364626991, | |
| "grad_norm": 2.107459545135498, | |
| "learning_rate": 3.0153689607045845e-06, | |
| "loss": 2.0836, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.06068734283319363, | |
| "grad_norm": 1.9093464612960815, | |
| "learning_rate": 2.724071220034158e-06, | |
| "loss": 2.0951, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.06102263202011735, | |
| "grad_norm": 1.8515557050704956, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.7775, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.061357921207041075, | |
| "grad_norm": 1.6800440549850464, | |
| "learning_rate": 2.1847622018482283e-06, | |
| "loss": 2.0027, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.0616932103939648, | |
| "grad_norm": 2.0286970138549805, | |
| "learning_rate": 1.9369152030840556e-06, | |
| "loss": 2.1018, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.06202849958088852, | |
| "grad_norm": 1.6535078287124634, | |
| "learning_rate": 1.70370868554659e-06, | |
| "loss": 2.0856, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.06236378876781224, | |
| "grad_norm": 1.7752177715301514, | |
| "learning_rate": 1.4852136862001764e-06, | |
| "loss": 2.1588, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.06269907795473596, | |
| "grad_norm": 1.6971702575683594, | |
| "learning_rate": 1.2814967607382432e-06, | |
| "loss": 2.2293, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.06303436714165968, | |
| "grad_norm": 1.8538073301315308, | |
| "learning_rate": 1.0926199633097157e-06, | |
| "loss": 2.1466, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.0633696563285834, | |
| "grad_norm": 1.5561541318893433, | |
| "learning_rate": 9.186408276168013e-07, | |
| "loss": 1.9356, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.06370494551550712, | |
| "grad_norm": 1.777517557144165, | |
| "learning_rate": 7.596123493895991e-07, | |
| "loss": 2.1566, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.06404023470243085, | |
| "grad_norm": 1.8275128602981567, | |
| "learning_rate": 6.15582970243117e-07, | |
| "loss": 2.0181, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.06437552388935457, | |
| "grad_norm": 1.6356531381607056, | |
| "learning_rate": 4.865965629214819e-07, | |
| "loss": 1.8176, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.0647108130762783, | |
| "grad_norm": 1.6933003664016724, | |
| "learning_rate": 3.7269241793390085e-07, | |
| "loss": 1.7686, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.06504610226320201, | |
| "grad_norm": 1.7050329446792603, | |
| "learning_rate": 2.7390523158633554e-07, | |
| "loss": 1.9721, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.06538139145012574, | |
| "grad_norm": 1.9019380807876587, | |
| "learning_rate": 1.9026509541272275e-07, | |
| "loss": 2.0865, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.06571668063704945, | |
| "grad_norm": 1.6882052421569824, | |
| "learning_rate": 1.2179748700879012e-07, | |
| "loss": 2.022, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.06605196982397318, | |
| "grad_norm": 2.0564017295837402, | |
| "learning_rate": 6.852326227130834e-08, | |
| "loss": 2.3897, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.0663872590108969, | |
| "grad_norm": 1.8937839269638062, | |
| "learning_rate": 3.04586490452119e-08, | |
| "loss": 1.8696, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.06672254819782061, | |
| "grad_norm": 1.7669646739959717, | |
| "learning_rate": 7.615242180436522e-09, | |
| "loss": 2.0887, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.06705783738474434, | |
| "grad_norm": 1.9383103847503662, | |
| "learning_rate": 0.0, | |
| "loss": 2.0162, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.06705783738474434, | |
| "eval_loss": 1.9086253643035889, | |
| "eval_runtime": 418.0779, | |
| "eval_samples_per_second": 12.015, | |
| "eval_steps_per_second": 6.008, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.017027157491712e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |