| { | |
| "best_metric": 0.8799543380737305, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.08983717012914093, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00044918585064570465, | |
| "grad_norm": 1.2916557788848877, | |
| "learning_rate": 1.0100000000000002e-05, | |
| "loss": 1.7692, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00044918585064570465, | |
| "eval_loss": 1.9869955778121948, | |
| "eval_runtime": 205.6388, | |
| "eval_samples_per_second": 145.863, | |
| "eval_steps_per_second": 4.561, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0008983717012914093, | |
| "grad_norm": 1.607176661491394, | |
| "learning_rate": 2.0200000000000003e-05, | |
| "loss": 1.8371, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.001347557551937114, | |
| "grad_norm": 1.799497365951538, | |
| "learning_rate": 3.0299999999999998e-05, | |
| "loss": 1.8926, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0017967434025828186, | |
| "grad_norm": 1.7198957204818726, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 1.9009, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0022459292532285235, | |
| "grad_norm": 1.7114357948303223, | |
| "learning_rate": 5.05e-05, | |
| "loss": 1.8579, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.002695115103874228, | |
| "grad_norm": 1.4471302032470703, | |
| "learning_rate": 6.0599999999999996e-05, | |
| "loss": 1.6855, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0031443009545199328, | |
| "grad_norm": 0.8383511304855347, | |
| "learning_rate": 7.07e-05, | |
| "loss": 1.4356, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.003593486805165637, | |
| "grad_norm": 0.8857292532920837, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 1.3539, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.004042672655811342, | |
| "grad_norm": 0.7878719568252563, | |
| "learning_rate": 9.09e-05, | |
| "loss": 1.2892, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.004491858506457047, | |
| "grad_norm": 0.8838219046592712, | |
| "learning_rate": 0.000101, | |
| "loss": 1.2533, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.004941044357102751, | |
| "grad_norm": 0.8204696774482727, | |
| "learning_rate": 0.00010046842105263158, | |
| "loss": 1.2432, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.005390230207748456, | |
| "grad_norm": 0.6225486397743225, | |
| "learning_rate": 9.993684210526315e-05, | |
| "loss": 1.184, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.00583941605839416, | |
| "grad_norm": 0.45047685503959656, | |
| "learning_rate": 9.940526315789473e-05, | |
| "loss": 1.1476, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0062886019090398655, | |
| "grad_norm": 0.43591365218162537, | |
| "learning_rate": 9.887368421052632e-05, | |
| "loss": 1.1241, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.00673778775968557, | |
| "grad_norm": 0.39855462312698364, | |
| "learning_rate": 9.83421052631579e-05, | |
| "loss": 1.1202, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.007186973610331274, | |
| "grad_norm": 0.32935649156570435, | |
| "learning_rate": 9.781052631578948e-05, | |
| "loss": 1.0963, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.007636159460976979, | |
| "grad_norm": 0.3512268364429474, | |
| "learning_rate": 9.727894736842106e-05, | |
| "loss": 1.0864, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.008085345311622683, | |
| "grad_norm": 0.36069127917289734, | |
| "learning_rate": 9.674736842105263e-05, | |
| "loss": 1.0866, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.00853453116226839, | |
| "grad_norm": 0.3602248728275299, | |
| "learning_rate": 9.621578947368421e-05, | |
| "loss": 1.0199, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.008983717012914094, | |
| "grad_norm": 0.3629809021949768, | |
| "learning_rate": 9.568421052631578e-05, | |
| "loss": 1.0797, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.009432902863559798, | |
| "grad_norm": 0.3103269934654236, | |
| "learning_rate": 9.515263157894737e-05, | |
| "loss": 1.0546, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.009882088714205503, | |
| "grad_norm": 0.30035093426704407, | |
| "learning_rate": 9.462105263157895e-05, | |
| "loss": 1.0643, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.010331274564851207, | |
| "grad_norm": 0.3492095172405243, | |
| "learning_rate": 9.408947368421054e-05, | |
| "loss": 1.0258, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.010780460415496912, | |
| "grad_norm": 0.4093751609325409, | |
| "learning_rate": 9.355789473684211e-05, | |
| "loss": 1.0746, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.011229646266142616, | |
| "grad_norm": 0.5422231554985046, | |
| "learning_rate": 9.302631578947369e-05, | |
| "loss": 0.975, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01167883211678832, | |
| "grad_norm": 0.344453364610672, | |
| "learning_rate": 9.249473684210526e-05, | |
| "loss": 1.0594, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.012128017967434027, | |
| "grad_norm": 0.3349956274032593, | |
| "learning_rate": 9.196315789473685e-05, | |
| "loss": 1.0283, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.012577203818079731, | |
| "grad_norm": 0.30147019028663635, | |
| "learning_rate": 9.143157894736843e-05, | |
| "loss": 1.0116, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.013026389668725435, | |
| "grad_norm": 0.25005072355270386, | |
| "learning_rate": 9.09e-05, | |
| "loss": 0.9857, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01347557551937114, | |
| "grad_norm": 0.262613981962204, | |
| "learning_rate": 9.036842105263158e-05, | |
| "loss": 1.0245, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.013924761370016844, | |
| "grad_norm": 0.35343340039253235, | |
| "learning_rate": 8.983684210526316e-05, | |
| "loss": 0.967, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.014373947220662549, | |
| "grad_norm": 0.24998489022254944, | |
| "learning_rate": 8.930526315789474e-05, | |
| "loss": 1.0017, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.014823133071308253, | |
| "grad_norm": 0.24498069286346436, | |
| "learning_rate": 8.877368421052632e-05, | |
| "loss": 1.0075, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.015272318921953958, | |
| "grad_norm": 0.21549324691295624, | |
| "learning_rate": 8.82421052631579e-05, | |
| "loss": 0.985, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.015721504772599662, | |
| "grad_norm": 0.22830170392990112, | |
| "learning_rate": 8.771052631578948e-05, | |
| "loss": 0.9629, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.016170690623245366, | |
| "grad_norm": 0.2677782475948334, | |
| "learning_rate": 8.717894736842105e-05, | |
| "loss": 1.014, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.01661987647389107, | |
| "grad_norm": 0.30911344289779663, | |
| "learning_rate": 8.664736842105263e-05, | |
| "loss": 0.9741, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.01706906232453678, | |
| "grad_norm": 0.26794004440307617, | |
| "learning_rate": 8.61157894736842e-05, | |
| "loss": 0.972, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.017518248175182483, | |
| "grad_norm": 0.26757925748825073, | |
| "learning_rate": 8.55842105263158e-05, | |
| "loss": 1.0005, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.017967434025828188, | |
| "grad_norm": 0.23241755366325378, | |
| "learning_rate": 8.505263157894737e-05, | |
| "loss": 0.9912, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.018416619876473892, | |
| "grad_norm": 0.23415440320968628, | |
| "learning_rate": 8.452105263157896e-05, | |
| "loss": 0.96, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.018865805727119597, | |
| "grad_norm": 0.2970597445964813, | |
| "learning_rate": 8.398947368421053e-05, | |
| "loss": 1.0017, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0193149915777653, | |
| "grad_norm": 0.3233031630516052, | |
| "learning_rate": 8.345789473684211e-05, | |
| "loss": 0.9603, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.019764177428411005, | |
| "grad_norm": 0.25598403811454773, | |
| "learning_rate": 8.292631578947368e-05, | |
| "loss": 0.9232, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.02021336327905671, | |
| "grad_norm": 0.28324705362319946, | |
| "learning_rate": 8.239473684210526e-05, | |
| "loss": 0.9762, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.020662549129702414, | |
| "grad_norm": 0.2578172981739044, | |
| "learning_rate": 8.186315789473683e-05, | |
| "loss": 0.9924, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.02111173498034812, | |
| "grad_norm": 0.23937109112739563, | |
| "learning_rate": 8.133157894736842e-05, | |
| "loss": 0.9744, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.021560920830993823, | |
| "grad_norm": 0.2387576401233673, | |
| "learning_rate": 8.080000000000001e-05, | |
| "loss": 0.9448, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.022010106681639528, | |
| "grad_norm": 0.3055512309074402, | |
| "learning_rate": 8.026842105263159e-05, | |
| "loss": 0.9624, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.022459292532285232, | |
| "grad_norm": 0.3693564534187317, | |
| "learning_rate": 7.973684210526316e-05, | |
| "loss": 0.8829, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022459292532285232, | |
| "eval_loss": 0.9505019783973694, | |
| "eval_runtime": 206.986, | |
| "eval_samples_per_second": 144.913, | |
| "eval_steps_per_second": 4.532, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.022908478382930936, | |
| "grad_norm": 0.27806970477104187, | |
| "learning_rate": 7.920526315789474e-05, | |
| "loss": 0.9497, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.02335766423357664, | |
| "grad_norm": 0.25498324632644653, | |
| "learning_rate": 7.867368421052631e-05, | |
| "loss": 0.9515, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.023806850084222345, | |
| "grad_norm": 0.2593478560447693, | |
| "learning_rate": 7.814210526315789e-05, | |
| "loss": 0.9594, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.024256035934868053, | |
| "grad_norm": 0.2518002688884735, | |
| "learning_rate": 7.761052631578946e-05, | |
| "loss": 0.9459, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.024705221785513758, | |
| "grad_norm": 0.26930755376815796, | |
| "learning_rate": 7.707894736842105e-05, | |
| "loss": 0.9266, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.025154407636159462, | |
| "grad_norm": 0.3588949143886566, | |
| "learning_rate": 7.654736842105264e-05, | |
| "loss": 0.9222, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.025603593486805167, | |
| "grad_norm": 0.2470785528421402, | |
| "learning_rate": 7.601578947368422e-05, | |
| "loss": 0.943, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.02605277933745087, | |
| "grad_norm": 0.24494914710521698, | |
| "learning_rate": 7.548421052631579e-05, | |
| "loss": 0.9628, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.026501965188096575, | |
| "grad_norm": 0.23059231042861938, | |
| "learning_rate": 7.495263157894737e-05, | |
| "loss": 0.9444, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.02695115103874228, | |
| "grad_norm": 0.2332954704761505, | |
| "learning_rate": 7.442105263157894e-05, | |
| "loss": 0.9433, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.027400336889387984, | |
| "grad_norm": 0.2612130045890808, | |
| "learning_rate": 7.388947368421053e-05, | |
| "loss": 0.9485, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.02784952274003369, | |
| "grad_norm": 0.3084039092063904, | |
| "learning_rate": 7.335789473684211e-05, | |
| "loss": 0.9346, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.028298708590679393, | |
| "grad_norm": 0.23960231244564056, | |
| "learning_rate": 7.282631578947368e-05, | |
| "loss": 0.955, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.028747894441325098, | |
| "grad_norm": 0.2437230795621872, | |
| "learning_rate": 7.229473684210527e-05, | |
| "loss": 0.9372, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.029197080291970802, | |
| "grad_norm": 0.2259586751461029, | |
| "learning_rate": 7.176315789473685e-05, | |
| "loss": 0.9379, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.029646266142616506, | |
| "grad_norm": 0.2495633065700531, | |
| "learning_rate": 7.123157894736842e-05, | |
| "loss": 0.9357, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.03009545199326221, | |
| "grad_norm": 0.27990928292274475, | |
| "learning_rate": 7.07e-05, | |
| "loss": 0.9611, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.030544637843907915, | |
| "grad_norm": 0.28341051936149597, | |
| "learning_rate": 7.016842105263159e-05, | |
| "loss": 0.9106, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.030993823694553623, | |
| "grad_norm": 0.26214614510536194, | |
| "learning_rate": 6.963684210526316e-05, | |
| "loss": 0.9311, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.031443009545199324, | |
| "grad_norm": 0.30308669805526733, | |
| "learning_rate": 6.910526315789474e-05, | |
| "loss": 0.94, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03189219539584503, | |
| "grad_norm": 0.2610565423965454, | |
| "learning_rate": 6.857368421052631e-05, | |
| "loss": 0.9241, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.03234138124649073, | |
| "grad_norm": 0.23850691318511963, | |
| "learning_rate": 6.80421052631579e-05, | |
| "loss": 0.9422, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.03279056709713644, | |
| "grad_norm": 0.28376665711402893, | |
| "learning_rate": 6.751052631578948e-05, | |
| "loss": 0.9415, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.03323975294778214, | |
| "grad_norm": 0.34418389201164246, | |
| "learning_rate": 6.697894736842105e-05, | |
| "loss": 0.9373, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.033688938798427846, | |
| "grad_norm": 0.4302990734577179, | |
| "learning_rate": 6.644736842105264e-05, | |
| "loss": 0.8997, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.03413812464907356, | |
| "grad_norm": 0.2855401933193207, | |
| "learning_rate": 6.591578947368422e-05, | |
| "loss": 0.945, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.03458731049971926, | |
| "grad_norm": 0.3258453607559204, | |
| "learning_rate": 6.538421052631579e-05, | |
| "loss": 0.933, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.035036496350364967, | |
| "grad_norm": 0.3375171422958374, | |
| "learning_rate": 6.485263157894737e-05, | |
| "loss": 0.9364, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.03548568220101067, | |
| "grad_norm": 0.259318083524704, | |
| "learning_rate": 6.432105263157894e-05, | |
| "loss": 0.8854, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.035934868051656375, | |
| "grad_norm": 0.28955981135368347, | |
| "learning_rate": 6.378947368421053e-05, | |
| "loss": 0.9347, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03638405390230208, | |
| "grad_norm": 0.4210306406021118, | |
| "learning_rate": 6.32578947368421e-05, | |
| "loss": 0.8895, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.036833239752947784, | |
| "grad_norm": 0.22932107746601105, | |
| "learning_rate": 6.27263157894737e-05, | |
| "loss": 0.9466, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.03728242560359349, | |
| "grad_norm": 0.22917746007442474, | |
| "learning_rate": 6.219473684210527e-05, | |
| "loss": 0.9374, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.03773161145423919, | |
| "grad_norm": 0.25457313656806946, | |
| "learning_rate": 6.166315789473685e-05, | |
| "loss": 0.9387, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0381807973048849, | |
| "grad_norm": 0.2656180262565613, | |
| "learning_rate": 6.113157894736842e-05, | |
| "loss": 0.9287, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.0386299831555306, | |
| "grad_norm": 0.26856109499931335, | |
| "learning_rate": 6.0599999999999996e-05, | |
| "loss": 0.9036, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.039079169006176306, | |
| "grad_norm": 0.30407968163490295, | |
| "learning_rate": 6.006842105263158e-05, | |
| "loss": 0.8876, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.03952835485682201, | |
| "grad_norm": 0.2707090675830841, | |
| "learning_rate": 5.953684210526315e-05, | |
| "loss": 0.9105, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.039977540707467715, | |
| "grad_norm": 0.26852765679359436, | |
| "learning_rate": 5.900526315789474e-05, | |
| "loss": 0.9667, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.04042672655811342, | |
| "grad_norm": 0.23827865719795227, | |
| "learning_rate": 5.847368421052632e-05, | |
| "loss": 0.9466, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.040875912408759124, | |
| "grad_norm": 0.25658079981803894, | |
| "learning_rate": 5.79421052631579e-05, | |
| "loss": 0.912, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.04132509825940483, | |
| "grad_norm": 0.28569671511650085, | |
| "learning_rate": 5.7410526315789475e-05, | |
| "loss": 0.9243, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.04177428411005053, | |
| "grad_norm": 0.3140634298324585, | |
| "learning_rate": 5.687894736842105e-05, | |
| "loss": 0.9092, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.04222346996069624, | |
| "grad_norm": 0.26119399070739746, | |
| "learning_rate": 5.6347368421052625e-05, | |
| "loss": 0.8836, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.04267265581134194, | |
| "grad_norm": 0.2906387448310852, | |
| "learning_rate": 5.5815789473684214e-05, | |
| "loss": 0.9251, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.043121841661987646, | |
| "grad_norm": 0.2718961834907532, | |
| "learning_rate": 5.5284210526315796e-05, | |
| "loss": 0.9486, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.04357102751263335, | |
| "grad_norm": 0.2514139711856842, | |
| "learning_rate": 5.475263157894737e-05, | |
| "loss": 0.906, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.044020213363279055, | |
| "grad_norm": 0.26873457431793213, | |
| "learning_rate": 5.422105263157895e-05, | |
| "loss": 0.9222, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.04446939921392476, | |
| "grad_norm": 0.3117465376853943, | |
| "learning_rate": 5.368947368421053e-05, | |
| "loss": 0.897, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.044918585064570464, | |
| "grad_norm": 0.4226572513580322, | |
| "learning_rate": 5.3157894736842104e-05, | |
| "loss": 0.847, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.044918585064570464, | |
| "eval_loss": 0.9066126942634583, | |
| "eval_runtime": 207.0827, | |
| "eval_samples_per_second": 144.846, | |
| "eval_steps_per_second": 4.53, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04536777091521617, | |
| "grad_norm": 0.25195518136024475, | |
| "learning_rate": 5.262631578947368e-05, | |
| "loss": 0.9138, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.04581695676586187, | |
| "grad_norm": 0.2436404973268509, | |
| "learning_rate": 5.209473684210527e-05, | |
| "loss": 0.9067, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.04626614261650758, | |
| "grad_norm": 0.27484673261642456, | |
| "learning_rate": 5.1563157894736844e-05, | |
| "loss": 0.9343, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.04671532846715328, | |
| "grad_norm": 0.27864891290664673, | |
| "learning_rate": 5.1031578947368426e-05, | |
| "loss": 0.9083, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.047164514317798986, | |
| "grad_norm": 0.28357595205307007, | |
| "learning_rate": 5.05e-05, | |
| "loss": 0.8889, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.04761370016844469, | |
| "grad_norm": 0.3293287754058838, | |
| "learning_rate": 4.9968421052631576e-05, | |
| "loss": 0.8996, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.0480628860190904, | |
| "grad_norm": 0.25651612877845764, | |
| "learning_rate": 4.943684210526316e-05, | |
| "loss": 0.9268, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.048512071869736106, | |
| "grad_norm": 0.26456665992736816, | |
| "learning_rate": 4.890526315789474e-05, | |
| "loss": 0.9306, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.04896125772038181, | |
| "grad_norm": 0.24413970112800598, | |
| "learning_rate": 4.8373684210526316e-05, | |
| "loss": 0.9318, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.049410443571027515, | |
| "grad_norm": 0.24325355887413025, | |
| "learning_rate": 4.784210526315789e-05, | |
| "loss": 0.9073, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04985962942167322, | |
| "grad_norm": 0.2712045907974243, | |
| "learning_rate": 4.731052631578947e-05, | |
| "loss": 0.8931, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.050308815272318924, | |
| "grad_norm": 0.3078647255897522, | |
| "learning_rate": 4.6778947368421055e-05, | |
| "loss": 0.8614, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.05075800112296463, | |
| "grad_norm": 0.2627927362918854, | |
| "learning_rate": 4.624736842105263e-05, | |
| "loss": 0.8967, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.05120718697361033, | |
| "grad_norm": 0.25500932335853577, | |
| "learning_rate": 4.571578947368421e-05, | |
| "loss": 0.9088, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.05165637282425604, | |
| "grad_norm": 0.275991827249527, | |
| "learning_rate": 4.518421052631579e-05, | |
| "loss": 0.8913, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.05210555867490174, | |
| "grad_norm": 0.2581343352794647, | |
| "learning_rate": 4.465263157894737e-05, | |
| "loss": 0.8865, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.052554744525547446, | |
| "grad_norm": 0.2798708975315094, | |
| "learning_rate": 4.412105263157895e-05, | |
| "loss": 0.9146, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.05300393037619315, | |
| "grad_norm": 0.3520914316177368, | |
| "learning_rate": 4.358947368421053e-05, | |
| "loss": 0.9002, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.053453116226838855, | |
| "grad_norm": 0.3063805401325226, | |
| "learning_rate": 4.30578947368421e-05, | |
| "loss": 0.8889, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.05390230207748456, | |
| "grad_norm": 0.268318772315979, | |
| "learning_rate": 4.2526315789473685e-05, | |
| "loss": 0.9306, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.054351487928130264, | |
| "grad_norm": 0.2569490075111389, | |
| "learning_rate": 4.199473684210527e-05, | |
| "loss": 0.9124, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.05480067377877597, | |
| "grad_norm": 0.24538756906986237, | |
| "learning_rate": 4.146315789473684e-05, | |
| "loss": 0.911, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.05524985962942167, | |
| "grad_norm": 0.2484789937734604, | |
| "learning_rate": 4.093157894736842e-05, | |
| "loss": 0.9066, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.05569904548006738, | |
| "grad_norm": 0.2707575261592865, | |
| "learning_rate": 4.0400000000000006e-05, | |
| "loss": 0.8681, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.05614823133071308, | |
| "grad_norm": 0.35909488797187805, | |
| "learning_rate": 3.986842105263158e-05, | |
| "loss": 0.8218, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.056597417181358786, | |
| "grad_norm": 0.23776692152023315, | |
| "learning_rate": 3.933684210526316e-05, | |
| "loss": 0.9211, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.05704660303200449, | |
| "grad_norm": 0.25755029916763306, | |
| "learning_rate": 3.880526315789473e-05, | |
| "loss": 0.8879, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.057495788882650195, | |
| "grad_norm": 0.2494814097881317, | |
| "learning_rate": 3.827368421052632e-05, | |
| "loss": 0.8906, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.0579449747332959, | |
| "grad_norm": 0.25940045714378357, | |
| "learning_rate": 3.7742105263157896e-05, | |
| "loss": 0.8868, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.058394160583941604, | |
| "grad_norm": 0.2853882908821106, | |
| "learning_rate": 3.721052631578947e-05, | |
| "loss": 0.8775, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.05884334643458731, | |
| "grad_norm": 0.32979917526245117, | |
| "learning_rate": 3.6678947368421054e-05, | |
| "loss": 0.8522, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.05929253228523301, | |
| "grad_norm": 0.255938321352005, | |
| "learning_rate": 3.6147368421052636e-05, | |
| "loss": 0.9059, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.05974171813587872, | |
| "grad_norm": 0.2526702284812927, | |
| "learning_rate": 3.561578947368421e-05, | |
| "loss": 0.9105, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.06019090398652442, | |
| "grad_norm": 0.2506902813911438, | |
| "learning_rate": 3.508421052631579e-05, | |
| "loss": 0.8951, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.060640089837170126, | |
| "grad_norm": 0.2672176957130432, | |
| "learning_rate": 3.455263157894737e-05, | |
| "loss": 0.8967, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.06108927568781583, | |
| "grad_norm": 0.2774716913700104, | |
| "learning_rate": 3.402105263157895e-05, | |
| "loss": 0.8891, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.06153846153846154, | |
| "grad_norm": 0.3068337142467499, | |
| "learning_rate": 3.3489473684210526e-05, | |
| "loss": 0.8719, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.061987647389107246, | |
| "grad_norm": 0.2592477798461914, | |
| "learning_rate": 3.295789473684211e-05, | |
| "loss": 0.9152, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.06243683323975295, | |
| "grad_norm": 0.24649055302143097, | |
| "learning_rate": 3.242631578947368e-05, | |
| "loss": 0.8869, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.06288601909039865, | |
| "grad_norm": 0.24601422250270844, | |
| "learning_rate": 3.1894736842105265e-05, | |
| "loss": 0.9147, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.06333520494104436, | |
| "grad_norm": 0.2531537115573883, | |
| "learning_rate": 3.136315789473685e-05, | |
| "loss": 0.9067, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.06378439079169006, | |
| "grad_norm": 0.27577677369117737, | |
| "learning_rate": 3.083157894736842e-05, | |
| "loss": 0.8926, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.06423357664233577, | |
| "grad_norm": 0.3049788475036621, | |
| "learning_rate": 3.0299999999999998e-05, | |
| "loss": 0.8536, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.06468276249298147, | |
| "grad_norm": 0.26886892318725586, | |
| "learning_rate": 2.9768421052631577e-05, | |
| "loss": 0.8746, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.06513194834362718, | |
| "grad_norm": 0.2674404978752136, | |
| "learning_rate": 2.923684210526316e-05, | |
| "loss": 0.8991, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.06558113419427287, | |
| "grad_norm": 0.2553546726703644, | |
| "learning_rate": 2.8705263157894737e-05, | |
| "loss": 0.8974, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.06603032004491859, | |
| "grad_norm": 0.2580017149448395, | |
| "learning_rate": 2.8173684210526313e-05, | |
| "loss": 0.8854, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.06647950589556428, | |
| "grad_norm": 0.27426302433013916, | |
| "learning_rate": 2.7642105263157898e-05, | |
| "loss": 0.8885, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.06692869174621, | |
| "grad_norm": 0.3027507960796356, | |
| "learning_rate": 2.7110526315789473e-05, | |
| "loss": 0.8973, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.06737787759685569, | |
| "grad_norm": 0.36818036437034607, | |
| "learning_rate": 2.6578947368421052e-05, | |
| "loss": 0.8283, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06737787759685569, | |
| "eval_loss": 0.8881184458732605, | |
| "eval_runtime": 207.2757, | |
| "eval_samples_per_second": 144.711, | |
| "eval_steps_per_second": 4.525, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0678270634475014, | |
| "grad_norm": 0.245889350771904, | |
| "learning_rate": 2.6047368421052634e-05, | |
| "loss": 0.9066, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.06827624929814712, | |
| "grad_norm": 0.24444212019443512, | |
| "learning_rate": 2.5515789473684213e-05, | |
| "loss": 0.9163, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.06872543514879281, | |
| "grad_norm": 0.24982236325740814, | |
| "learning_rate": 2.4984210526315788e-05, | |
| "loss": 0.8824, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.06917462099943852, | |
| "grad_norm": 0.2683698534965515, | |
| "learning_rate": 2.445263157894737e-05, | |
| "loss": 0.8958, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.06962380685008422, | |
| "grad_norm": 0.27906811237335205, | |
| "learning_rate": 2.3921052631578946e-05, | |
| "loss": 0.8996, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.07007299270072993, | |
| "grad_norm": 0.3476739823818207, | |
| "learning_rate": 2.3389473684210528e-05, | |
| "loss": 0.8366, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.07052217855137563, | |
| "grad_norm": 0.26626768708229065, | |
| "learning_rate": 2.2857894736842106e-05, | |
| "loss": 0.9163, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.07097136440202134, | |
| "grad_norm": 0.26524534821510315, | |
| "learning_rate": 2.2326315789473685e-05, | |
| "loss": 0.895, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.07142055025266704, | |
| "grad_norm": 0.26299136877059937, | |
| "learning_rate": 2.1794736842105264e-05, | |
| "loss": 0.8973, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.07186973610331275, | |
| "grad_norm": 0.2615616023540497, | |
| "learning_rate": 2.1263157894736842e-05, | |
| "loss": 0.901, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.07231892195395845, | |
| "grad_norm": 0.2764776945114136, | |
| "learning_rate": 2.073157894736842e-05, | |
| "loss": 0.8718, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.07276810780460416, | |
| "grad_norm": 0.30812153220176697, | |
| "learning_rate": 2.0200000000000003e-05, | |
| "loss": 0.877, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.07321729365524986, | |
| "grad_norm": 0.2510615289211273, | |
| "learning_rate": 1.966842105263158e-05, | |
| "loss": 0.8739, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.07366647950589557, | |
| "grad_norm": 0.24000753462314606, | |
| "learning_rate": 1.913684210526316e-05, | |
| "loss": 0.913, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.07411566535654127, | |
| "grad_norm": 0.24215757846832275, | |
| "learning_rate": 1.8605263157894736e-05, | |
| "loss": 0.9063, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.07456485120718698, | |
| "grad_norm": 0.26448169350624084, | |
| "learning_rate": 1.8073684210526318e-05, | |
| "loss": 0.8805, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.07501403705783267, | |
| "grad_norm": 0.2861745357513428, | |
| "learning_rate": 1.7542105263157897e-05, | |
| "loss": 0.9049, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.07546322290847839, | |
| "grad_norm": 0.30888983607292175, | |
| "learning_rate": 1.7010526315789475e-05, | |
| "loss": 0.867, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.07591240875912408, | |
| "grad_norm": 0.27784985303878784, | |
| "learning_rate": 1.6478947368421054e-05, | |
| "loss": 0.8749, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.0763615946097698, | |
| "grad_norm": 0.2495727688074112, | |
| "learning_rate": 1.5947368421052633e-05, | |
| "loss": 0.8959, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.07681078046041549, | |
| "grad_norm": 0.24338717758655548, | |
| "learning_rate": 1.541578947368421e-05, | |
| "loss": 0.8821, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.0772599663110612, | |
| "grad_norm": 0.25039950013160706, | |
| "learning_rate": 1.4884210526315788e-05, | |
| "loss": 0.8733, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.0777091521617069, | |
| "grad_norm": 0.27508071064949036, | |
| "learning_rate": 1.4352631578947369e-05, | |
| "loss": 0.9056, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.07815833801235261, | |
| "grad_norm": 0.2863399386405945, | |
| "learning_rate": 1.3821052631578949e-05, | |
| "loss": 0.8748, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.07860752386299831, | |
| "grad_norm": 0.3646789491176605, | |
| "learning_rate": 1.3289473684210526e-05, | |
| "loss": 0.8383, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.07905670971364402, | |
| "grad_norm": 0.2623049020767212, | |
| "learning_rate": 1.2757894736842106e-05, | |
| "loss": 0.8969, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.07950589556428972, | |
| "grad_norm": 0.26813623309135437, | |
| "learning_rate": 1.2226315789473685e-05, | |
| "loss": 0.899, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.07995508141493543, | |
| "grad_norm": 0.2505576014518738, | |
| "learning_rate": 1.1694736842105264e-05, | |
| "loss": 0.8629, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.08040426726558113, | |
| "grad_norm": 0.25836649537086487, | |
| "learning_rate": 1.1163157894736842e-05, | |
| "loss": 0.8929, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.08085345311622684, | |
| "grad_norm": 0.27238643169403076, | |
| "learning_rate": 1.0631578947368421e-05, | |
| "loss": 0.8732, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.08130263896687254, | |
| "grad_norm": 0.33109569549560547, | |
| "learning_rate": 1.0100000000000002e-05, | |
| "loss": 0.8555, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.08175182481751825, | |
| "grad_norm": 0.24706503748893738, | |
| "learning_rate": 9.56842105263158e-06, | |
| "loss": 0.8922, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.08220101066816396, | |
| "grad_norm": 0.24443064630031586, | |
| "learning_rate": 9.036842105263159e-06, | |
| "loss": 0.9011, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.08265019651880966, | |
| "grad_norm": 0.24768005311489105, | |
| "learning_rate": 8.505263157894738e-06, | |
| "loss": 0.91, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.08309938236945537, | |
| "grad_norm": 0.2456647902727127, | |
| "learning_rate": 7.973684210526316e-06, | |
| "loss": 0.8728, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.08354856822010107, | |
| "grad_norm": 0.2784600555896759, | |
| "learning_rate": 7.442105263157894e-06, | |
| "loss": 0.8891, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.08399775407074678, | |
| "grad_norm": 0.3673495948314667, | |
| "learning_rate": 6.9105263157894745e-06, | |
| "loss": 0.8639, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.08444693992139247, | |
| "grad_norm": 0.253378301858902, | |
| "learning_rate": 6.378947368421053e-06, | |
| "loss": 0.8923, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.08489612577203819, | |
| "grad_norm": 0.23032832145690918, | |
| "learning_rate": 5.847368421052632e-06, | |
| "loss": 0.9097, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.08534531162268388, | |
| "grad_norm": 0.24500040709972382, | |
| "learning_rate": 5.315789473684211e-06, | |
| "loss": 0.9037, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.0857944974733296, | |
| "grad_norm": 0.2614372670650482, | |
| "learning_rate": 4.78421052631579e-06, | |
| "loss": 0.9011, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.08624368332397529, | |
| "grad_norm": 0.2819826900959015, | |
| "learning_rate": 4.252631578947369e-06, | |
| "loss": 0.8751, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.086692869174621, | |
| "grad_norm": 0.29293060302734375, | |
| "learning_rate": 3.721052631578947e-06, | |
| "loss": 0.8822, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.0871420550252667, | |
| "grad_norm": 0.2739028036594391, | |
| "learning_rate": 3.1894736842105266e-06, | |
| "loss": 0.8568, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.08759124087591241, | |
| "grad_norm": 0.23401287198066711, | |
| "learning_rate": 2.6578947368421053e-06, | |
| "loss": 0.8958, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.08804042672655811, | |
| "grad_norm": 0.23881946504116058, | |
| "learning_rate": 2.1263157894736844e-06, | |
| "loss": 0.8922, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.08848961257720382, | |
| "grad_norm": 0.2490690052509308, | |
| "learning_rate": 1.5947368421052633e-06, | |
| "loss": 0.8816, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.08893879842784952, | |
| "grad_norm": 0.26390817761421204, | |
| "learning_rate": 1.0631578947368422e-06, | |
| "loss": 0.8778, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.08938798427849523, | |
| "grad_norm": 0.27074772119522095, | |
| "learning_rate": 5.315789473684211e-07, | |
| "loss": 0.8248, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.08983717012914093, | |
| "grad_norm": 0.3729373812675476, | |
| "learning_rate": 0.0, | |
| "loss": 0.8088, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08983717012914093, | |
| "eval_loss": 0.8799543380737305, | |
| "eval_runtime": 207.3734, | |
| "eval_samples_per_second": 144.642, | |
| "eval_steps_per_second": 4.523, | |
| "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": 2.1532281667584e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |