| { | |
| "best_metric": 11.5, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-50", | |
| "epoch": 0.0020129432249363405, | |
| "eval_steps": 50, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.0129432249363408e-05, | |
| "grad_norm": 1.6507690816069953e-05, | |
| "learning_rate": 1.004e-05, | |
| "loss": 23.0, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 2.0129432249363408e-05, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 285.9668, | |
| "eval_samples_per_second": 73.148, | |
| "eval_steps_per_second": 18.289, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 4.0258864498726816e-05, | |
| "grad_norm": 1.639904803596437e-05, | |
| "learning_rate": 2.008e-05, | |
| "loss": 23.0, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 6.038829674809022e-05, | |
| "grad_norm": 2.0076164219062775e-05, | |
| "learning_rate": 3.012e-05, | |
| "loss": 23.0, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 8.051772899745363e-05, | |
| "grad_norm": 9.847400178841781e-06, | |
| "learning_rate": 4.016e-05, | |
| "loss": 23.0, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00010064716124681703, | |
| "grad_norm": 9.781237167771906e-06, | |
| "learning_rate": 5.02e-05, | |
| "loss": 23.0, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00012077659349618043, | |
| "grad_norm": 2.2412741600419395e-05, | |
| "learning_rate": 6.024e-05, | |
| "loss": 23.0, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00014090602574554385, | |
| "grad_norm": 1.6903981304494664e-05, | |
| "learning_rate": 7.028e-05, | |
| "loss": 23.0, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.00016103545799490726, | |
| "grad_norm": 2.2363303287420422e-05, | |
| "learning_rate": 8.032e-05, | |
| "loss": 23.0, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.00018116489024427065, | |
| "grad_norm": 1.0626799848978408e-05, | |
| "learning_rate": 9.036000000000001e-05, | |
| "loss": 23.0, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.00020129432249363407, | |
| "grad_norm": 1.4076299521548208e-05, | |
| "learning_rate": 0.0001004, | |
| "loss": 23.0, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.00022142375474299748, | |
| "grad_norm": 2.3633083401364274e-05, | |
| "learning_rate": 9.987157894736842e-05, | |
| "loss": 23.0, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.00024155318699236087, | |
| "grad_norm": 2.4257647964986973e-05, | |
| "learning_rate": 9.934315789473684e-05, | |
| "loss": 23.0, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0002616826192417243, | |
| "grad_norm": 2.6146915843128227e-05, | |
| "learning_rate": 9.881473684210525e-05, | |
| "loss": 23.0, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0002818120514910877, | |
| "grad_norm": 1.611110747035127e-05, | |
| "learning_rate": 9.828631578947369e-05, | |
| "loss": 23.0, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0003019414837404511, | |
| "grad_norm": 1.668692311795894e-05, | |
| "learning_rate": 9.77578947368421e-05, | |
| "loss": 23.0, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.00032207091598981453, | |
| "grad_norm": 2.7145209969603457e-05, | |
| "learning_rate": 9.722947368421052e-05, | |
| "loss": 23.0, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0003422003482391779, | |
| "grad_norm": 1.7148986444226466e-05, | |
| "learning_rate": 9.670105263157895e-05, | |
| "loss": 23.0, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0003623297804885413, | |
| "grad_norm": 1.813442213460803e-05, | |
| "learning_rate": 9.617263157894737e-05, | |
| "loss": 23.0, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.00038245921273790474, | |
| "grad_norm": 3.107384327449836e-05, | |
| "learning_rate": 9.564421052631579e-05, | |
| "loss": 23.0, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.00040258864498726813, | |
| "grad_norm": 2.8771428333129734e-05, | |
| "learning_rate": 9.511578947368421e-05, | |
| "loss": 23.0, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0004227180772366315, | |
| "grad_norm": 1.3413990927801933e-05, | |
| "learning_rate": 9.458736842105264e-05, | |
| "loss": 23.0, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.00044284750948599496, | |
| "grad_norm": 1.6914329535211436e-05, | |
| "learning_rate": 9.405894736842106e-05, | |
| "loss": 23.0, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.00046297694173535835, | |
| "grad_norm": 1.504943247709889e-05, | |
| "learning_rate": 9.353052631578947e-05, | |
| "loss": 23.0, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.00048310637398472174, | |
| "grad_norm": 2.3936727302498184e-05, | |
| "learning_rate": 9.300210526315789e-05, | |
| "loss": 23.0, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0005032358062340851, | |
| "grad_norm": 4.2899515392491594e-05, | |
| "learning_rate": 9.247368421052631e-05, | |
| "loss": 23.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0005233652384834486, | |
| "grad_norm": 1.9880957552231848e-05, | |
| "learning_rate": 9.194526315789473e-05, | |
| "loss": 23.0, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.000543494670732812, | |
| "grad_norm": 2.101472273352556e-05, | |
| "learning_rate": 9.141684210526316e-05, | |
| "loss": 23.0, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0005636241029821754, | |
| "grad_norm": 2.740858144534286e-05, | |
| "learning_rate": 9.088842105263158e-05, | |
| "loss": 23.0, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0005837535352315388, | |
| "grad_norm": 8.637041173642501e-05, | |
| "learning_rate": 9.036000000000001e-05, | |
| "loss": 23.0, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0006038829674809022, | |
| "grad_norm": 4.428947431733832e-05, | |
| "learning_rate": 8.983157894736843e-05, | |
| "loss": 23.0, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0006240123997302656, | |
| "grad_norm": 6.015045801177621e-05, | |
| "learning_rate": 8.930315789473684e-05, | |
| "loss": 23.0, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0006441418319796291, | |
| "grad_norm": 5.055875953985378e-05, | |
| "learning_rate": 8.877473684210526e-05, | |
| "loss": 23.0, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0006642712642289924, | |
| "grad_norm": 3.0454708394245245e-05, | |
| "learning_rate": 8.824631578947368e-05, | |
| "loss": 23.0, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0006844006964783558, | |
| "grad_norm": 5.01158538099844e-05, | |
| "learning_rate": 8.771789473684211e-05, | |
| "loss": 23.0, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0007045301287277192, | |
| "grad_norm": 5.602340752375312e-05, | |
| "learning_rate": 8.718947368421053e-05, | |
| "loss": 23.0, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0007246595609770826, | |
| "grad_norm": 3.0846869776723906e-05, | |
| "learning_rate": 8.666105263157895e-05, | |
| "loss": 23.0, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0007447889932264461, | |
| "grad_norm": 5.5015039833961055e-05, | |
| "learning_rate": 8.613263157894737e-05, | |
| "loss": 23.0, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0007649184254758095, | |
| "grad_norm": 3.8054538890719414e-05, | |
| "learning_rate": 8.560421052631578e-05, | |
| "loss": 23.0, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0007850478577251729, | |
| "grad_norm": 2.42203677771613e-05, | |
| "learning_rate": 8.50757894736842e-05, | |
| "loss": 23.0, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0008051772899745363, | |
| "grad_norm": 4.2876483348663896e-05, | |
| "learning_rate": 8.454736842105263e-05, | |
| "loss": 23.0, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0008253067222238997, | |
| "grad_norm": 3.370803460711613e-05, | |
| "learning_rate": 8.401894736842106e-05, | |
| "loss": 23.0, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.000845436154473263, | |
| "grad_norm": 4.167450242675841e-05, | |
| "learning_rate": 8.349052631578948e-05, | |
| "loss": 23.0, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0008655655867226265, | |
| "grad_norm": 4.232488572597504e-05, | |
| "learning_rate": 8.29621052631579e-05, | |
| "loss": 23.0, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0008856950189719899, | |
| "grad_norm": 2.6339081159676425e-05, | |
| "learning_rate": 8.243368421052632e-05, | |
| "loss": 23.0, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0009058244512213533, | |
| "grad_norm": 4.41583979409188e-05, | |
| "learning_rate": 8.190526315789474e-05, | |
| "loss": 23.0, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0009259538834707167, | |
| "grad_norm": 7.86347227403894e-05, | |
| "learning_rate": 8.137684210526315e-05, | |
| "loss": 23.0, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0009460833157200801, | |
| "grad_norm": 4.264622475602664e-05, | |
| "learning_rate": 8.084842105263157e-05, | |
| "loss": 23.0, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.0009662127479694435, | |
| "grad_norm": 3.8661462895106524e-05, | |
| "learning_rate": 8.032e-05, | |
| "loss": 23.0, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.000986342180218807, | |
| "grad_norm": 7.720923167653382e-05, | |
| "learning_rate": 7.979157894736842e-05, | |
| "loss": 23.0, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0010064716124681702, | |
| "grad_norm": 4.70045160909649e-05, | |
| "learning_rate": 7.926315789473684e-05, | |
| "loss": 23.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0010064716124681702, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 285.9875, | |
| "eval_samples_per_second": 73.143, | |
| "eval_steps_per_second": 18.288, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0010266010447175337, | |
| "grad_norm": 8.320242159243207e-06, | |
| "learning_rate": 7.873473684210526e-05, | |
| "loss": 23.0, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0010467304769668972, | |
| "grad_norm": 1.5145632460189518e-05, | |
| "learning_rate": 7.820631578947369e-05, | |
| "loss": 23.0, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.0010668599092162605, | |
| "grad_norm": 1.1512304809002671e-05, | |
| "learning_rate": 7.76778947368421e-05, | |
| "loss": 23.0, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.001086989341465624, | |
| "grad_norm": 1.528786015114747e-05, | |
| "learning_rate": 7.714947368421052e-05, | |
| "loss": 23.0, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0011071187737149873, | |
| "grad_norm": 2.5456898583797738e-05, | |
| "learning_rate": 7.662105263157896e-05, | |
| "loss": 23.0, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0011272482059643508, | |
| "grad_norm": 2.668505476322025e-05, | |
| "learning_rate": 7.609263157894737e-05, | |
| "loss": 23.0, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0011473776382137143, | |
| "grad_norm": 1.551048808323685e-05, | |
| "learning_rate": 7.556421052631579e-05, | |
| "loss": 23.0, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.0011675070704630776, | |
| "grad_norm": 2.366452827118337e-05, | |
| "learning_rate": 7.503578947368421e-05, | |
| "loss": 23.0, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.001187636502712441, | |
| "grad_norm": 2.2682577764499e-05, | |
| "learning_rate": 7.450736842105263e-05, | |
| "loss": 23.0, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.0012077659349618043, | |
| "grad_norm": 5.866533319931477e-05, | |
| "learning_rate": 7.397894736842105e-05, | |
| "loss": 23.0, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0012278953672111678, | |
| "grad_norm": 1.637460991332773e-05, | |
| "learning_rate": 7.345052631578948e-05, | |
| "loss": 23.0, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.0012480247994605311, | |
| "grad_norm": 1.6382557078031823e-05, | |
| "learning_rate": 7.29221052631579e-05, | |
| "loss": 23.0, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0012681542317098946, | |
| "grad_norm": 2.3177421098807827e-05, | |
| "learning_rate": 7.239368421052631e-05, | |
| "loss": 23.0, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0012882836639592581, | |
| "grad_norm": 6.291714089456946e-05, | |
| "learning_rate": 7.186526315789474e-05, | |
| "loss": 23.0, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.0013084130962086214, | |
| "grad_norm": 7.432617712765932e-05, | |
| "learning_rate": 7.133684210526316e-05, | |
| "loss": 23.0, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0013285425284579849, | |
| "grad_norm": 3.0208946554921567e-05, | |
| "learning_rate": 7.080842105263158e-05, | |
| "loss": 23.0, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0013486719607073482, | |
| "grad_norm": 3.2406318496214226e-05, | |
| "learning_rate": 7.028e-05, | |
| "loss": 23.0, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0013688013929567117, | |
| "grad_norm": 3.3409804018447176e-05, | |
| "learning_rate": 6.975157894736843e-05, | |
| "loss": 23.0, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0013889308252060752, | |
| "grad_norm": 4.153459667577408e-05, | |
| "learning_rate": 6.922315789473685e-05, | |
| "loss": 23.0, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.0014090602574554384, | |
| "grad_norm": 3.1347994081443176e-05, | |
| "learning_rate": 6.869473684210527e-05, | |
| "loss": 23.0, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.001429189689704802, | |
| "grad_norm": 3.106446092715487e-05, | |
| "learning_rate": 6.816631578947368e-05, | |
| "loss": 23.0, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0014493191219541652, | |
| "grad_norm": 2.5425708372495137e-05, | |
| "learning_rate": 6.76378947368421e-05, | |
| "loss": 23.0, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0014694485542035287, | |
| "grad_norm": 8.160302240867168e-05, | |
| "learning_rate": 6.710947368421052e-05, | |
| "loss": 23.0, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0014895779864528922, | |
| "grad_norm": 8.36909530335106e-05, | |
| "learning_rate": 6.658105263157894e-05, | |
| "loss": 23.0, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.0015097074187022555, | |
| "grad_norm": 5.806803164887242e-05, | |
| "learning_rate": 6.605263157894737e-05, | |
| "loss": 23.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.001529836850951619, | |
| "grad_norm": 2.5655506760813296e-05, | |
| "learning_rate": 6.55242105263158e-05, | |
| "loss": 23.0, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0015499662832009823, | |
| "grad_norm": 5.792943193227984e-05, | |
| "learning_rate": 6.499578947368422e-05, | |
| "loss": 23.0, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0015700957154503458, | |
| "grad_norm": 4.383608029456809e-05, | |
| "learning_rate": 6.446736842105264e-05, | |
| "loss": 23.0, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.001590225147699709, | |
| "grad_norm": 4.217608875478618e-05, | |
| "learning_rate": 6.393894736842105e-05, | |
| "loss": 23.0, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0016103545799490725, | |
| "grad_norm": 0.0001017999675241299, | |
| "learning_rate": 6.341052631578947e-05, | |
| "loss": 23.0, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.001630484012198436, | |
| "grad_norm": 7.253052172018215e-05, | |
| "learning_rate": 6.288210526315789e-05, | |
| "loss": 23.0, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0016506134444477993, | |
| "grad_norm": 4.4168151362100616e-05, | |
| "learning_rate": 6.235368421052632e-05, | |
| "loss": 23.0, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0016707428766971628, | |
| "grad_norm": 4.690126297646202e-05, | |
| "learning_rate": 6.182526315789474e-05, | |
| "loss": 23.0, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.001690872308946526, | |
| "grad_norm": 5.543129373108968e-05, | |
| "learning_rate": 6.129684210526316e-05, | |
| "loss": 23.0, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0017110017411958896, | |
| "grad_norm": 3.422376903472468e-05, | |
| "learning_rate": 6.076842105263158e-05, | |
| "loss": 23.0, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.001731131173445253, | |
| "grad_norm": 0.00011534958321135491, | |
| "learning_rate": 6.024e-05, | |
| "loss": 23.0, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.0017512606056946164, | |
| "grad_norm": 5.733649231842719e-05, | |
| "learning_rate": 5.971157894736842e-05, | |
| "loss": 23.0, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0017713900379439798, | |
| "grad_norm": 9.903619502438232e-05, | |
| "learning_rate": 5.9183157894736835e-05, | |
| "loss": 23.0, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.0017915194701933431, | |
| "grad_norm": 7.510482100769877e-05, | |
| "learning_rate": 5.8654736842105267e-05, | |
| "loss": 23.0, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0018116489024427066, | |
| "grad_norm": 4.391166294226423e-05, | |
| "learning_rate": 5.8126315789473684e-05, | |
| "loss": 23.0, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0018317783346920701, | |
| "grad_norm": 0.00010185052815359086, | |
| "learning_rate": 5.759789473684211e-05, | |
| "loss": 23.0, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0018519077669414334, | |
| "grad_norm": 9.979342576116323e-05, | |
| "learning_rate": 5.706947368421053e-05, | |
| "loss": 23.0, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.001872037199190797, | |
| "grad_norm": 9.317202056990936e-05, | |
| "learning_rate": 5.6541052631578945e-05, | |
| "loss": 23.0, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0018921666314401602, | |
| "grad_norm": 7.172553159762174e-05, | |
| "learning_rate": 5.601263157894736e-05, | |
| "loss": 23.0, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0019122960636895237, | |
| "grad_norm": 5.0452064897399396e-05, | |
| "learning_rate": 5.5484210526315794e-05, | |
| "loss": 23.0, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.001932425495938887, | |
| "grad_norm": 0.00011866455315612257, | |
| "learning_rate": 5.495578947368421e-05, | |
| "loss": 23.0, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0019525549281882504, | |
| "grad_norm": 6.334228237392381e-05, | |
| "learning_rate": 5.442736842105264e-05, | |
| "loss": 23.0, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.001972684360437614, | |
| "grad_norm": 6.854250386822969e-05, | |
| "learning_rate": 5.3898947368421055e-05, | |
| "loss": 23.0, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.0019928137926869772, | |
| "grad_norm": 0.00011966691090492532, | |
| "learning_rate": 5.337052631578947e-05, | |
| "loss": 23.0, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0020129432249363405, | |
| "grad_norm": 9.832581417867914e-05, | |
| "learning_rate": 5.284210526315789e-05, | |
| "loss": 23.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0020129432249363405, | |
| "eval_loss": 11.5, | |
| "eval_runtime": 286.0829, | |
| "eval_samples_per_second": 73.119, | |
| "eval_steps_per_second": 18.281, | |
| "step": 100 | |
| } | |
| ], | |
| "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": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 4120303042560.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |