|
{ |
|
"best_metric": 11.5, |
|
"best_model_checkpoint": "miner_id_24/checkpoint-50", |
|
"epoch": 0.004025886449872681, |
|
"eval_steps": 50, |
|
"global_step": 200, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 2.0129432249363408e-05, |
|
"grad_norm": 1.510836500528967e-05, |
|
"learning_rate": 1.0018000000000001e-05, |
|
"loss": 23.0, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 2.0129432249363408e-05, |
|
"eval_loss": 11.5, |
|
"eval_runtime": 90.192, |
|
"eval_samples_per_second": 231.927, |
|
"eval_steps_per_second": 57.987, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 4.0258864498726816e-05, |
|
"grad_norm": 1.5213749975373503e-05, |
|
"learning_rate": 2.0036000000000003e-05, |
|
"loss": 23.0, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 6.038829674809022e-05, |
|
"grad_norm": 1.9886019799741916e-05, |
|
"learning_rate": 3.0054e-05, |
|
"loss": 23.0, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 8.051772899745363e-05, |
|
"grad_norm": 8.898293344827835e-06, |
|
"learning_rate": 4.0072000000000005e-05, |
|
"loss": 23.0, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.00010064716124681703, |
|
"grad_norm": 9.695926564745605e-06, |
|
"learning_rate": 5.009e-05, |
|
"loss": 23.0, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.00012077659349618043, |
|
"grad_norm": 1.9819382941932417e-05, |
|
"learning_rate": 6.0108e-05, |
|
"loss": 23.0, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.00014090602574554385, |
|
"grad_norm": 1.6191377653740346e-05, |
|
"learning_rate": 7.0126e-05, |
|
"loss": 23.0, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.00016103545799490726, |
|
"grad_norm": 2.2024883946869522e-05, |
|
"learning_rate": 8.014400000000001e-05, |
|
"loss": 23.0, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.00018116489024427065, |
|
"grad_norm": 9.54006736719748e-06, |
|
"learning_rate": 9.016200000000001e-05, |
|
"loss": 23.0, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.00020129432249363407, |
|
"grad_norm": 1.384806091664359e-05, |
|
"learning_rate": 0.00010018, |
|
"loss": 23.0, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.00022142375474299748, |
|
"grad_norm": 2.238066190329846e-05, |
|
"learning_rate": 9.965273684210526e-05, |
|
"loss": 23.0, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.00024155318699236087, |
|
"grad_norm": 2.1979891243972816e-05, |
|
"learning_rate": 9.912547368421053e-05, |
|
"loss": 23.0, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.0002616826192417243, |
|
"grad_norm": 2.554473394411616e-05, |
|
"learning_rate": 9.859821052631579e-05, |
|
"loss": 23.0, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.0002818120514910877, |
|
"grad_norm": 1.498403707955731e-05, |
|
"learning_rate": 9.807094736842106e-05, |
|
"loss": 23.0, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.0003019414837404511, |
|
"grad_norm": 1.537183197797276e-05, |
|
"learning_rate": 9.754368421052633e-05, |
|
"loss": 23.0, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.00032207091598981453, |
|
"grad_norm": 2.7295525796944275e-05, |
|
"learning_rate": 9.701642105263158e-05, |
|
"loss": 23.0, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.0003422003482391779, |
|
"grad_norm": 1.6758611309342086e-05, |
|
"learning_rate": 9.648915789473685e-05, |
|
"loss": 23.0, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.0003623297804885413, |
|
"grad_norm": 1.7154054148704745e-05, |
|
"learning_rate": 9.596189473684211e-05, |
|
"loss": 23.0, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.00038245921273790474, |
|
"grad_norm": 2.9581389753730036e-05, |
|
"learning_rate": 9.543463157894737e-05, |
|
"loss": 23.0, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.00040258864498726813, |
|
"grad_norm": 2.7940845029661432e-05, |
|
"learning_rate": 9.490736842105264e-05, |
|
"loss": 23.0, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.0004227180772366315, |
|
"grad_norm": 1.2358163985481951e-05, |
|
"learning_rate": 9.43801052631579e-05, |
|
"loss": 23.0, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.00044284750948599496, |
|
"grad_norm": 1.571402754052542e-05, |
|
"learning_rate": 9.385284210526316e-05, |
|
"loss": 23.0, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.00046297694173535835, |
|
"grad_norm": 1.4946596820664126e-05, |
|
"learning_rate": 9.332557894736843e-05, |
|
"loss": 23.0, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.00048310637398472174, |
|
"grad_norm": 2.304871304659173e-05, |
|
"learning_rate": 9.279831578947369e-05, |
|
"loss": 23.0, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.0005032358062340851, |
|
"grad_norm": 4.3125663069076836e-05, |
|
"learning_rate": 9.227105263157896e-05, |
|
"loss": 23.0, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.0005233652384834486, |
|
"grad_norm": 1.9486795281409286e-05, |
|
"learning_rate": 9.174378947368421e-05, |
|
"loss": 23.0, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.000543494670732812, |
|
"grad_norm": 2.001772736548446e-05, |
|
"learning_rate": 9.121652631578948e-05, |
|
"loss": 23.0, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.0005636241029821754, |
|
"grad_norm": 2.597480488475412e-05, |
|
"learning_rate": 9.068926315789475e-05, |
|
"loss": 23.0, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.0005837535352315388, |
|
"grad_norm": 8.075709774857387e-05, |
|
"learning_rate": 9.016200000000001e-05, |
|
"loss": 23.0, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.0006038829674809022, |
|
"grad_norm": 4.117946809856221e-05, |
|
"learning_rate": 8.963473684210526e-05, |
|
"loss": 23.0, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.0006240123997302656, |
|
"grad_norm": 5.761471402365714e-05, |
|
"learning_rate": 8.910747368421053e-05, |
|
"loss": 23.0, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.0006441418319796291, |
|
"grad_norm": 4.664399602916092e-05, |
|
"learning_rate": 8.858021052631579e-05, |
|
"loss": 23.0, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.0006642712642289924, |
|
"grad_norm": 2.7544720069272444e-05, |
|
"learning_rate": 8.805294736842106e-05, |
|
"loss": 23.0, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.0006844006964783558, |
|
"grad_norm": 4.744183024740778e-05, |
|
"learning_rate": 8.752568421052633e-05, |
|
"loss": 23.0, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.0007045301287277192, |
|
"grad_norm": 5.79297193326056e-05, |
|
"learning_rate": 8.699842105263159e-05, |
|
"loss": 23.0, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.0007246595609770826, |
|
"grad_norm": 2.9632537916768342e-05, |
|
"learning_rate": 8.647115789473686e-05, |
|
"loss": 23.0, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.0007447889932264461, |
|
"grad_norm": 5.4365860705729574e-05, |
|
"learning_rate": 8.594389473684211e-05, |
|
"loss": 23.0, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.0007649184254758095, |
|
"grad_norm": 3.470982119324617e-05, |
|
"learning_rate": 8.541663157894737e-05, |
|
"loss": 23.0, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.0007850478577251729, |
|
"grad_norm": 2.286666858708486e-05, |
|
"learning_rate": 8.488936842105264e-05, |
|
"loss": 23.0, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.0008051772899745363, |
|
"grad_norm": 3.8486385165015236e-05, |
|
"learning_rate": 8.43621052631579e-05, |
|
"loss": 23.0, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.0008253067222238997, |
|
"grad_norm": 3.153427678626031e-05, |
|
"learning_rate": 8.383484210526316e-05, |
|
"loss": 23.0, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.000845436154473263, |
|
"grad_norm": 3.871660373988561e-05, |
|
"learning_rate": 8.330757894736843e-05, |
|
"loss": 23.0, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.0008655655867226265, |
|
"grad_norm": 4.188347884337418e-05, |
|
"learning_rate": 8.278031578947369e-05, |
|
"loss": 23.0, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.0008856950189719899, |
|
"grad_norm": 2.5132152586593293e-05, |
|
"learning_rate": 8.225305263157896e-05, |
|
"loss": 23.0, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.0009058244512213533, |
|
"grad_norm": 4.127430293010548e-05, |
|
"learning_rate": 8.172578947368422e-05, |
|
"loss": 23.0, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.0009259538834707167, |
|
"grad_norm": 7.43012860766612e-05, |
|
"learning_rate": 8.119852631578947e-05, |
|
"loss": 23.0, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.0009460833157200801, |
|
"grad_norm": 4.009860276710242e-05, |
|
"learning_rate": 8.067126315789474e-05, |
|
"loss": 23.0, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.0009662127479694435, |
|
"grad_norm": 3.689499862957746e-05, |
|
"learning_rate": 8.014400000000001e-05, |
|
"loss": 23.0, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.000986342180218807, |
|
"grad_norm": 6.844763993285596e-05, |
|
"learning_rate": 7.961673684210527e-05, |
|
"loss": 23.0, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.0010064716124681702, |
|
"grad_norm": 4.541236921795644e-05, |
|
"learning_rate": 7.908947368421054e-05, |
|
"loss": 23.0, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.0010064716124681702, |
|
"eval_loss": 11.5, |
|
"eval_runtime": 90.1841, |
|
"eval_samples_per_second": 231.948, |
|
"eval_steps_per_second": 57.992, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.0010266010447175337, |
|
"grad_norm": 7.401086350000696e-06, |
|
"learning_rate": 7.856221052631579e-05, |
|
"loss": 23.0, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.0010467304769668972, |
|
"grad_norm": 1.3705683159059845e-05, |
|
"learning_rate": 7.803494736842106e-05, |
|
"loss": 23.0, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.0010668599092162605, |
|
"grad_norm": 1.0768772881419864e-05, |
|
"learning_rate": 7.750768421052632e-05, |
|
"loss": 23.0, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.001086989341465624, |
|
"grad_norm": 1.4973178622312844e-05, |
|
"learning_rate": 7.698042105263157e-05, |
|
"loss": 23.0, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.0011071187737149873, |
|
"grad_norm": 2.5119483325397596e-05, |
|
"learning_rate": 7.645315789473686e-05, |
|
"loss": 23.0, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.0011272482059643508, |
|
"grad_norm": 2.6307834559702314e-05, |
|
"learning_rate": 7.592589473684211e-05, |
|
"loss": 23.0, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.0011473776382137143, |
|
"grad_norm": 1.5282301319530234e-05, |
|
"learning_rate": 7.539863157894737e-05, |
|
"loss": 23.0, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.0011675070704630776, |
|
"grad_norm": 2.157269773306325e-05, |
|
"learning_rate": 7.487136842105264e-05, |
|
"loss": 23.0, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.001187636502712441, |
|
"grad_norm": 1.986996539926622e-05, |
|
"learning_rate": 7.43441052631579e-05, |
|
"loss": 23.0, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.0012077659349618043, |
|
"grad_norm": 5.521572893485427e-05, |
|
"learning_rate": 7.381684210526315e-05, |
|
"loss": 23.0, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.0012278953672111678, |
|
"grad_norm": 1.4630297300755046e-05, |
|
"learning_rate": 7.328957894736844e-05, |
|
"loss": 23.0, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.0012480247994605311, |
|
"grad_norm": 1.5226029063342139e-05, |
|
"learning_rate": 7.276231578947369e-05, |
|
"loss": 23.0, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.0012681542317098946, |
|
"grad_norm": 2.2780744984629564e-05, |
|
"learning_rate": 7.223505263157895e-05, |
|
"loss": 23.0, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.0012882836639592581, |
|
"grad_norm": 5.959494592389092e-05, |
|
"learning_rate": 7.170778947368422e-05, |
|
"loss": 23.0, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.0013084130962086214, |
|
"grad_norm": 6.706695421598852e-05, |
|
"learning_rate": 7.118052631578947e-05, |
|
"loss": 23.0, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.0013285425284579849, |
|
"grad_norm": 2.7232843422098085e-05, |
|
"learning_rate": 7.065326315789474e-05, |
|
"loss": 23.0, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.0013486719607073482, |
|
"grad_norm": 2.8405625926097855e-05, |
|
"learning_rate": 7.0126e-05, |
|
"loss": 23.0, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.0013688013929567117, |
|
"grad_norm": 3.136721352348104e-05, |
|
"learning_rate": 6.959873684210527e-05, |
|
"loss": 23.0, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.0013889308252060752, |
|
"grad_norm": 3.615265086409636e-05, |
|
"learning_rate": 6.907147368421054e-05, |
|
"loss": 23.0, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.0014090602574554384, |
|
"grad_norm": 2.7944166504312307e-05, |
|
"learning_rate": 6.85442105263158e-05, |
|
"loss": 23.0, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.001429189689704802, |
|
"grad_norm": 2.80793356068898e-05, |
|
"learning_rate": 6.801694736842105e-05, |
|
"loss": 23.0, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.0014493191219541652, |
|
"grad_norm": 2.3540800611954182e-05, |
|
"learning_rate": 6.748968421052632e-05, |
|
"loss": 23.0, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.0014694485542035287, |
|
"grad_norm": 7.162094698287547e-05, |
|
"learning_rate": 6.696242105263158e-05, |
|
"loss": 23.0, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.0014895779864528922, |
|
"grad_norm": 8.528814214514568e-05, |
|
"learning_rate": 6.643515789473685e-05, |
|
"loss": 23.0, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.0015097074187022555, |
|
"grad_norm": 4.904473462374881e-05, |
|
"learning_rate": 6.590789473684212e-05, |
|
"loss": 23.0, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.001529836850951619, |
|
"grad_norm": 2.2618882212555036e-05, |
|
"learning_rate": 6.538063157894737e-05, |
|
"loss": 23.0, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.0015499662832009823, |
|
"grad_norm": 5.1661736506503075e-05, |
|
"learning_rate": 6.485336842105264e-05, |
|
"loss": 23.0, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.0015700957154503458, |
|
"grad_norm": 3.921652751159854e-05, |
|
"learning_rate": 6.43261052631579e-05, |
|
"loss": 23.0, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.001590225147699709, |
|
"grad_norm": 3.7143105146242306e-05, |
|
"learning_rate": 6.379884210526315e-05, |
|
"loss": 23.0, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.0016103545799490725, |
|
"grad_norm": 8.97124846233055e-05, |
|
"learning_rate": 6.327157894736842e-05, |
|
"loss": 23.0, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.001630484012198436, |
|
"grad_norm": 6.435957766370848e-05, |
|
"learning_rate": 6.274431578947368e-05, |
|
"loss": 23.0, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.0016506134444477993, |
|
"grad_norm": 4.061551589984447e-05, |
|
"learning_rate": 6.221705263157895e-05, |
|
"loss": 23.0, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.0016707428766971628, |
|
"grad_norm": 4.216823435854167e-05, |
|
"learning_rate": 6.168978947368422e-05, |
|
"loss": 23.0, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.001690872308946526, |
|
"grad_norm": 4.681897917180322e-05, |
|
"learning_rate": 6.116252631578948e-05, |
|
"loss": 23.0, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.0017110017411958896, |
|
"grad_norm": 3.0376580980373546e-05, |
|
"learning_rate": 6.063526315789474e-05, |
|
"loss": 23.0, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.001731131173445253, |
|
"grad_norm": 0.00010400725295767188, |
|
"learning_rate": 6.0108e-05, |
|
"loss": 23.0, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.0017512606056946164, |
|
"grad_norm": 4.685055318986997e-05, |
|
"learning_rate": 5.9580736842105264e-05, |
|
"loss": 23.0, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.0017713900379439798, |
|
"grad_norm": 8.442284888587892e-05, |
|
"learning_rate": 5.905347368421053e-05, |
|
"loss": 23.0, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.0017915194701933431, |
|
"grad_norm": 6.63155151414685e-05, |
|
"learning_rate": 5.85262105263158e-05, |
|
"loss": 23.0, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.0018116489024427066, |
|
"grad_norm": 3.886529157171026e-05, |
|
"learning_rate": 5.799894736842106e-05, |
|
"loss": 23.0, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.0018317783346920701, |
|
"grad_norm": 8.638443250674754e-05, |
|
"learning_rate": 5.747168421052632e-05, |
|
"loss": 23.0, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.0018519077669414334, |
|
"grad_norm": 8.624787005828694e-05, |
|
"learning_rate": 5.694442105263158e-05, |
|
"loss": 23.0, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.001872037199190797, |
|
"grad_norm": 7.803425978636369e-05, |
|
"learning_rate": 5.641715789473684e-05, |
|
"loss": 23.0, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.0018921666314401602, |
|
"grad_norm": 5.7351535360794514e-05, |
|
"learning_rate": 5.5889894736842104e-05, |
|
"loss": 23.0, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.0019122960636895237, |
|
"grad_norm": 4.696400719694793e-05, |
|
"learning_rate": 5.5362631578947374e-05, |
|
"loss": 23.0, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.001932425495938887, |
|
"grad_norm": 0.00011351658031344414, |
|
"learning_rate": 5.483536842105264e-05, |
|
"loss": 23.0, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.0019525549281882504, |
|
"grad_norm": 5.443142435979098e-05, |
|
"learning_rate": 5.43081052631579e-05, |
|
"loss": 23.0, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.001972684360437614, |
|
"grad_norm": 5.9478716138983145e-05, |
|
"learning_rate": 5.378084210526316e-05, |
|
"loss": 23.0, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.0019928137926869772, |
|
"grad_norm": 0.00010627473966451362, |
|
"learning_rate": 5.3253578947368426e-05, |
|
"loss": 23.0, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.0020129432249363405, |
|
"grad_norm": 8.376106416108087e-05, |
|
"learning_rate": 5.272631578947368e-05, |
|
"loss": 23.0, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.0020129432249363405, |
|
"eval_loss": 11.5, |
|
"eval_runtime": 90.2118, |
|
"eval_samples_per_second": 231.877, |
|
"eval_steps_per_second": 57.975, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.002033072657185704, |
|
"grad_norm": 3.085847492911853e-05, |
|
"learning_rate": 5.2199052631578945e-05, |
|
"loss": 23.0, |
|
"step": 101 |
|
}, |
|
{ |
|
"epoch": 0.0020532020894350675, |
|
"grad_norm": 1.759108818077948e-05, |
|
"learning_rate": 5.167178947368422e-05, |
|
"loss": 23.0, |
|
"step": 102 |
|
}, |
|
{ |
|
"epoch": 0.0020733315216844308, |
|
"grad_norm": 2.8313368602539413e-05, |
|
"learning_rate": 5.114452631578948e-05, |
|
"loss": 23.0, |
|
"step": 103 |
|
}, |
|
{ |
|
"epoch": 0.0020934609539337945, |
|
"grad_norm": 3.0228055038605817e-05, |
|
"learning_rate": 5.061726315789474e-05, |
|
"loss": 23.0, |
|
"step": 104 |
|
}, |
|
{ |
|
"epoch": 0.0021135903861831578, |
|
"grad_norm": 2.273799509566743e-05, |
|
"learning_rate": 5.009e-05, |
|
"loss": 23.0, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.002133719818432521, |
|
"grad_norm": 5.4774165619164705e-05, |
|
"learning_rate": 4.9562736842105266e-05, |
|
"loss": 23.0, |
|
"step": 106 |
|
}, |
|
{ |
|
"epoch": 0.0021538492506818843, |
|
"grad_norm": 1.822214835556224e-05, |
|
"learning_rate": 4.903547368421053e-05, |
|
"loss": 23.0, |
|
"step": 107 |
|
}, |
|
{ |
|
"epoch": 0.002173978682931248, |
|
"grad_norm": 3.304668643977493e-05, |
|
"learning_rate": 4.850821052631579e-05, |
|
"loss": 23.0, |
|
"step": 108 |
|
}, |
|
{ |
|
"epoch": 0.0021941081151806113, |
|
"grad_norm": 2.679731187527068e-05, |
|
"learning_rate": 4.7980947368421055e-05, |
|
"loss": 23.0, |
|
"step": 109 |
|
}, |
|
{ |
|
"epoch": 0.0022142375474299746, |
|
"grad_norm": 4.0604074456496164e-05, |
|
"learning_rate": 4.745368421052632e-05, |
|
"loss": 23.0, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.0022343669796793383, |
|
"grad_norm": 3.5479981306707487e-05, |
|
"learning_rate": 4.692642105263158e-05, |
|
"loss": 23.0, |
|
"step": 111 |
|
}, |
|
{ |
|
"epoch": 0.0022544964119287016, |
|
"grad_norm": 2.652855073392857e-05, |
|
"learning_rate": 4.6399157894736844e-05, |
|
"loss": 23.0, |
|
"step": 112 |
|
}, |
|
{ |
|
"epoch": 0.002274625844178065, |
|
"grad_norm": 3.958947127102874e-05, |
|
"learning_rate": 4.5871894736842107e-05, |
|
"loss": 23.0, |
|
"step": 113 |
|
}, |
|
{ |
|
"epoch": 0.0022947552764274286, |
|
"grad_norm": 4.629440081771463e-05, |
|
"learning_rate": 4.5344631578947376e-05, |
|
"loss": 23.0, |
|
"step": 114 |
|
}, |
|
{ |
|
"epoch": 0.002314884708676792, |
|
"grad_norm": 4.368358349893242e-05, |
|
"learning_rate": 4.481736842105263e-05, |
|
"loss": 23.0, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.002335014140926155, |
|
"grad_norm": 2.5798855858738534e-05, |
|
"learning_rate": 4.4290105263157895e-05, |
|
"loss": 23.0, |
|
"step": 116 |
|
}, |
|
{ |
|
"epoch": 0.0023551435731755184, |
|
"grad_norm": 5.440914537757635e-05, |
|
"learning_rate": 4.3762842105263165e-05, |
|
"loss": 23.0, |
|
"step": 117 |
|
}, |
|
{ |
|
"epoch": 0.002375273005424882, |
|
"grad_norm": 3.2479510991834104e-05, |
|
"learning_rate": 4.323557894736843e-05, |
|
"loss": 23.0, |
|
"step": 118 |
|
}, |
|
{ |
|
"epoch": 0.0023954024376742454, |
|
"grad_norm": 3.85025778086856e-05, |
|
"learning_rate": 4.2708315789473684e-05, |
|
"loss": 23.0, |
|
"step": 119 |
|
}, |
|
{ |
|
"epoch": 0.0024155318699236087, |
|
"grad_norm": 5.463774505187757e-05, |
|
"learning_rate": 4.218105263157895e-05, |
|
"loss": 23.0, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.0024356613021729724, |
|
"grad_norm": 1.561428689456079e-05, |
|
"learning_rate": 4.1653789473684217e-05, |
|
"loss": 23.0, |
|
"step": 121 |
|
}, |
|
{ |
|
"epoch": 0.0024557907344223357, |
|
"grad_norm": 4.556514249998145e-05, |
|
"learning_rate": 4.112652631578948e-05, |
|
"loss": 23.0, |
|
"step": 122 |
|
}, |
|
{ |
|
"epoch": 0.002475920166671699, |
|
"grad_norm": 7.233426003949717e-05, |
|
"learning_rate": 4.0599263157894736e-05, |
|
"loss": 23.0, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 0.0024960495989210622, |
|
"grad_norm": 3.5000026400666684e-05, |
|
"learning_rate": 4.0072000000000005e-05, |
|
"loss": 23.0, |
|
"step": 124 |
|
}, |
|
{ |
|
"epoch": 0.002516179031170426, |
|
"grad_norm": 4.47658276243601e-05, |
|
"learning_rate": 3.954473684210527e-05, |
|
"loss": 23.0, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 0.0025363084634197892, |
|
"grad_norm": 4.98553563375026e-05, |
|
"learning_rate": 3.901747368421053e-05, |
|
"loss": 23.0, |
|
"step": 126 |
|
}, |
|
{ |
|
"epoch": 0.0025564378956691525, |
|
"grad_norm": 5.663814954459667e-05, |
|
"learning_rate": 3.849021052631579e-05, |
|
"loss": 23.0, |
|
"step": 127 |
|
}, |
|
{ |
|
"epoch": 0.0025765673279185162, |
|
"grad_norm": 7.687522884225473e-05, |
|
"learning_rate": 3.796294736842106e-05, |
|
"loss": 23.0, |
|
"step": 128 |
|
}, |
|
{ |
|
"epoch": 0.0025966967601678795, |
|
"grad_norm": 0.00012378401879686862, |
|
"learning_rate": 3.743568421052632e-05, |
|
"loss": 23.0, |
|
"step": 129 |
|
}, |
|
{ |
|
"epoch": 0.0026168261924172428, |
|
"grad_norm": 7.353304681601003e-05, |
|
"learning_rate": 3.6908421052631576e-05, |
|
"loss": 23.0, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.0026369556246666065, |
|
"grad_norm": 3.6857436498394236e-05, |
|
"learning_rate": 3.6381157894736846e-05, |
|
"loss": 23.0, |
|
"step": 131 |
|
}, |
|
{ |
|
"epoch": 0.0026570850569159698, |
|
"grad_norm": 5.3032148571219295e-05, |
|
"learning_rate": 3.585389473684211e-05, |
|
"loss": 23.0, |
|
"step": 132 |
|
}, |
|
{ |
|
"epoch": 0.002677214489165333, |
|
"grad_norm": 5.596369373961352e-05, |
|
"learning_rate": 3.532663157894737e-05, |
|
"loss": 23.0, |
|
"step": 133 |
|
}, |
|
{ |
|
"epoch": 0.0026973439214146963, |
|
"grad_norm": 6.214417953742668e-05, |
|
"learning_rate": 3.4799368421052634e-05, |
|
"loss": 23.0, |
|
"step": 134 |
|
}, |
|
{ |
|
"epoch": 0.00271747335366406, |
|
"grad_norm": 0.00012366275768727064, |
|
"learning_rate": 3.42721052631579e-05, |
|
"loss": 23.0, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 0.0027376027859134233, |
|
"grad_norm": 0.00010686524183256552, |
|
"learning_rate": 3.374484210526316e-05, |
|
"loss": 23.0, |
|
"step": 136 |
|
}, |
|
{ |
|
"epoch": 0.0027577322181627866, |
|
"grad_norm": 6.914123514434323e-05, |
|
"learning_rate": 3.321757894736842e-05, |
|
"loss": 23.0, |
|
"step": 137 |
|
}, |
|
{ |
|
"epoch": 0.0027778616504121503, |
|
"grad_norm": 8.196205453714356e-05, |
|
"learning_rate": 3.2690315789473686e-05, |
|
"loss": 23.0, |
|
"step": 138 |
|
}, |
|
{ |
|
"epoch": 0.0027979910826615136, |
|
"grad_norm": 5.320821946952492e-05, |
|
"learning_rate": 3.216305263157895e-05, |
|
"loss": 23.0, |
|
"step": 139 |
|
}, |
|
{ |
|
"epoch": 0.002818120514910877, |
|
"grad_norm": 7.619400275871158e-05, |
|
"learning_rate": 3.163578947368421e-05, |
|
"loss": 23.0, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.00283824994716024, |
|
"grad_norm": 8.360709034604952e-05, |
|
"learning_rate": 3.1108526315789475e-05, |
|
"loss": 23.0, |
|
"step": 141 |
|
}, |
|
{ |
|
"epoch": 0.002858379379409604, |
|
"grad_norm": 8.154918759828433e-05, |
|
"learning_rate": 3.058126315789474e-05, |
|
"loss": 23.0, |
|
"step": 142 |
|
}, |
|
{ |
|
"epoch": 0.002878508811658967, |
|
"grad_norm": 8.84397595655173e-05, |
|
"learning_rate": 3.0054e-05, |
|
"loss": 23.0, |
|
"step": 143 |
|
}, |
|
{ |
|
"epoch": 0.0028986382439083304, |
|
"grad_norm": 9.481957386014983e-05, |
|
"learning_rate": 2.9526736842105263e-05, |
|
"loss": 23.0, |
|
"step": 144 |
|
}, |
|
{ |
|
"epoch": 0.002918767676157694, |
|
"grad_norm": 0.00012848826008848846, |
|
"learning_rate": 2.899947368421053e-05, |
|
"loss": 23.0, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 0.0029388971084070574, |
|
"grad_norm": 4.97991859447211e-05, |
|
"learning_rate": 2.847221052631579e-05, |
|
"loss": 23.0, |
|
"step": 146 |
|
}, |
|
{ |
|
"epoch": 0.0029590265406564207, |
|
"grad_norm": 0.00014037324581295252, |
|
"learning_rate": 2.7944947368421052e-05, |
|
"loss": 23.0, |
|
"step": 147 |
|
}, |
|
{ |
|
"epoch": 0.0029791559729057844, |
|
"grad_norm": 9.983405470848083e-05, |
|
"learning_rate": 2.741768421052632e-05, |
|
"loss": 23.0, |
|
"step": 148 |
|
}, |
|
{ |
|
"epoch": 0.0029992854051551477, |
|
"grad_norm": 0.00019755310495384037, |
|
"learning_rate": 2.689042105263158e-05, |
|
"loss": 23.0, |
|
"step": 149 |
|
}, |
|
{ |
|
"epoch": 0.003019414837404511, |
|
"grad_norm": 0.0001168611051980406, |
|
"learning_rate": 2.636315789473684e-05, |
|
"loss": 23.0, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.003019414837404511, |
|
"eval_loss": 11.5, |
|
"eval_runtime": 90.2465, |
|
"eval_samples_per_second": 231.787, |
|
"eval_steps_per_second": 57.952, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.0030395442696538742, |
|
"grad_norm": 2.2188000002643093e-05, |
|
"learning_rate": 2.583589473684211e-05, |
|
"loss": 23.0, |
|
"step": 151 |
|
}, |
|
{ |
|
"epoch": 0.003059673701903238, |
|
"grad_norm": 3.3306067052762955e-05, |
|
"learning_rate": 2.530863157894737e-05, |
|
"loss": 23.0, |
|
"step": 152 |
|
}, |
|
{ |
|
"epoch": 0.0030798031341526012, |
|
"grad_norm": 7.252175419125706e-05, |
|
"learning_rate": 2.4781368421052633e-05, |
|
"loss": 23.0, |
|
"step": 153 |
|
}, |
|
{ |
|
"epoch": 0.0030999325664019645, |
|
"grad_norm": 2.680643774510827e-05, |
|
"learning_rate": 2.4254105263157896e-05, |
|
"loss": 23.0, |
|
"step": 154 |
|
}, |
|
{ |
|
"epoch": 0.0031200619986513282, |
|
"grad_norm": 3.1729945476399735e-05, |
|
"learning_rate": 2.372684210526316e-05, |
|
"loss": 23.0, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 0.0031401914309006915, |
|
"grad_norm": 2.2617728973273188e-05, |
|
"learning_rate": 2.3199578947368422e-05, |
|
"loss": 23.0, |
|
"step": 156 |
|
}, |
|
{ |
|
"epoch": 0.003160320863150055, |
|
"grad_norm": 3.523330087773502e-05, |
|
"learning_rate": 2.2672315789473688e-05, |
|
"loss": 23.0, |
|
"step": 157 |
|
}, |
|
{ |
|
"epoch": 0.003180450295399418, |
|
"grad_norm": 1.8790780814015307e-05, |
|
"learning_rate": 2.2145052631578948e-05, |
|
"loss": 23.0, |
|
"step": 158 |
|
}, |
|
{ |
|
"epoch": 0.0032005797276487818, |
|
"grad_norm": 1.8678092601476237e-05, |
|
"learning_rate": 2.1617789473684214e-05, |
|
"loss": 23.0, |
|
"step": 159 |
|
}, |
|
{ |
|
"epoch": 0.003220709159898145, |
|
"grad_norm": 2.8310923880781047e-05, |
|
"learning_rate": 2.1090526315789473e-05, |
|
"loss": 23.0, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.0032408385921475083, |
|
"grad_norm": 4.5371118176262826e-05, |
|
"learning_rate": 2.056326315789474e-05, |
|
"loss": 23.0, |
|
"step": 161 |
|
}, |
|
{ |
|
"epoch": 0.003260968024396872, |
|
"grad_norm": 5.299994154484011e-05, |
|
"learning_rate": 2.0036000000000003e-05, |
|
"loss": 23.0, |
|
"step": 162 |
|
}, |
|
{ |
|
"epoch": 0.0032810974566462353, |
|
"grad_norm": 4.766113852383569e-05, |
|
"learning_rate": 1.9508736842105266e-05, |
|
"loss": 23.0, |
|
"step": 163 |
|
}, |
|
{ |
|
"epoch": 0.0033012268888955986, |
|
"grad_norm": 3.349925464135595e-05, |
|
"learning_rate": 1.898147368421053e-05, |
|
"loss": 23.0, |
|
"step": 164 |
|
}, |
|
{ |
|
"epoch": 0.0033213563211449623, |
|
"grad_norm": 6.378292164299637e-05, |
|
"learning_rate": 1.8454210526315788e-05, |
|
"loss": 23.0, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 0.0033414857533943256, |
|
"grad_norm": 4.2118925193790346e-05, |
|
"learning_rate": 1.7926947368421054e-05, |
|
"loss": 23.0, |
|
"step": 166 |
|
}, |
|
{ |
|
"epoch": 0.003361615185643689, |
|
"grad_norm": 5.257881275610998e-05, |
|
"learning_rate": 1.7399684210526317e-05, |
|
"loss": 23.0, |
|
"step": 167 |
|
}, |
|
{ |
|
"epoch": 0.003381744617893052, |
|
"grad_norm": 5.806943227071315e-05, |
|
"learning_rate": 1.687242105263158e-05, |
|
"loss": 23.0, |
|
"step": 168 |
|
}, |
|
{ |
|
"epoch": 0.003401874050142416, |
|
"grad_norm": 5.3176350775174797e-05, |
|
"learning_rate": 1.6345157894736843e-05, |
|
"loss": 23.0, |
|
"step": 169 |
|
}, |
|
{ |
|
"epoch": 0.003422003482391779, |
|
"grad_norm": 7.09736195858568e-05, |
|
"learning_rate": 1.5817894736842106e-05, |
|
"loss": 23.0, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.0034421329146411424, |
|
"grad_norm": 4.30285872425884e-05, |
|
"learning_rate": 1.529063157894737e-05, |
|
"loss": 23.0, |
|
"step": 171 |
|
}, |
|
{ |
|
"epoch": 0.003462262346890506, |
|
"grad_norm": 4.993961192667484e-05, |
|
"learning_rate": 1.4763368421052632e-05, |
|
"loss": 23.0, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 0.0034823917791398694, |
|
"grad_norm": 5.322639117366634e-05, |
|
"learning_rate": 1.4236105263157895e-05, |
|
"loss": 23.0, |
|
"step": 173 |
|
}, |
|
{ |
|
"epoch": 0.0035025212113892327, |
|
"grad_norm": 6.130256224423647e-05, |
|
"learning_rate": 1.370884210526316e-05, |
|
"loss": 23.0, |
|
"step": 174 |
|
}, |
|
{ |
|
"epoch": 0.003522650643638596, |
|
"grad_norm": 6.289948214543983e-05, |
|
"learning_rate": 1.318157894736842e-05, |
|
"loss": 23.0, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 0.0035427800758879597, |
|
"grad_norm": 3.8983846025075763e-05, |
|
"learning_rate": 1.2654315789473685e-05, |
|
"loss": 23.0, |
|
"step": 176 |
|
}, |
|
{ |
|
"epoch": 0.003562909508137323, |
|
"grad_norm": 7.520937651861459e-05, |
|
"learning_rate": 1.2127052631578948e-05, |
|
"loss": 23.0, |
|
"step": 177 |
|
}, |
|
{ |
|
"epoch": 0.0035830389403866863, |
|
"grad_norm": 3.817744072875939e-05, |
|
"learning_rate": 1.1599789473684211e-05, |
|
"loss": 23.0, |
|
"step": 178 |
|
}, |
|
{ |
|
"epoch": 0.00360316837263605, |
|
"grad_norm": 7.804012420820072e-05, |
|
"learning_rate": 1.1072526315789474e-05, |
|
"loss": 23.0, |
|
"step": 179 |
|
}, |
|
{ |
|
"epoch": 0.0036232978048854132, |
|
"grad_norm": 5.756770769949071e-05, |
|
"learning_rate": 1.0545263157894737e-05, |
|
"loss": 23.0, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.0036434272371347765, |
|
"grad_norm": 8.38518826640211e-05, |
|
"learning_rate": 1.0018000000000001e-05, |
|
"loss": 23.0, |
|
"step": 181 |
|
}, |
|
{ |
|
"epoch": 0.0036635566693841402, |
|
"grad_norm": 4.2767831473611295e-05, |
|
"learning_rate": 9.490736842105264e-06, |
|
"loss": 23.0, |
|
"step": 182 |
|
}, |
|
{ |
|
"epoch": 0.0036836861016335035, |
|
"grad_norm": 4.7194145736284554e-05, |
|
"learning_rate": 8.963473684210527e-06, |
|
"loss": 23.0, |
|
"step": 183 |
|
}, |
|
{ |
|
"epoch": 0.003703815533882867, |
|
"grad_norm": 8.454011549474671e-05, |
|
"learning_rate": 8.43621052631579e-06, |
|
"loss": 23.0, |
|
"step": 184 |
|
}, |
|
{ |
|
"epoch": 0.00372394496613223, |
|
"grad_norm": 0.00015326568973250687, |
|
"learning_rate": 7.908947368421053e-06, |
|
"loss": 23.0, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 0.003744074398381594, |
|
"grad_norm": 6.415565439965576e-05, |
|
"learning_rate": 7.381684210526316e-06, |
|
"loss": 23.0, |
|
"step": 186 |
|
}, |
|
{ |
|
"epoch": 0.003764203830630957, |
|
"grad_norm": 5.757789404015057e-05, |
|
"learning_rate": 6.85442105263158e-06, |
|
"loss": 23.0, |
|
"step": 187 |
|
}, |
|
{ |
|
"epoch": 0.0037843332628803203, |
|
"grad_norm": 0.0001279564603464678, |
|
"learning_rate": 6.3271578947368425e-06, |
|
"loss": 23.0, |
|
"step": 188 |
|
}, |
|
{ |
|
"epoch": 0.003804462695129684, |
|
"grad_norm": 0.00011098594404757023, |
|
"learning_rate": 5.7998947368421054e-06, |
|
"loss": 23.0, |
|
"step": 189 |
|
}, |
|
{ |
|
"epoch": 0.0038245921273790473, |
|
"grad_norm": 7.304556493181735e-05, |
|
"learning_rate": 5.272631578947368e-06, |
|
"loss": 23.0, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.0038447215596284106, |
|
"grad_norm": 5.6200729886768386e-05, |
|
"learning_rate": 4.745368421052632e-06, |
|
"loss": 23.0, |
|
"step": 191 |
|
}, |
|
{ |
|
"epoch": 0.003864850991877774, |
|
"grad_norm": 7.393292617052794e-05, |
|
"learning_rate": 4.218105263157895e-06, |
|
"loss": 23.0, |
|
"step": 192 |
|
}, |
|
{ |
|
"epoch": 0.0038849804241271376, |
|
"grad_norm": 9.384314762428403e-05, |
|
"learning_rate": 3.690842105263158e-06, |
|
"loss": 23.0, |
|
"step": 193 |
|
}, |
|
{ |
|
"epoch": 0.003905109856376501, |
|
"grad_norm": 7.593140617245808e-05, |
|
"learning_rate": 3.1635789473684213e-06, |
|
"loss": 23.0, |
|
"step": 194 |
|
}, |
|
{ |
|
"epoch": 0.003925239288625864, |
|
"grad_norm": 0.00012393394717946649, |
|
"learning_rate": 2.636315789473684e-06, |
|
"loss": 23.0, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 0.003945368720875228, |
|
"grad_norm": 5.397694985731505e-05, |
|
"learning_rate": 2.1090526315789475e-06, |
|
"loss": 23.0, |
|
"step": 196 |
|
}, |
|
{ |
|
"epoch": 0.003965498153124591, |
|
"grad_norm": 0.00010385041969129816, |
|
"learning_rate": 1.5817894736842106e-06, |
|
"loss": 23.0, |
|
"step": 197 |
|
}, |
|
{ |
|
"epoch": 0.0039856275853739544, |
|
"grad_norm": 9.945422061719e-05, |
|
"learning_rate": 1.0545263157894738e-06, |
|
"loss": 23.0, |
|
"step": 198 |
|
}, |
|
{ |
|
"epoch": 0.004005757017623318, |
|
"grad_norm": 0.00011496926163090393, |
|
"learning_rate": 5.272631578947369e-07, |
|
"loss": 23.0, |
|
"step": 199 |
|
}, |
|
{ |
|
"epoch": 0.004025886449872681, |
|
"grad_norm": 0.0001539279764983803, |
|
"learning_rate": 0.0, |
|
"loss": 23.0, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.004025886449872681, |
|
"eval_loss": 11.5, |
|
"eval_runtime": 90.2378, |
|
"eval_samples_per_second": 231.81, |
|
"eval_steps_per_second": 57.958, |
|
"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": 3 |
|
} |
|
}, |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": true |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 8280804163584.0, |
|
"train_batch_size": 4, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|