|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.9433962264150946, |
|
"global_step": 78, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 0.00013333333333333334, |
|
"loss": 1.5444, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 0.0002666666666666667, |
|
"loss": 1.543, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 0.0004, |
|
"loss": 1.3151, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 0.0003998245660197717, |
|
"loss": 1.2922, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 0.0003992985718499009, |
|
"loss": 1.1978, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 0.0003984229402628956, |
|
"loss": 1.1409, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 0.000397199207414101, |
|
"loss": 1.1181, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 0.00039562952014676116, |
|
"loss": 1.1124, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 0.0003937166322257262, |
|
"loss": 1.0952, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 0.00039146389950641346, |
|
"loss": 1.0885, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 0.00038887527404749624, |
|
"loss": 1.0787, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 0.00038595529717765027, |
|
"loss": 1.0854, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 0.0003827090915285202, |
|
"loss": 1.0807, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 0.0003791423520478826, |
|
"loss": 1.055, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 0.0003752613360087727, |
|
"loss": 1.056, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 0.00037107285203210135, |
|
"loss": 1.0526, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 0.0003665842481420199, |
|
"loss": 1.0693, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 0.0003618033988749895, |
|
"loss": 1.0471, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 0.000356738691465168, |
|
"loss": 1.0356, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 0.0003513990111303513, |
|
"loss": 1.0433, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 0.00034579372548428235, |
|
"loss": 1.0463, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 0.0003399326681026731, |
|
"loss": 1.0564, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 0.00033382612127177166, |
|
"loss": 1.0391, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 0.000327484797949738, |
|
"loss": 1.035, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 0.000320919822972475, |
|
"loss": 1.0371, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 0.0003141427135368864, |
|
"loss": 1.0296, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 0.0003071653589957993, |
|
"loss": 0.93, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 0.00030000000000000003, |
|
"loss": 0.9064, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 0.00029265920702397236, |
|
"loss": 0.904, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 0.00028515585831301456, |
|
"loss": 0.893, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 0.0002775031172904206, |
|
"loss": 0.8945, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 0.0002697144094643631, |
|
"loss": 0.8945, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 0.00026180339887498953, |
|
"loss": 0.8792, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 0.00025378396412305314, |
|
"loss": 0.8908, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 0.00024567017402213115, |
|
"loss": 0.8892, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 0.00023747626291714498, |
|
"loss": 0.8754, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 0.00022921660571248236, |
|
"loss": 0.8414, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 0.00022090569265353072, |
|
"loss": 0.8629, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 0.0002125581039058627, |
|
"loss": 0.8653, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 0.0002041884839766714, |
|
"loss": 0.8702, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 0.00019581151602332864, |
|
"loss": 0.8799, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 0.00018744189609413734, |
|
"loss": 0.8628, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 0.00017909430734646935, |
|
"loss": 0.8751, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 0.0001707833942875177, |
|
"loss": 0.8683, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 0.00016252373708285504, |
|
"loss": 0.8522, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 0.00015432982597786884, |
|
"loss": 0.8431, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 0.00014621603587694688, |
|
"loss": 0.8575, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 0.00013819660112501054, |
|
"loss": 0.8612, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 0.000130285590535637, |
|
"loss": 0.8791, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 0.0001224968827095794, |
|
"loss": 0.8104, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 0.00011484414168698547, |
|
"loss": 0.8223, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 0.00010734079297602771, |
|
"loss": 0.8642, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 0.00010000000000000005, |
|
"loss": 0.7986, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 9.283464100420063e-05, |
|
"loss": 0.7106, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 2.08, |
|
"learning_rate": 8.585728646311369e-05, |
|
"loss": 0.7073, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 7.908017702752504e-05, |
|
"loss": 0.7, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 2.15, |
|
"learning_rate": 7.251520205026205e-05, |
|
"loss": 0.7022, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 6.617387872822842e-05, |
|
"loss": 0.6812, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 6.0067331897326895e-05, |
|
"loss": 0.6941, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 5.4206274515717736e-05, |
|
"loss": 0.6749, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 2.3, |
|
"learning_rate": 4.8600988869648745e-05, |
|
"loss": 0.6833, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 4.326130853483206e-05, |
|
"loss": 0.6684, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 2.38, |
|
"learning_rate": 3.819660112501053e-05, |
|
"loss": 0.6482, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 2.42, |
|
"learning_rate": 3.3415751857980115e-05, |
|
"loss": 0.6638, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 2.45, |
|
"learning_rate": 2.8927147967898682e-05, |
|
"loss": 0.6586, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 2.49, |
|
"learning_rate": 2.473866399122733e-05, |
|
"loss": 0.6709, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 2.53, |
|
"learning_rate": 2.085764795211742e-05, |
|
"loss": 0.6518, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 2.57, |
|
"learning_rate": 1.7290908471479805e-05, |
|
"loss": 0.6567, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 2.6, |
|
"learning_rate": 1.4044702822349731e-05, |
|
"loss": 0.678, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 2.64, |
|
"learning_rate": 1.1124725952503801e-05, |
|
"loss": 0.6848, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 2.68, |
|
"learning_rate": 8.536100493586552e-06, |
|
"loss": 0.6601, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 2.72, |
|
"learning_rate": 6.283367774273785e-06, |
|
"loss": 0.6749, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 4.370479853238863e-06, |
|
"loss": 0.6864, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 2.79, |
|
"learning_rate": 2.800792585899026e-06, |
|
"loss": 0.6621, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 1.577059737104447e-06, |
|
"loss": 0.6514, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 2.87, |
|
"learning_rate": 7.01428150099126e-07, |
|
"loss": 0.6516, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 2.91, |
|
"learning_rate": 1.7543398022832336e-07, |
|
"loss": 0.6504, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"learning_rate": 0.0, |
|
"loss": 0.6665, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"step": 78, |
|
"total_flos": 183654252937216.0, |
|
"train_loss": 0.8923629743930621, |
|
"train_runtime": 2340.4996, |
|
"train_samples_per_second": 4.275, |
|
"train_steps_per_second": 0.033 |
|
} |
|
], |
|
"max_steps": 78, |
|
"num_train_epochs": 3, |
|
"total_flos": 183654252937216.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|