|
{ |
|
"best_metric": 5.6417765617370605, |
|
"best_model_checkpoint": "bert_base_48/checkpoint-70000", |
|
"epoch": 0.6501007717642432, |
|
"global_step": 79350, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 5.000000000000001e-07, |
|
"loss": 6.0105, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.0000000000000002e-06, |
|
"loss": 6.0097, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 1.5e-06, |
|
"loss": 6.0142, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 2.0000000000000003e-06, |
|
"loss": 6.008, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 2.5e-06, |
|
"loss": 6.0069, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 3e-06, |
|
"loss": 6.0074, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 3.5e-06, |
|
"loss": 6.0085, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.000000000000001e-06, |
|
"loss": 6.0159, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.5e-06, |
|
"loss": 6.0081, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 5e-06, |
|
"loss": 6.0135, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 5.500000000000001e-06, |
|
"loss": 6.0059, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 6e-06, |
|
"loss": 6.0055, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 6.5000000000000004e-06, |
|
"loss": 6.0125, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 7e-06, |
|
"loss": 6.0067, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 7.500000000000001e-06, |
|
"loss": 6.01, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 8.000000000000001e-06, |
|
"loss": 6.0015, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 8.5e-06, |
|
"loss": 5.9958, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 9e-06, |
|
"loss": 6.0018, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 9.5e-06, |
|
"loss": 6.0014, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 1e-05, |
|
"loss": 6.005, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"eval_accuracy": 0.15133511156733304, |
|
"eval_loss": 5.999912261962891, |
|
"eval_runtime": 2027.3389, |
|
"eval_samples_per_second": 152.099, |
|
"eval_steps_per_second": 3.169, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.991670692498626e-06, |
|
"loss": 6.0055, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.983341384997253e-06, |
|
"loss": 6.0101, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 9.975012077495878e-06, |
|
"loss": 6.005, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.966682769994504e-06, |
|
"loss": 6.0053, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 9.95835346249313e-06, |
|
"loss": 5.9975, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.950024154991756e-06, |
|
"loss": 5.9999, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.941694847490381e-06, |
|
"loss": 6.0007, |
|
"step": 13500 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 9.933365539989006e-06, |
|
"loss": 5.9992, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.925036232487631e-06, |
|
"loss": 5.9928, |
|
"step": 14500 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 9.916706924986258e-06, |
|
"loss": 5.9949, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.908377617484883e-06, |
|
"loss": 5.9857, |
|
"step": 15500 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 9.90004830998351e-06, |
|
"loss": 5.9919, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.891719002482134e-06, |
|
"loss": 5.9964, |
|
"step": 16500 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.883389694980761e-06, |
|
"loss": 5.9864, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 9.875060387479386e-06, |
|
"loss": 5.9927, |
|
"step": 17500 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.866731079978011e-06, |
|
"loss": 5.9906, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 9.858401772476636e-06, |
|
"loss": 5.9938, |
|
"step": 18500 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.850072464975263e-06, |
|
"loss": 5.987, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.84174315747389e-06, |
|
"loss": 5.9845, |
|
"step": 19500 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 9.833413849972514e-06, |
|
"loss": 5.9845, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"eval_accuracy": 0.15109292721310397, |
|
"eval_loss": 5.981122016906738, |
|
"eval_runtime": 2027.3895, |
|
"eval_samples_per_second": 152.095, |
|
"eval_steps_per_second": 3.169, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.82508454247114e-06, |
|
"loss": 5.9837, |
|
"step": 20500 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 9.816755234969766e-06, |
|
"loss": 5.981, |
|
"step": 21000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.808425927468391e-06, |
|
"loss": 5.9865, |
|
"step": 21500 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.800096619967016e-06, |
|
"loss": 5.9776, |
|
"step": 22000 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 9.791767312465643e-06, |
|
"loss": 5.9836, |
|
"step": 22500 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.783438004964268e-06, |
|
"loss": 5.9718, |
|
"step": 23000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 9.775108697462895e-06, |
|
"loss": 5.9737, |
|
"step": 23500 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.76677938996152e-06, |
|
"loss": 5.9699, |
|
"step": 24000 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.758450082460145e-06, |
|
"loss": 5.9756, |
|
"step": 24500 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 9.750120774958771e-06, |
|
"loss": 5.9744, |
|
"step": 25000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.741791467457396e-06, |
|
"loss": 5.9701, |
|
"step": 25500 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 9.733462159956021e-06, |
|
"loss": 5.9688, |
|
"step": 26000 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.725132852454648e-06, |
|
"loss": 5.9758, |
|
"step": 26500 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 9.716803544953275e-06, |
|
"loss": 5.9749, |
|
"step": 27000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.7084742374519e-06, |
|
"loss": 5.9696, |
|
"step": 27500 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.700144929950525e-06, |
|
"loss": 5.9697, |
|
"step": 28000 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 9.69181562244915e-06, |
|
"loss": 5.9726, |
|
"step": 28500 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.683486314947776e-06, |
|
"loss": 5.9701, |
|
"step": 29000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 9.675157007446401e-06, |
|
"loss": 5.9674, |
|
"step": 29500 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.666827699945028e-06, |
|
"loss": 5.9695, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"eval_accuracy": 0.15177876498535858, |
|
"eval_loss": 5.967329025268555, |
|
"eval_runtime": 2024.8948, |
|
"eval_samples_per_second": 152.282, |
|
"eval_steps_per_second": 3.173, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.658498392443653e-06, |
|
"loss": 5.9639, |
|
"step": 30500 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 9.65016908494228e-06, |
|
"loss": 5.9663, |
|
"step": 31000 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 9.641839777440905e-06, |
|
"loss": 5.97, |
|
"step": 31500 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 9.63351046993953e-06, |
|
"loss": 5.9643, |
|
"step": 32000 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 9.625181162438155e-06, |
|
"loss": 5.9634, |
|
"step": 32500 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 9.616851854936781e-06, |
|
"loss": 5.9626, |
|
"step": 33000 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 9.608522547435406e-06, |
|
"loss": 5.9595, |
|
"step": 33500 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 9.600193239934033e-06, |
|
"loss": 5.962, |
|
"step": 34000 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 9.591863932432658e-06, |
|
"loss": 5.9563, |
|
"step": 34500 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 9.583534624931285e-06, |
|
"loss": 5.9599, |
|
"step": 35000 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 9.57520531742991e-06, |
|
"loss": 5.9666, |
|
"step": 35500 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 9.566876009928535e-06, |
|
"loss": 5.9626, |
|
"step": 36000 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 9.55854670242716e-06, |
|
"loss": 5.9612, |
|
"step": 36500 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 9.550217394925787e-06, |
|
"loss": 5.9589, |
|
"step": 37000 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 9.541888087424413e-06, |
|
"loss": 5.9577, |
|
"step": 37500 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 9.533558779923038e-06, |
|
"loss": 5.9486, |
|
"step": 38000 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 9.525229472421663e-06, |
|
"loss": 5.9533, |
|
"step": 38500 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 9.51690016492029e-06, |
|
"loss": 5.9502, |
|
"step": 39000 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 9.508570857418915e-06, |
|
"loss": 5.9535, |
|
"step": 39500 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 9.50024154991754e-06, |
|
"loss": 5.9599, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"eval_accuracy": 0.1515399918178128, |
|
"eval_loss": 5.951314449310303, |
|
"eval_runtime": 2027.5096, |
|
"eval_samples_per_second": 152.086, |
|
"eval_steps_per_second": 3.169, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 9.491912242416167e-06, |
|
"loss": 5.9516, |
|
"step": 40500 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 9.483582934914792e-06, |
|
"loss": 5.9521, |
|
"step": 41000 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 9.475253627413418e-06, |
|
"loss": 5.9509, |
|
"step": 41500 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 9.466924319912043e-06, |
|
"loss": 5.9557, |
|
"step": 42000 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 9.458595012410668e-06, |
|
"loss": 5.9461, |
|
"step": 42500 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 9.450265704909295e-06, |
|
"loss": 5.9464, |
|
"step": 43000 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 9.44193639740792e-06, |
|
"loss": 5.9435, |
|
"step": 43500 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 9.433607089906545e-06, |
|
"loss": 5.9501, |
|
"step": 44000 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 9.425277782405172e-06, |
|
"loss": 5.9461, |
|
"step": 44500 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 9.416948474903798e-06, |
|
"loss": 5.9454, |
|
"step": 45000 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 9.408619167402423e-06, |
|
"loss": 5.9485, |
|
"step": 45500 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 9.400289859901048e-06, |
|
"loss": 5.941, |
|
"step": 46000 |
|
}, |
|
{ |
|
"epoch": 0.38, |
|
"learning_rate": 9.391960552399673e-06, |
|
"loss": 5.9398, |
|
"step": 46500 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 9.3836312448983e-06, |
|
"loss": 5.9368, |
|
"step": 47000 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 9.375301937396925e-06, |
|
"loss": 5.9387, |
|
"step": 47500 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 9.366972629895552e-06, |
|
"loss": 5.9411, |
|
"step": 48000 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 9.358643322394177e-06, |
|
"loss": 5.9376, |
|
"step": 48500 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 9.350314014892804e-06, |
|
"loss": 5.9365, |
|
"step": 49000 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 9.341984707391429e-06, |
|
"loss": 5.9397, |
|
"step": 49500 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 9.333655399890054e-06, |
|
"loss": 5.9402, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"eval_accuracy": 0.15165703958938734, |
|
"eval_loss": 5.941145896911621, |
|
"eval_runtime": 2026.3662, |
|
"eval_samples_per_second": 152.172, |
|
"eval_steps_per_second": 3.171, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 9.325326092388679e-06, |
|
"loss": 5.9347, |
|
"step": 50500 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 9.316996784887305e-06, |
|
"loss": 5.9378, |
|
"step": 51000 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 9.30866747738593e-06, |
|
"loss": 5.941, |
|
"step": 51500 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 9.300338169884557e-06, |
|
"loss": 5.9373, |
|
"step": 52000 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 9.292008862383182e-06, |
|
"loss": 5.9347, |
|
"step": 52500 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 9.283679554881809e-06, |
|
"loss": 5.938, |
|
"step": 53000 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 9.275350247380434e-06, |
|
"loss": 5.9324, |
|
"step": 53500 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 9.267020939879059e-06, |
|
"loss": 5.937, |
|
"step": 54000 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 9.258691632377685e-06, |
|
"loss": 5.9361, |
|
"step": 54500 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 9.25036232487631e-06, |
|
"loss": 5.9299, |
|
"step": 55000 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 9.242033017374937e-06, |
|
"loss": 5.9377, |
|
"step": 55500 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 9.233703709873562e-06, |
|
"loss": 5.9326, |
|
"step": 56000 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 9.225374402372189e-06, |
|
"loss": 5.9291, |
|
"step": 56500 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 9.217045094870814e-06, |
|
"loss": 5.9213, |
|
"step": 57000 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 9.208715787369439e-06, |
|
"loss": 5.9325, |
|
"step": 57500 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 9.200386479868064e-06, |
|
"loss": 5.9333, |
|
"step": 58000 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 9.19205717236669e-06, |
|
"loss": 5.9311, |
|
"step": 58500 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 9.183727864865315e-06, |
|
"loss": 5.9276, |
|
"step": 59000 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 9.175398557363942e-06, |
|
"loss": 5.931, |
|
"step": 59500 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 9.167069249862567e-06, |
|
"loss": 5.9291, |
|
"step": 60000 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"eval_accuracy": 0.1517665705603436, |
|
"eval_loss": 5.921680450439453, |
|
"eval_runtime": 2027.9259, |
|
"eval_samples_per_second": 152.055, |
|
"eval_steps_per_second": 3.168, |
|
"step": 60000 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 9.158739942361194e-06, |
|
"loss": 5.9136, |
|
"step": 60500 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 9.150410634859819e-06, |
|
"loss": 5.8956, |
|
"step": 61000 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 9.142081327358444e-06, |
|
"loss": 5.8873, |
|
"step": 61500 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 9.133752019857069e-06, |
|
"loss": 5.8676, |
|
"step": 62000 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 9.125422712355696e-06, |
|
"loss": 5.8634, |
|
"step": 62500 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 9.11709340485432e-06, |
|
"loss": 5.8416, |
|
"step": 63000 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 9.108764097352947e-06, |
|
"loss": 5.8083, |
|
"step": 63500 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 9.100434789851572e-06, |
|
"loss": 5.7886, |
|
"step": 64000 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.092105482350199e-06, |
|
"loss": 5.7781, |
|
"step": 64500 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 9.083776174848824e-06, |
|
"loss": 5.7662, |
|
"step": 65000 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.075446867347449e-06, |
|
"loss": 5.7568, |
|
"step": 65500 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.067117559846076e-06, |
|
"loss": 5.7486, |
|
"step": 66000 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 9.0587882523447e-06, |
|
"loss": 5.7321, |
|
"step": 66500 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.050458944843327e-06, |
|
"loss": 5.7241, |
|
"step": 67000 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 9.042129637341952e-06, |
|
"loss": 5.7168, |
|
"step": 67500 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 9.033800329840577e-06, |
|
"loss": 5.7137, |
|
"step": 68000 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 9.025471022339204e-06, |
|
"loss": 5.7048, |
|
"step": 68500 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 9.017141714837829e-06, |
|
"loss": 5.6969, |
|
"step": 69000 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 9.008812407336454e-06, |
|
"loss": 5.6905, |
|
"step": 69500 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 9.00048309983508e-06, |
|
"loss": 5.6816, |
|
"step": 70000 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"eval_accuracy": 0.1662817301507095, |
|
"eval_loss": 5.6417765617370605, |
|
"eval_runtime": 2026.7686, |
|
"eval_samples_per_second": 152.142, |
|
"eval_steps_per_second": 3.17, |
|
"step": 70000 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.992153792333706e-06, |
|
"loss": 5.6705, |
|
"step": 70500 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 8.983824484832332e-06, |
|
"loss": 5.6607, |
|
"step": 71000 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.975495177330957e-06, |
|
"loss": 5.6629, |
|
"step": 71500 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.967165869829582e-06, |
|
"loss": 5.6536, |
|
"step": 72000 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 8.958836562328209e-06, |
|
"loss": 5.6492, |
|
"step": 72500 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 8.950507254826834e-06, |
|
"loss": 5.6365, |
|
"step": 73000 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 8.942177947325459e-06, |
|
"loss": 5.6359, |
|
"step": 73500 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 8.933848639824086e-06, |
|
"loss": 5.6252, |
|
"step": 74000 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 8.925519332322713e-06, |
|
"loss": 5.6182, |
|
"step": 74500 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 8.917190024821338e-06, |
|
"loss": 5.6179, |
|
"step": 75000 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 8.908860717319963e-06, |
|
"loss": 5.6039, |
|
"step": 75500 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 8.900531409818588e-06, |
|
"loss": 5.6015, |
|
"step": 76000 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 8.892202102317214e-06, |
|
"loss": 5.5923, |
|
"step": 76500 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 8.88387279481584e-06, |
|
"loss": 5.5887, |
|
"step": 77000 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 8.875543487314466e-06, |
|
"loss": 5.5808, |
|
"step": 77500 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 8.867214179813091e-06, |
|
"loss": 5.5747, |
|
"step": 78000 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 8.858884872311718e-06, |
|
"loss": 5.5714, |
|
"step": 78500 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 8.850555564810343e-06, |
|
"loss": 5.5637, |
|
"step": 79000 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"step": 79350, |
|
"total_flos": 1.0024944886957998e+18, |
|
"train_loss": 5.903401354265123, |
|
"train_runtime": 86382.9395, |
|
"train_samples_per_second": 339.115, |
|
"train_steps_per_second": 7.065 |
|
} |
|
], |
|
"max_steps": 610290, |
|
"num_train_epochs": 5, |
|
"total_flos": 1.0024944886957998e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|