smolm-autoreg-bpe-counterfactual_babylm_aann_high_variability_numeral-seed_1024-1e-3
/
train_results.json
{ | |
"epoch": 20.0, | |
"total_flos": 1.5670047538944e+18, | |
"train_loss": 3.0279207580395733, | |
"train_runtime": 82395.3872, | |
"train_samples": 595065, | |
"train_samples_per_second": 144.441, | |
"train_steps_per_second": 4.514 | |
} |