mega-ar-350m-L3t-v0.08-ultraTBfw / train_results.json
pszemraj's picture
End of training
adc5ed7 verified
{
"epoch": 0.9998921074234783,
"num_input_tokens_seen": 3492282368,
"total_flos": 4.5723602603621745e+18,
"train_loss": 2.154333936901462,
"train_runtime": 99522.0612,
"train_samples": 852698,
"train_samples_per_second": 8.568,
"train_steps_per_second": 0.067
}