DiscoPOP-zephyr-7b-gemma / train_results.json
chrlu's picture
Duplicate from chrlu/zephyr-7b-gemma-log_ratio_modulated_loss
2a22bc1 verified
raw
history blame
230 Bytes
{
"epoch": 1.971563981042654,
"total_flos": 0.0,
"train_loss": 0.5405629426240921,
"train_runtime": 2172.3856,
"train_samples": 6750,
"train_samples_per_second": 6.214,
"train_steps_per_second": 0.048
}