mamba_0_5_dpo_ep3 / train_results.json
Junxiong Wang
add models
24a7edb
raw
history blame contribute delete
219 Bytes
{
"epoch": 3.0,
"total_flos": 0.0,
"train_loss": 0.22024025758457094,
"train_runtime": 20844.5321,
"train_samples": 61134,
"train_samples_per_second": 8.799,
"train_steps_per_second": 0.275
}