llama3.2-1b-tamil / eval_results.json
HARISHSENTHIL
finetuning with llamafactory using H100 epoch over 30
89de10b
raw
history blame contribute delete
162 Bytes
{
"epoch": 28.2,
"eval_loss": 3.401175022125244,
"eval_runtime": 0.0709,
"eval_samples_per_second": 126.952,
"eval_steps_per_second": 28.212
}