Llama3.1-Mamba-8B-distill / train_results.json
Junxiong Wang
add models
0c247b3
raw
history blame contribute delete
254 Bytes
{
"epoch": 0.999995191868488,
"total_flos": 7.252179007674778e+16,
"train_loss": 276.11489062246443,
"train_runtime": 988432.6936,
"train_samples": 19473081,
"train_samples_per_second": 3.367,
"train_steps_per_second": 0.053
}