QC-Llama-3.2-1B_6e_CPT / train_results.json
e3ham's picture
Initial adapter (6E CPT) + tokenizer + config
f12b8d1 verified
raw
history blame contribute delete
208 Bytes
{
"epoch": 6.0,
"total_flos": 1.7755774072777605e+18,
"train_loss": 3.266454470654329,
"train_runtime": 95277.1031,
"train_samples_per_second": 6.163,
"train_steps_per_second": 0.012
}