TinyLlama-DPO / all_results.json
TheGuyWithoutH's picture
Deliverable 1: DPO model trained on custom data v1.3
8fcdfa9 verified
raw
history blame contribute delete
173 Bytes
{
"epoch": 2.93,
"train_loss": 0.6931471824645996,
"train_runtime": 1837.2852,
"train_samples_per_second": 0.558,
"train_steps_per_second": 0.034
}