qwen2.5-0.5b-expo-DPO-ES-100 / train_results.json
hZzy's picture
Model save
f04c3bc verified
raw
history blame
230 Bytes
{
"epoch": 2.69248937175248,
"total_flos": 0.0,
"train_loss": 70.38767440143384,
"train_runtime": 26138.5983,
"train_samples": 50802,
"train_samples_per_second": 9.718,
"train_steps_per_second": 0.067
}