Qwen2.5-3B-Instruct-OT3-8K-R1 / train_results.json
RZ412's picture
End of training
a11ec91 verified
{
"epoch": 5.0,
"total_flos": 620641233469440.0,
"train_loss": 0.5814858378295478,
"train_runtime": 136874.8782,
"train_samples_per_second": 0.304,
"train_steps_per_second": 0.038
}