qwenvl-2B-cadica-direction-scale4 / train_results.json
ben81828's picture
End of training
093ca76 verified
raw
history blame contribute delete
261 Bytes
{
"epoch": 0.6185938707185166,
"num_input_tokens_seen": 11990784,
"total_flos": 791107965943808.0,
"train_loss": 6.11944748344618e-08,
"train_runtime": 102.3684,
"train_samples_per_second": 281.337,
"train_steps_per_second": 11.722
}