SmolVLM-256M-Instruct-SFT / all_results.json
osman93's picture
Add checkpoint -1 post-trained on curated_deepscaler
8b9b791 verified
raw
history blame
180 Bytes
{
"total_flos": 259411378176.0,
"train_loss": 2.135700225830078,
"train_runtime": 126.1654,
"train_samples_per_second": 0.024,
"train_steps_per_second": 0.024
}