gsstbst / train_state.json
wind77's picture
Upload folder using huggingface_hub
21064ba verified
{
"step": 1300,
"kind": "best",
"train_rolling_sparse_kl": 0.1843354889191687,
"teacher": "Qwen/Qwen3.5-35B-A3B",
"data_position": 54690,
"step_mean_sparse_kl": 0.2065462488681078
}