autoprogrammer's picture
Upload checkpoint from deepseek_esft_summary_lr2e-4_ste_ckpt921
2165b1a verified
raw
history blame contribute delete
181 Bytes
{
"_from_model_config": true,
"bos_token_id": 100000,
"do_sample": true,
"eos_token_id": 100001,
"temperature": 0.3,
"top_p": 0.95,
"transformers_version": "4.51.3"
}