DeepSeek-V4-Flash-W4A16-FP8-MTP / generation_config.json
pastapaul's picture
Phase 2 full GPTQ + BF16 MTP: 89.1% MTP acceptance
e910552 verified
{
"_from_model_config": true,
"bos_token_id": 0,
"do_sample": true,
"eos_token_id": 1,
"temperature": 1.0,
"top_p": 1.0,
"transformers_version": "5.8.1"
}