DeepSeek-V3-0324-AWQ / generation_config.json
v2ray's picture
Added modeling code, fixed cache, and added prefill ability.
a9608c6
raw
history blame contribute delete
137 Bytes
{
"_from_model_config": true,
"bos_token_id": 0,
"do_sample": true,
"eos_token_id": 1,
"transformers_version": "4.48.0.dev0"
}