Devstral-24B-AWQ / generation_config.json
mattbucci's picture
Devstral 24B AWQ: GPTQ-calibrated, BOS-fixed chat template, 37 tok/s on RDNA4
df87209 verified
raw
history blame contribute delete
153 Bytes
{
"_from_model_config": true,
"bos_token_id": 1,
"eos_token_id": 2,
"pad_token_id": 11,
"transformers_version": "5.3.0",
"use_cache": true
}