Mistral-lb-wura-9b / config.json
KoseiUemura's picture
Upload entire model folder
4d1106c verified
raw
history blame contribute delete
323 Bytes
{
"alignments": "linear",
"architectures": [
"LangBridgeModel"
],
"dim_enc": 2048,
"dim_lm": 4096,
"enc": "llama-lang-adapt/mt5-xl-lm-wura",
"freeze_encoder": true,
"freeze_language_model": true,
"lm": "meta-math/MetaMath-Mistral-7B",
"torch_dtype": "bfloat16",
"transformers_version": "4.37.2"
}