vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

convert-*.py: base_model is actually in spec for model cards

Browse source
This commit is contained in:
brian khuu 2024-06-04 00:28:16 +10:00
parent 4d5cd0670a
commit 32e80e094c
1 changed files with 0 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

2
gguf-py/gguf/metadata.py
View file

@ -51,8 +51,6 @@ class Metadata:
metadata.name = model_card.get("model_name") metadata.name = model_card.get("model_name")
if "base_model" in model_card: if "base_model" in model_card:
# Not part of huggingface model card standard but notice some model creator using it
# such as TheBloke who would encode 'mistralai/Mixtral-8x7B-Instruct-v0.1' into base_model
model_id = model_card.get("base_model") model_id = model_card.get("base_model")
model_name_normal, organization_name, base_name, fine_tune, version_string, parameter_weight_class = Metadata.get_model_name_components(model_id) model_name_normal, organization_name, base_name, fine_tune, version_string, parameter_weight_class = Metadata.get_model_name_components(model_id)