add deepseekv3 and llama4
This commit is contained in:
@@ -483,6 +483,7 @@ class _ModelRegistry:
|
||||
architectures = [architectures]
|
||||
if not architectures:
|
||||
logger.warning("No model architectures are specified")
|
||||
return []
|
||||
|
||||
return architectures
|
||||
|
||||
|
||||
@@ -229,6 +229,14 @@ def get_config(
|
||||
model_type = MODEL_FOR_CAUSAL_LM_MAPPING_NAMES[config.model_type]
|
||||
config.update({"architectures": [model_type]})
|
||||
|
||||
# Some composite config classes (e.g. Llama4Config) may not preserve
|
||||
# the 'architectures' field from config.json. Restore it from the
|
||||
# raw config_dict if needed.
|
||||
if config_format == ConfigFormat.HF:
|
||||
raw_archs = config_dict.get("architectures")
|
||||
if raw_archs and not getattr(config, "architectures", None):
|
||||
config.architectures = raw_archs
|
||||
|
||||
patch_rope_scaling(config)
|
||||
|
||||
if trust_remote_code:
|
||||
|
||||
Reference in New Issue
Block a user