[Misc] Reduce LoRA-related static variable (#13166)

This commit is contained in:
Jee Jee Li
2025-02-22 00:21:30 -08:00
committed by GitHub
parent 2cb8c1540e
commit 105b8ce4c0
41 changed files with 120 additions and 395 deletions
-11
View File
@@ -390,17 +390,6 @@ class Gemma2ForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
],
}
# LoRA specific attributes
supported_lora_modules = [
"qkv_proj",
"o_proj",
"gate_up_proj",
"down_proj",
]
# Gemma does not apply LoRA to the embedding layer.
embedding_modules = {}
embedding_padding_modules = []
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
config = vllm_config.model_config.hf_config
quant_config = vllm_config.quant_config