[Misc]Further reduce BNB static variable (#10597)

Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
2024-11-27 14:54:12 +08:00
parent e85250b1d1
commit 15cc2a9f1a
14 changed files with 131 additions and 219 deletions
--- a/vllm/model_executor/models/qwen2.py
+++ b/vllm/model_executor/models/qwen2.py
@@ -419,15 +419,6 @@ class Qwen2ForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
    embedding_padding_modules = []

    # BitandBytes specific attributes
-    default_bitsandbytes_target_modules = [
-        ".gate_proj.",
-        ".down_proj.",
-        ".up_proj.",
-        ".q_proj.",
-        ".k_proj.",
-        ".v_proj.",
-        ".o_proj.",
-    ]
    bitsandbytes_stacked_params_mapping = {
        # shard_name, weight_name, index
        "q_proj": ("qkv_proj", 0),