[Model] support bitsandbytes quantization with minicpm model (#10842)

Signed-off-by: Ubuntu <zixuanzhang@bytedance.com>

vllm/model_executor/models/minicpm.py

@@ -534,6 +534,16 @@ class MiniCPMForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
     }
     embedding_padding_modules = ["lm_head"]
 
+    # BitandBytes specific attributes
+    bitsandbytes_stacked_params_mapping = {
+        # shard_name, weight_name, index
+        "q_proj": ("qkv_proj", 0),
+        "k_proj": ("qkv_proj", 1),
+        "v_proj": ("qkv_proj", 2),
+        "gate_proj": ("gate_up_proj", 0),
+        "up_proj": ("gate_up_proj", 1),
+    }
+
     def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
         super().__init__()
         config = vllm_config.model_config.hf_config