[Misc]Add BNB quantization for Qwen2VL (#11719)

Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
Signed-off-by: Isotr0py <2037008807@qq.com>
Co-authored-by: Isotr0py <2037008807@qq.com>

vllm/model_executor/models/qwen2_vl.py

@@ -38,7 +38,7 @@ from transformers.models.qwen2_vl.image_processing_qwen2_vl import smart_resize
 
 from vllm.attention import AttentionMetadata
 from vllm.config import VllmConfig
-from vllm.distributed import parallel_state
+from vllm.distributed import parallel_state, tensor_model_parallel_all_gather
 from vllm.distributed import utils as dist_utils
 from vllm.logger import init_logger
 from vllm.model_executor import SamplingMetadata
@@ -239,6 +239,8 @@ class Qwen2VisionAttention(nn.Module):
         super().__init__()
         # Per attention head and per partition values.
         world_size = parallel_state.get_tensor_model_parallel_world_size()
+        self.tp_size = world_size
+        self.tp_rank = parallel_state.get_tensor_model_parallel_rank()
         self.hidden_size_per_attention_head = dist_utils.divide(
             projection_size, num_heads)
         self.num_attention_heads_per_partition = dist_utils.divide(
@@ -261,24 +263,41 @@ class Qwen2VisionAttention(nn.Module):
             raise RuntimeError(
                 f"Qwen2-VL does not support {self.attn_backend} backend now.")
 
+    def split_qkv(self, qkv: torch.Tensor) -> tuple[torch.Tensor, ...]:
+        # [s, b, 3 * head * head_dim]
+        seq_len, bs, _ = qkv.shape
+        if self.tp_size > 1:
+            qkv = tensor_model_parallel_all_gather(qkv)
+
+        # [s, b, 3 * head * head_dim] -> 3 * [s, b, head * head_dim]
+        q, k, v = qkv.chunk(3, dim=2)
+
+        # 3 * [s, b, head * head_dim]
+        if self.tp_size > 1:
+            splitter = partial(dist_utils.split_tensor_along_last_dim,
+                               num_partitions=self.tp_size)
+            q = splitter(q)[self.tp_rank]
+            k = splitter(k)[self.tp_rank]
+            v = splitter(v)[self.tp_rank]
+
+        # 3 * [s, b, head * head_dim] -> 3 * [s, b, head, head_dim]
+        new_shape = (seq_len, bs, self.num_attention_heads_per_partition,
+                     self.hidden_size_per_attention_head)
+        q, k, v = (x.view(*new_shape) for x in (q, k, v))
+        return q, k, v
+
     def forward(
         self,
         x: torch.Tensor,
         cu_seqlens: torch.Tensor,
         rotary_pos_emb: torch.Tensor,
     ) -> torch.Tensor:
-        # [s, b, c] --> [s, b, head * 3 * head_dim]
+
+        # [s, b, c] --> [s, b, 3 * head * head_dim]
         x, _ = self.qkv(x)
 
-        # [s, b, head * 3 * head_dim] --> [s, b, head, 3 * head_dim]
+        # [s, b, 3 * head * head_dim] -> 3 * [s, b, head, head_dim]
-        new_x_shape = x.size()[:-1] + (
+        q, k, v = self.split_qkv(x)
-            self.num_attention_heads_per_partition,
-            3 * self.hidden_size_per_attention_head,
-        )
-        x = x.view(*new_x_shape)
-
-        # [s, b, head, 3 * head_dim] --> 3 [s, b, head, head_dim]
-        q, k, v = dist_utils.split_tensor_along_last_dim(x, 3)
         batch_size = q.shape[1]
 
         q, k, v = (rearrange(x, "s b ... -> b s ...").contiguous()
@@ -614,24 +633,6 @@ class Qwen2VisionTransformer(nn.Module):
                 weight_loader(param, loaded_weight, shard_id)
                 break
             else:
-                if name.endswith("qkv.weight"):
-                    visual_num_heads = self.num_heads
-                    visual_embed_dim = self.embed_dim
-                    head_size = visual_embed_dim // visual_num_heads
-                    loaded_weight = loaded_weight.view(3, visual_num_heads,
-                                                       head_size,
-                                                       visual_embed_dim)
-                    loaded_weight = loaded_weight.transpose(0, 1)
-                    loaded_weight = loaded_weight.reshape(-1, visual_embed_dim)
-                elif name.endswith("qkv.bias"):
-                    visual_num_heads = self.num_heads
-                    visual_embed_dim = self.embed_dim
-                    head_size = visual_embed_dim // visual_num_heads
-                    loaded_weight = loaded_weight.view(3, visual_num_heads,
-                                                       head_size)
-                    loaded_weight = loaded_weight.transpose(0, 1)
-                    loaded_weight = loaded_weight.reshape(-1)
-
                 param = params_dict[name]
                 weight_loader = getattr(param, "weight_loader",
                                         default_weight_loader)
@@ -935,6 +936,16 @@ class Qwen2VLForConditionalGeneration(nn.Module, SupportsMultiModal,
     embedding_modules = {}
     embedding_padding_modules = []
 
+    # 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),
+    }
+
     # To ensure correct weight loading and mapping.
     hf_to_vllm_mapper = WeightsMapper(orig_to_new_prefix={
         "lm_head.": "language_model.lm_head.",