Support custom multimodal dataset

vllm/benchmarks/datasets.py

@@ -1626,9 +1626,33 @@ class CustomDataset(BenchmarkDataset):
             raise NotImplementedError(
                 "Only JSONL format is supported for CustomDataset.")
 
-        random.seed(self.random_seed)
-        random.shuffle(self.data)
+        # random.seed(self.random_seed)
+        # random.shuffle(self.data)
 
+    def apply_multimodal_chat_transformation(
+            self,
+            prompt: str,
+            mm_content: Optional[
+                        Union[MultiModalDataDict, dict, list[dict]]
+                             ] = None) -> list[dict]:
+        """
+        Transform a prompt and optional multimodal content into a chat format.
+        This method is used for chat models that expect a specific conversation
+        format.
+        """
+        content = [{"text": prompt, "type": "text"}]
+        if mm_content is not None:
+            if isinstance(mm_content, list):
+                content.extend(cast(list[dict[str, Any]], mm_content))
+            elif isinstance(mm_content, dict):
+                content.append(mm_content)
+            else:
+                raise TypeError(
+                    "Could not process multimodal content of type: " +
+                    f"{type(mm_content)}"
+                )
+        return [{"role": "user", "content": content}]
+    
     def sample(
         self,
         tokenizer: PreTrainedTokenizerBase,
@@ -1655,24 +1679,16 @@ class CustomDataset(BenchmarkDataset):
             if len(sampled_requests) >= num_requests:
                 break
             prompt = item["prompt"]
-
-            # apply template
-            if not skip_chat_template:
-                prompt = tokenizer.apply_chat_template(
-                    [{
-                        "role": "user",
-                        "content": prompt
-                    }],
-                    add_generation_prompt=True,
-                    tokenize=False,
-                )
-
+ 
             prompt_len = len(tokenizer(prompt).input_ids)
+            mm_contents = item["image"]
+            prompt = self.apply_multimodal_chat_transformation(prompt, mm_contents)
+
             sampled_requests.append(
                 SampleRequest(
                     prompt=prompt,
                     prompt_len=prompt_len,
-                    expected_output_len=output_len,
+                    expected_output_len=item["expected_output_len"],
                     request_id=request_id_prefix + str(i),
                 ))
         self.maybe_oversample_requests(sampled_requests, num_requests,
@@ -2720,4 +2736,4 @@ class PrefixRepetitionRandomDataset(BenchmarkDataset):
                 )
 
         random.shuffle(requests)
-        return requests
+        return requests
\ No newline at end of file

vllm/benchmarks/lib/endpoint_request_func.py

@@ -250,12 +250,7 @@ async def async_request_openai_chat_completions(
         "model":
         request_func_input.model_name
         if request_func_input.model_name else request_func_input.model,
-        "messages": [
-            {
-                "role": "user",
-                "content": content
-            },
-        ],
+        "messages": request_func_input.prompt,
         "temperature":
         0.0,
         "max_completion_tokens":
@@ -281,7 +276,6 @@ async def async_request_openai_chat_completions(
 
     output = RequestFuncOutput()
     output.prompt_len = request_func_input.prompt_len
-
     generated_text = ""
     ttft = 0.0
     st = time.perf_counter()