[Model] Add MiDashengLM model support (vllm-project#23652)

Signed-off-by: chenbing8 <[email protected]>
Signed-off-by: bingchen-mi <[email protected]>
Co-authored-by: Jee Jee Li <[email protected]>
Co-authored-by: Isotr0py <[email protected]>

Author: 3 people

docs/models/supported_models.md

@@ -643,6 +643,7 @@ These models primarily accept the [`LLM.generate`](./generative_models.md#llmgen
 | `LlavaNextForConditionalGeneration` | LLaVA-NeXT | T + I<sup>E+</sup> | `llava-hf/llava-v1.6-mistral-7b-hf`, `llava-hf/llava-v1.6-vicuna-7b-hf`, etc. | | ✅︎ | ✅︎ |
 | `LlavaNextVideoForConditionalGeneration` | LLaVA-NeXT-Video | T + V | `llava-hf/LLaVA-NeXT-Video-7B-hf`, etc. | | ✅︎ | ✅︎ |
 | `LlavaOnevisionForConditionalGeneration` | LLaVA-Onevision | T + I<sup>+</sup> + V<sup>+</sup> | `llava-hf/llava-onevision-qwen2-7b-ov-hf`, `llava-hf/llava-onevision-qwen2-0.5b-ov-hf`, etc. | | ✅︎ | ✅︎ |
+| `MiDashengLMModel` | MiDashengLM | T + A<sup>+</sup> | `mispeech/midashenglm-7b` | | ✅︎ | ✅︎ |
 | `MiniCPMO` | MiniCPM-O | T + I<sup>E+</sup> + V<sup>E+</sup> + A<sup>E+</sup> | `openbmb/MiniCPM-o-2_6`, etc. | ✅︎ | ✅︎ | ✅︎ |
 | `MiniCPMV` | MiniCPM-V | T + I<sup>E+</sup> + V<sup>E+</sup> | `openbmb/MiniCPM-V-2` (see note), `openbmb/MiniCPM-Llama3-V-2_5`, `openbmb/MiniCPM-V-2_6`, `openbmb/MiniCPM-V-4`, `openbmb/MiniCPM-V-4_5`, etc. | ✅︎ | | ✅︎ |
 | `MiniMaxVL01ForConditionalGeneration` | MiniMax-VL | T + I<sup>E+</sup> | `MiniMaxAI/MiniMax-VL-01`, etc. | | ✅︎ | ✅︎ |

examples/offline_inference/audio_language.py

@@ -146,6 +146,36 @@ def run_granite_speech(question: str, audio_count: int) -> ModelRequestData:
     )
 
 
+# MiDashengLM
+def run_midashenglm(question: str, audio_count: int):
+    model_name = "mispeech/midashenglm-7b"
+
+    engine_args = EngineArgs(
+        model=model_name,
+        trust_remote_code=True,
+        max_model_len=4096,
+        max_num_seqs=5,
+        limit_mm_per_prompt={"audio": audio_count},
+    )
+
+    audio_in_prompt = "".join(
+        ["<|audio_bos|><|AUDIO|><|audio_eos|>" for idx in range(audio_count)]
+    )
+
+    default_system = "You are a helpful language and speech assistant."
+
+    prompt = (
+        f"<|im_start|>system\n{default_system}<|im_end|>\n"
+        "<|im_start|>user\n"
+        f"{audio_in_prompt}{question}<|im_end|>\n"
+        "<|im_start|>assistant\n"
+    )
+    return ModelRequestData(
+        engine_args=engine_args,
+        prompt=prompt,
+    )
+
+
 # MiniCPM-O
 def run_minicpmo(question: str, audio_count: int) -> ModelRequestData:
     model_name = "openbmb/MiniCPM-o-2_6"
@@ -352,6 +382,7 @@ def run_whisper(question: str, audio_count: int) -> ModelRequestData:
     "voxtral": run_voxtral,
     "gemma3n": run_gemma3n,
     "granite_speech": run_granite_speech,
+    "midashenglm": run_midashenglm,
     "minicpmo": run_minicpmo,
     "phi4_mm": run_phi4mm,
     "phi4_multimodal": run_phi4_multimodal,

tests/models/multimodal/processing/test_common.py

@@ -302,6 +302,7 @@ def _test_processing_correctness_one(
     "llava-hf/llava-onevision-qwen2-0.5b-ov-hf",
     "meta-llama/Llama-3.2-11B-Vision-Instruct",
     "TIGER-Lab/Mantis-8B-siglip-llama3",
+    "mispeech/midashenglm-7b",
     "openbmb/MiniCPM-Llama3-V-2_5",
     "openbmb/MiniCPM-o-2_6",
     "openbmb/MiniCPM-V-2_6",

tests/models/registry.py

@@ -460,6 +460,8 @@ def check_available_online(
                                                       max_transformers_version="4.48",  # noqa: E501
                                                       transformers_version_reason="HF model is not compatible.",  # noqa: E501
                                                       hf_overrides={"architectures": ["MantisForConditionalGeneration"]}),  # noqa: E501
+    "MiDashengLMModel": _HfExamplesInfo("mispeech/midashenglm-7b",
+                            trust_remote_code=True),
     "MiniCPMO": _HfExamplesInfo("openbmb/MiniCPM-o-2_6",
                                 trust_remote_code=True),
     "MiniCPMV": _HfExamplesInfo("openbmb/MiniCPM-Llama3-V-2_5",