[Model] Support InternS2 Preview (#42705)

Signed-off-by: Isotr0py <[email protected]>
Co-authored-by: zxy <[email protected]>

docs/models/supported_models.md

@@ -575,6 +575,7 @@ These models primarily accept the [`LLM.generate`](./generative_models.md#llmgen
 | `IsaacForConditionalGeneration` | Isaac | T + I<sup>+</sup> | `PerceptronAI/Isaac-0.1` | ✅︎ | ✅︎ |
 | `InternS1ForConditionalGeneration` | Intern-S1 | T + I<sup>E+</sup> + V<sup>E+</sup> | `internlm/Intern-S1`, `internlm/Intern-S1-mini`, etc. | ✅︎ | ✅︎ |
 | `InternS1ProForConditionalGeneration` | Intern-S1-Pro | T + I<sup>E+</sup> + V<sup>E+</sup> | `internlm/Intern-S1-Pro`, etc. | ✅︎ | ✅︎ |
+| `InternS2PreviewForConditionalGeneration` | Intern-S2-Preview | T + I<sup>E+</sup> + V<sup>E+</sup> | `internlm/Intern-S2-Preview`, etc. | ✅︎ | ✅︎ |
 | `InternVLChatModel` | InternVL 3.5, InternVL 3.0, InternVideo 2.5, InternVL 2.5, Mono-InternVL, InternVL 2.0 | T + I<sup>E+</sup> + (V<sup>E+</sup>) | `OpenGVLab/InternVL3_5-14B`, `OpenGVLab/InternVL3-9B`, `OpenGVLab/InternVideo2_5_Chat_8B`, `OpenGVLab/InternVL2_5-4B`, `OpenGVLab/Mono-InternVL-2B`, `OpenGVLab/InternVL2-4B`, etc. | ✅︎ | ✅︎ |
 | `InternVLForConditionalGeneration` | InternVL 3.0 (HF format) | T + I<sup>E+</sup> + V<sup>E+</sup> | `OpenGVLab/InternVL3-1B-hf`, etc. | ✅︎ | ✅︎ |
 | `KananaVForConditionalGeneration` | Kanana-V | T + I<sup>+</sup> | `kakaocorp/kanana-1.5-v-3b-instruct`, etc. | | ✅︎ |

tests/models/registry.py

@@ -988,6 +988,11 @@ _MULTIMODAL_EXAMPLE_MODELS = {
         "internlm/Intern-S1-Pro",
         trust_remote_code=True,
     ),
+    "InternS2PreviewForConditionalGeneration": _HfExamplesInfo(
+        "internlm/Intern-S2-Preview",
+        trust_remote_code=True,
+        is_available_online=False,
+    ),
     "InternVLChatModel": _HfExamplesInfo(
         "OpenGVLab/InternVL2-1B",
         extras={

vllm/config/speculative.py

@@ -467,6 +467,17 @@ class SpeculativeConfig:
                     "architectures": ["Qwen3_5MoeMTP" if is_moe else "Qwen3_5MTP"],
                 }
             )
+        if hf_config.model_type == "intern_s2_preview":
+            text_config = getattr(hf_config, "text_config", None)
+            is_moe = getattr(text_config, "model_type", None) == "qwen3_5_moe_text"
+            hf_config.model_type = "qwen3_5_mtp"
+            n_predict = getattr(text_config, "mtp_num_hidden_layers", None)
+            hf_config.update(
+                {
+                    "n_predict": n_predict,
+                    "architectures": ["Qwen3_5MoeMTP" if is_moe else "Qwen3_5MTP"],
+                }
+            )
         if hf_config.model_type == "longcat_flash":
             hf_config.model_type = "longcat_flash_mtp"
             n_predict = getattr(hf_config, "num_nextn_predict_layers", 1)

vllm/model_executor/models/interns2_preview.py

@@ -0,0 +1,38 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+from collections.abc import Iterable
+
+import torch
+from transformers import AutoProcessor
+
+from vllm.multimodal import MULTIMODAL_REGISTRY
+
+from .qwen3_5 import Qwen3_5MoeForConditionalGeneration
+from .qwen3_vl import (
+    Qwen3VLDummyInputsBuilder,
+    Qwen3VLMultiModalProcessor,
+    Qwen3VLProcessingInfo,
+)
+from .utils import AutoWeightsLoader
+
+
+class InternS2PreviewProcessingInfo(Qwen3VLProcessingInfo):
+    def get_hf_config(self):
+        return self.ctx.get_hf_config()
+
+    def get_hf_processor(self, **kwargs: object) -> AutoProcessor:
+        return self.ctx.get_hf_processor(**kwargs)
+
+
+@MULTIMODAL_REGISTRY.register_processor(
+    Qwen3VLMultiModalProcessor,
+    info=InternS2PreviewProcessingInfo,
+    dummy_inputs=Qwen3VLDummyInputsBuilder,
+)
+class InternS2PreviewForConditionalGeneration(Qwen3_5MoeForConditionalGeneration):
+    def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]) -> set[str]:
+        loader = AutoWeightsLoader(
+            self,
+            skip_prefixes=["mtp.", "model.time_series.", "time_series."],
+        )
+        return loader.load_weights(weights, mapper=self.hf_to_vllm_mapper)

vllm/model_executor/models/registry.py

@@ -435,6 +435,10 @@ _MULTIMODAL_MODELS = {
         "interns1_pro",
         "InternS1ProForConditionalGeneration",
     ),
+    "InternS2PreviewForConditionalGeneration": (
+        "interns2_preview",
+        "InternS2PreviewForConditionalGeneration",
+    ),
     "Idefics3ForConditionalGeneration": (
         "idefics3",
         "Idefics3ForConditionalGeneration",

vllm/v1/spec_decode/llm_base_proposer.py

@@ -1218,6 +1218,7 @@ class SpecDecodeBaseProposer:
                 "Exaone4_5_ForConditionalGeneration",
                 "GlmOcrForConditionalGeneration",
                 "HunYuanVLForConditionalGeneration",
+                "InternS2PreviewForConditionalGeneration",
                 "MiMoV2OmniForCausalLM",
                 "Qwen2_5_VLForConditionalGeneration",
                 "Qwen3_5ForConditionalGeneration",