[feat] Support EAGLE for Qwen2

tests/models/registry.py

@@ -558,6 +558,10 @@ def check_available_online(
         trust_remote_code=True,
         speculative_model="morgendave/EAGLE-Llama-4-Scout-17B-16E-Instruct",
         tokenizer="meta-llama/Llama-4-Scout-17B-16E-Instruct"),  # noqa: E501
+    "Qwen2ForCausalLMEagle": _HfExamplesInfo("Qwen/QwQ-32B",
+                                             trust_remote_code=True,
+                                             speculative_model="reinforce20001/QwQ-32B-Eagle",
+                                             tokenizer="Qwen/QwQ-32B"),
     "EagleMiniCPMForCausalLM": _HfExamplesInfo("openbmb/MiniCPM-1B-sft-bf16",
                                             trust_remote_code=True,
                                             is_available_online=False,

vllm/model_executor/models/qwen2.py

@@ -52,7 +52,7 @@
 from vllm.sequence import IntermediateTensors
 from vllm.transformers_utils.config import is_interleaved
 
-from .interfaces import SupportsLoRA, SupportsPP
+from .interfaces import SupportsEagle3, SupportsLoRA, SupportsPP
 from .utils import (AutoWeightsLoader, PPMissingLayer, extract_layer_index,
                     is_pp_missing_parameter,
                     make_empty_intermediate_tensors_factory, make_layers,
@@ -442,7 +442,7 @@ def load_weights(self, weights: Iterable[tuple[str,
         return loaded_params
 
 
-class Qwen2ForCausalLM(nn.Module, SupportsLoRA, SupportsPP):
+class Qwen2ForCausalLM(nn.Module, SupportsLoRA, SupportsPP, SupportsEagle3):
     packed_modules_mapping = {
         "qkv_proj": [
             "q_proj",
@@ -499,6 +499,13 @@ def forward(
                                    inputs_embeds)
         return hidden_states
 
+    def set_aux_hidden_state_layers(self, layers: tuple[int]) -> None:
+        self.model.aux_hidden_state_layers = layers
+
+    def get_eagle3_aux_hidden_state_layers(self) -> tuple[int]:
+        num_layers = len(self.model.layers)
+        return (2, num_layers // 2, num_layers - 3)
+
     def compute_logits(
         self,
         hidden_states: torch.Tensor,

vllm/model_executor/models/qwen2_eagle.py

@@ -0,0 +1,172 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
+from collections.abc import Iterable
+from typing import Optional
+
+import torch
+import torch.nn as nn
+from transformers import Qwen2Config
+
+from vllm.compilation.decorators import support_torch_compile
+from vllm.config import VllmConfig
+from vllm.distributed.parallel_state import get_pp_group
+from vllm.logger import init_logger
+from vllm.model_executor.layers.logits_processor import LogitsProcessor
+from vllm.model_executor.layers.vocab_parallel_embedding import (
+    VocabParallelEmbedding)
+from vllm.model_executor.model_loader.weight_utils import default_weight_loader
+from vllm.model_executor.models.qwen2 import (Qwen2DecoderLayer,
+                                              Qwen2ForCausalLM)
+
+from .utils import AutoWeightsLoader, maybe_prefix
+
+logger = init_logger(__name__)
+
+
+class Qwen2DecoderLayer(Qwen2DecoderLayer):
+
+    def __init__(
+        self,
+        config: Qwen2Config,
+        disable_input_layernorm: bool,
+        prefix: str = "",
+    ) -> None:
+        super().__init__(config, prefix=prefix)
+
+        # Skip the input_layernorm
+        # https://github.com/SafeAILab/EAGLE/blob/35c78f6cdc19a73e05cf5c330b4c358dad970c6a/eagle/model/cnets.py#L427
+        if disable_input_layernorm:
+            del self.input_layernorm
+            self.input_layernorm = nn.Identity()
+
+
+@support_torch_compile
+class Qwen2Model(nn.Module):
+
+    def __init__(
+        self,
+        *,
+        vllm_config: VllmConfig,
+        prefix: str = "",
+        start_layer_id: int = 0,
+    ) -> None:
+        super().__init__()
+        self.config = vllm_config. \
+            speculative_config.draft_model_config.hf_config
+        self.vocab_size = self.config.vocab_size
+
+        self.embed_tokens = VocabParallelEmbedding(
+            self.config.vocab_size,
+            self.config.hidden_size,
+            prefix=maybe_prefix(prefix, "embed_tokens"),
+        )
+
+        self.layers = nn.ModuleList([
+            Qwen2DecoderLayer(
+                self.config,
+                i == 0,
+                prefix=maybe_prefix(prefix, f"layers.{i + start_layer_id}"),
+            ) for i in range(self.config.num_hidden_layers)
+        ])
+        self.fc = torch.nn.Linear(self.config.hidden_size * 2,
+                                  self.config.hidden_size,
+                                  bias=False)
+
+    def forward(
+        self,
+        input_ids: torch.Tensor,
+        positions: torch.Tensor,
+        hidden_states: torch.Tensor,
+    ) -> tuple[torch.Tensor, torch.Tensor]:
+        input_embeds = self.embed_tokens(input_ids)
+        hidden_states = self.fc(
+            torch.cat((input_embeds, hidden_states), dim=-1))
+        residual = None
+        for layer in self.layers:
+            hidden_states, residual = layer(
+                positions,
+                hidden_states,
+                residual,
+            )
+        hidden_states = hidden_states + residual
+        return hidden_states, hidden_states
+
+    def load_weights(self, weights: Iterable[tuple[str,
+                                                   torch.Tensor]]) -> set[str]:
+        stacked_params_mapping = [
+            # (param_name, shard_name, shard_id)
+            ("qkv_proj", "q_proj", "q"),
+            ("qkv_proj", "k_proj", "k"),
+            ("qkv_proj", "v_proj", "v"),
+            ("gate_up_proj", "gate_proj", 0),
+            ("gate_up_proj", "up_proj", 1),
+        ]
+        params_dict = dict(self.named_parameters())
+        loaded_params: set[str] = set()
+        for name, loaded_weight in weights:
+            for param_name, weight_name, shard_id in stacked_params_mapping:
+                if weight_name not in name:
+                    continue
+                name = name.replace(weight_name, param_name)
+                param = params_dict[name]
+                weight_loader = param.weight_loader
+                weight_loader(param, loaded_weight, shard_id)
+                break
+            else:
+
+                # if PP disabled then draft will share embed with target
+                if get_pp_group().world_size == 1 and \
+                    "embed_tokens." in name:
+                    continue
+
+                param = params_dict[name]
+                weight_loader = getattr(param, "weight_loader",
+                                        default_weight_loader)
+                weight_loader(param, loaded_weight)
+            loaded_params.add(name)
+        return loaded_params
+
+
+class Qwen2ForCausalLMEagle(Qwen2ForCausalLM):
+
+    def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
+        nn.Module.__init__(self)
+        self.config = vllm_config. \
+            speculative_config.draft_model_config.hf_config
+        target_layer_num = vllm_config.model_config.get_num_layers(
+            vllm_config.parallel_config)
+        self.model = Qwen2Model(vllm_config=vllm_config,
+                                prefix="model",
+                                start_layer_id=target_layer_num)
+
+        logit_scale = getattr(self.config, "logit_scale", 1.0)
+        self.logits_processor = LogitsProcessor(self.config.vocab_size,
+                                                scale=logit_scale)
+
+    def forward(
+        self,
+        input_ids: torch.Tensor,
+        positions: torch.Tensor,
+        hidden_states: torch.Tensor,
+        inputs_embeds: Optional[torch.Tensor] = None,
+    ) -> tuple[torch.Tensor, torch.Tensor]:
+        if inputs_embeds is not None:
+            raise NotImplementedError(
+                f"{type(self).__name__} does not support multimodal inputs yet."
+            )
+        return self.model(input_ids, positions, hidden_states)
+
+    def load_weights(self, weights: Iterable[tuple[str, torch.Tensor]]):
+        loader = AutoWeightsLoader(
+            self,
+            skip_prefixes=None,
+        )
+
+        model_weights = {}
+        for name, loaded_weight in weights:
+            if "lm_head" not in name:
+                name = "model." + name
+            model_weights[name] = loaded_weight
+        return loader.load_weights(model_weights.items())

vllm/model_executor/models/registry.py

@@ -265,6 +265,7 @@
     "EagleLlama4ForCausalLM": ("llama4_eagle", "EagleLlama4ForCausalLM"),
     "EagleMiniCPMForCausalLM": ("minicpm_eagle", "EagleMiniCPMForCausalLM"),
     "Eagle3LlamaForCausalLM": ("llama_eagle3", "Eagle3LlamaForCausalLM"),
+    "Qwen2ForCausalLMEagle": ("qwen2_eagle", "Qwen2ForCausalLMEagle"),
     # TODO: Re-enable this once tests/models/test_initialization.py is fixed, see PR #22333 #22611  # noqa: E501
     # "LlamaForCausalLMEagle3": ("llama_eagle3", "Eagle3LlamaForCausalLM"),
     "EagleDeepSeekMTPModel": ("deepseek_eagle", "EagleDeepseekV3ForCausalLM"),

vllm/transformers_utils/configs/eagle.py

@@ -50,8 +50,8 @@ def __init__(self,
             assert self.model is not None, \
                 "model should not be None when method is eagle"
             kwargs["architectures"] = [
-                f"Eagle{arch}" if not arch.startswith("Eagle") \
-                    else arch for arch in self.model.architectures
+                arch if arch.startswith("Eagle") or arch.endswith("Eagle") else
+                f"Eagle{arch}" for arch in self.model.architectures
             ]
         elif method == "eagle3":
             assert self.model is not None, \