xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-10 07:15:01 +08:00
Code Issues Packages Projects Releases Wiki Activity

[LoRA]: Add lora support to qwen-2.5-omni (#24231)

Browse Source
This commit is contained in:
Yash Pratap Singh 2025-09-04 18:20:50 +05:30 committed by GitHub
parent 16ded21eeb
commit c9f7081f9c
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 14 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
docs/models/supported_models.md
View File

@ -662,7 +662,7 @@ These models primarily accept the [`LLM.generate`](./generative_models.md#llmgen
| `Qwen2AudioForConditionalGeneration` | Qwen2-Audio | T + A<sup>+</sup> | `Qwen/Qwen2-Audio-7B-Instruct` | | ✅︎ | ✅︎ |
| `Qwen2VLForConditionalGeneration` | QVQ, Qwen2-VL | T + I<sup>E+</sup> + V<sup>E+</sup> | `Qwen/QVQ-72B-Preview`, `Qwen/Qwen2-VL-7B-Instruct`, `Qwen/Qwen2-VL-72B-Instruct`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Qwen2_5_VLForConditionalGeneration` | Qwen2.5-VL | T + I<sup>E+</sup> + V<sup>E+</sup> | `Qwen/Qwen2.5-VL-3B-Instruct`, `Qwen/Qwen2.5-VL-72B-Instruct`, etc. | ✅︎ | ✅︎ | ✅︎ |
| `Qwen2_5OmniThinkerForConditionalGeneration` | Qwen2.5-Omni | T + I<sup>E+</sup> + V<sup>E+</sup> + A<sup>+</sup> | `Qwen/Qwen2.5-Omni-7B` | | ✅︎ | ✅︎ |
| `Qwen2_5OmniThinkerForConditionalGeneration` | Qwen2.5-Omni | T + I<sup>E+</sup> + V<sup>E+</sup> + A<sup>+</sup> | `Qwen/Qwen2.5-Omni-3B`, `Qwen/Qwen2.5-Omni-7B` | ✅︎ | ✅︎ | ✅︎ |
| `RForConditionalGeneration` | R-VL-4B | T + I<sup>E+</sup> | `YannQi/R-4B` | | ✅︎ | ✅︎ |
| `SkyworkR1VChatModel` | Skywork-R1V-38B | T + I | `Skywork/Skywork-R1V-38B` | | ✅︎ | ✅︎ |
| `SmolVLMForConditionalGeneration` | SmolVLM2 | T + I | `SmolVLM2-2.2B-Instruct` | ✅︎ | | ✅︎ |

15
vllm/model_executor/models/qwen2_5_omni_thinker.py
View File

@ -41,6 +41,7 @@ from transformers.models.whisper import WhisperFeatureExtractor
from vllm.config import VllmConfig
from vllm.logger import init_logger
from vllm.model_executor.layers.rotary_embedding import MRotaryEmbedding
from vllm.model_executor.models.module_mapping import MultiModelKeys
from vllm.model_executor.models.qwen2_5_vl import (
Qwen2_5_VisionTransformer, Qwen2_5_VLImageEmbeddingInputs,
Qwen2_5_VLImageInputs, Qwen2_5_VLImagePixelInputs,
@ -66,7 +67,8 @@ from vllm.multimodal.profiling import BaseDummyInputsBuilder
from vllm.sequence import IntermediateTensors
from vllm.transformers_utils.tokenizer import decode_tokens, encode_tokens
from .interfaces import MultiModalEmbeddings, SupportsMultiModal, SupportsPP
from .interfaces import (MultiModalEmbeddings, SupportsLoRA,
SupportsMultiModal, SupportsPP)
from .utils import (AutoWeightsLoader, WeightsMapper,
init_vllm_registered_model, maybe_prefix,
merge_multimodal_embeddings)
@ -705,7 +707,7 @@ class Qwen2_5OmniConditionalGenerationMixin:
dummy_inputs=Qwen2_5OmniThinkerDummyInputsBuilder,
)
class Qwen2_5OmniThinkerForConditionalGeneration(
nn.Module, SupportsMultiModal, SupportsPP,
nn.Module, SupportsMultiModal, SupportsPP, SupportsLoRA,
Qwen2_5OmniConditionalGenerationMixin):
hf_to_vllm_mapper = WeightsMapper(
orig_to_new_prefix={
@ -798,6 +800,15 @@ class Qwen2_5OmniThinkerForConditionalGeneration(
def get_language_model(self) -> torch.nn.Module:
return self.language_model
def get_mm_mapping(self) -> MultiModelKeys:
"""Get module prefix for multimodal models to filter LoRA modules."""
return MultiModelKeys.from_string_field(
language_model="language_model",
connector=[], # No explicit connector in this model
tower_model=["visual",
"audio_tower"], # Exclude vision and audio towers
)
def get_multimodal_embeddings(self,
**kwargs: object) -> MultiModalEmbeddings: