biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

Implicit language-model-only mode via limit-mm-per-prompt (#22299)

Browse Source 
Signed-off-by: Roger Wang <hey@rogerw.me>
Signed-off-by: Andy Xie <andy.xning@gmail.com>
Signed-off-by: tjtanaa <tunjian.tan@embeddedllm.com>
Signed-off-by: Andrew Sansom <andrew@protopia.ai>
Signed-off-by: Zhiyu Cheng <zhiyuc@nvidia.com>
Signed-off-by: Shu Wang <shuw@nvidia.com>
Signed-off-by: Po-Han Huang <pohanh@nvidia.com>
Signed-off-by: Shu Wang. <shuw@nvidia.com>
Signed-off-by: XIn Li <xinli@nvidia.com>
Signed-off-by: Junhao Li <junhao@ubicloud.com>
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Signed-off-by: zRzRzRzRzRzRzR <2448370773@qq.com>
Signed-off-by: zitian.zhao <zitian.zhao@tencentmusic.com>
Signed-off-by: zitian zhao <zitian.zhao@tencentmusic.com>
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
Signed-off-by: iAmir97 <Amir.balwel@embeddedllm.com>
Signed-off-by: iAmir97 <71513472+iAmir97@users.noreply.github.com>
Signed-off-by: Linkun <github@lkchen.net>
Co-authored-by: Ning Xie <andy.xning@gmail.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
Co-authored-by: Andrew Sansom <andrew@protopia.ai>
Co-authored-by: Zhiyu <zhiyuc@nvidia.com>
Co-authored-by: Shu Wang <shuw@nvidia.com>
Co-authored-by: XIn Li <xinli@nvidia.com>
Co-authored-by: Junhao Li <streaver91@gmail.com>
Co-authored-by: Chauncey <chaunceyjiang@gmail.com>
Co-authored-by: Yuxuan Zhang <2448370773@qq.com>
Co-authored-by: ZiTian Zhao <zitian.zhao@tencentmusic.com>
Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Co-authored-by: Po-Han Huang (NVIDIA) <53919306+nvpohanh@users.noreply.github.com>
Co-authored-by: iAmir97 <71513472+iAmir97@users.noreply.github.com>
Co-authored-by: iAmir97 <Amir.balwel@embeddedllm.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
Co-authored-by: Hong Hanh <hanh.usth@gmail.com>
Co-authored-by: youkaichao <youkaichao@gmail.com>
Co-authored-by: lkchen <github@lkchen.net>
This commit is contained in:
Roger Wang
2025-08-08 22:21:40 -07:00
committed by GitHub
parent 429e4e2d42
commit 08b751ba74
16 changed files with 271 additions and 116 deletions
Download Patch File Download Diff File Expand all files Collapse all files

38
vllm/model_executor/models/mistral3.py
View File

@@ -428,20 +428,24 @@ class Mistral3ForConditionalGeneration(nn.Module, SupportsLoRA,
config.projector_hidden_act = "gelu"
# TODO: Optionally initializes this for supporting embeddings.
self.vision_tower = init_vision_tower_for_llava(
config,
quant_config,
require_post_norm=False,
prefix=maybe_prefix(prefix, "vision_tower"))
self.multi_modal_projector = Mistral3MultiModalProjector(
vision_hidden_size=config.vision_config.hidden_size,
text_hidden_size=config.text_config.hidden_size,
projector_hidden_act=config.projector_hidden_act,
spatial_merge_size=config.spatial_merge_size,
patch_size=config.vision_config.patch_size,
multimodal_projector_bias=config.multimodal_projector_bias,
quant_config=quant_config,
prefix=maybe_prefix(prefix, "multi_modal_projector"))
if multimodal_config.get_limit_per_prompt("image"):
self.vision_tower = init_vision_tower_for_llava(
config,
quant_config,
require_post_norm=False,
prefix=maybe_prefix(prefix, "vision_tower"))
self.multi_modal_projector = Mistral3MultiModalProjector(
vision_hidden_size=config.vision_config.hidden_size,
text_hidden_size=config.text_config.hidden_size,
projector_hidden_act=config.projector_hidden_act,
spatial_merge_size=config.spatial_merge_size,
patch_size=config.vision_config.patch_size,
multimodal_projector_bias=config.multimodal_projector_bias,
quant_config=quant_config,
prefix=maybe_prefix(prefix, "multi_modal_projector"))
else:
self.vision_tower = None
self.multi_modal_projector = None
self.language_model = init_vllm_registered_model(
vllm_config=vllm_config,
@@ -611,7 +615,11 @@ class Mistral3ForConditionalGeneration(nn.Module, SupportsLoRA,
def load_weights(self, weights: Iterable[tuple[str,
torch.Tensor]]) -> set[str]:
loader = AutoWeightsLoader(self)
skip_prefixes = []
if self.vision_tower is None and self.multi_modal_projector is None:
skip_prefixes = ["vision_tower.", "multi_modal_projector."]
loader = AutoWeightsLoader(self, skip_prefixes=skip_prefixes)
return loader.load_weights(weights, mapper=self.hf_to_vllm_mapper)
def get_mm_mapping(self) -> MultiModelKeys: