[Model] Add Kimi K3 support: model files and kernels [1/N] (#50089)
Signed-off-by: zjy0516 <riverclouds.zhu@qq.com> Signed-off-by: Isotr0py <Isotr0py@outlook.com> Signed-off-by: Bugen Zhao <i@bugenzhao.com> Signed-off-by: Yongye Zhu <zyy1102000@gmail.com> Co-authored-by: Isotr0py <Isotr0py@outlook.com> Co-authored-by: Thien Tran <gau.nernst@yahoo.com.sg> Co-authored-by: Bugen Zhao <i@bugenzhao.com> Co-authored-by: Yongye Zhu <zyy1102000@gmail.com> Co-authored-by: Ziming Huang <zelda.huanghuang@gmail.com> Co-authored-by: Roger Wang <hey@rogerw.io> Co-authored-by: Isotr0py <mozf@inferact.ai> Co-authored-by: aoshen02 <aoshen02@users.noreply.github.com> Co-authored-by: Woosuk Kwon <woosuk@inferact.ai> Co-authored-by: Jee Jee Li <jeejeelee@inferact.ai> Co-authored-by: aoshen02 <aoshen@inferact.ai> Co-authored-by: Summer Yang <girasoleyang@gmail.com> Co-authored-by: Kevin H. Luu <khluu000@gmail.com> Co-authored-by: Bowen Wang <abmfy@icloud.com> Co-authored-by: gnovack <novackgm@gmail.com> Co-authored-by: Nick Hill <nickhill123@gmail.com> Co-authored-by: Yifan Qiao <yifanqiao@inferact.ai> Co-authored-by: xiaozhoupy <peiyuanzhou1994@gmail.com> Co-authored-by: Roy Wang <yasong.wang@inferact.ai> Co-authored-by: Jeff (Junze) Ma <93145857+majunze2001@users.noreply.github.com> Co-authored-by: Codex <codex@openai.com>
This commit is contained in:
co-authored by
Isotr0py
Thien Tran
Bugen Zhao
Yongye Zhu
Ziming Huang
Roger Wang
Isotr0py
aoshen02
Woosuk Kwon
Jee Jee Li
aoshen02
Summer Yang
Kevin H. Luu
Bowen Wang
gnovack
Nick Hill
Yifan Qiao
xiaozhoupy
Roy Wang
Jeff Ma
Codex
parent
6f00a1ae3b
commit
7c6729b769
@@ -0,0 +1,61 @@
|
||||
# SPDX-License-Identifier: Apache-2.0
|
||||
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
||||
from transformers import BaseImageProcessor, BatchFeature, TensorType
|
||||
from transformers.processing_utils import ProcessorMixin
|
||||
|
||||
from vllm.tokenizers.hf import HfTokenizer
|
||||
|
||||
|
||||
class KimiK3Processor(ProcessorMixin):
|
||||
"""HF-style processor wrapper for the image-only Kimi-K3 model.
|
||||
|
||||
K3 exposes the standard ``image`` modality, so vLLM calls this processor
|
||||
with ``images=[PIL, ...]``. The underlying checkpoint image processor
|
||||
(``KimiK3VisionProcessor``) works on ``{"type": "image", "image": PIL}``
|
||||
media dicts, so this wrapper adapts bare PIL images into that shape before
|
||||
delegating to ``preprocess``.
|
||||
|
||||
Text is only tokenized here; the single ``<|kimi_image_placeholder|>``
|
||||
token per image is expanded into the resolution-aware media block by the
|
||||
model's ``_get_prompt_updates`` on the vLLM side.
|
||||
"""
|
||||
|
||||
attributes = ["image_processor", "tokenizer"]
|
||||
|
||||
def __init__(
|
||||
self,
|
||||
image_processor: BaseImageProcessor,
|
||||
tokenizer: HfTokenizer,
|
||||
) -> None:
|
||||
self.image_processor = image_processor
|
||||
self.tokenizer = tokenizer
|
||||
|
||||
def __call__(
|
||||
self,
|
||||
text: str | list[str] | None = None,
|
||||
images: object | list[object] | None = None,
|
||||
return_tensors: str | TensorType | None = None,
|
||||
**kwargs,
|
||||
) -> BatchFeature:
|
||||
if images is not None:
|
||||
if not isinstance(images, list):
|
||||
images = [images]
|
||||
medias = [{"type": "image", "image": image} for image in images]
|
||||
mm_inputs = self.image_processor.preprocess(
|
||||
medias,
|
||||
return_tensors=return_tensors,
|
||||
)
|
||||
else:
|
||||
mm_inputs = {}
|
||||
|
||||
if text is not None:
|
||||
if not isinstance(text, list):
|
||||
text = [text]
|
||||
text_inputs = self.tokenizer(text)
|
||||
else:
|
||||
text_inputs = {}
|
||||
|
||||
return BatchFeature(
|
||||
data={**text_inputs, **mm_inputs},
|
||||
tensor_type=return_tensors,
|
||||
)
|
||||
Reference in New Issue
Block a user