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

[VLM] merged multimodal processor and V1 support for idefics3 (#12660)

Browse Source 
Signed-off-by: Isotr0py <2037008807@qq.com>
Co-authored-by: Cyrus Leung <cyrus.tl.leung@gmail.com>
This commit is contained in:
Isotr0py
2025-02-04 20:00:51 +08:00
committed by GitHub
parent 18a88fcccc
commit 815079de8e
7 changed files with 320 additions and 462 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/models/decoder_only/vision_language/test_models.py
View File

@@ -254,14 +254,14 @@ VLM_TEST_SETTINGS = {
patch_hf_runner=model_utils.h2ovl_patch_hf_runner,
),
"idefics3": VLMTestInfo(
models=["HuggingFaceM4/Idefics3-8B-Llama3"],
models=["HuggingFaceTB/SmolVLM-256M-Instruct"],
test_type=(VLMTestType.IMAGE, VLMTestType.MULTI_IMAGE),
prompt_formatter=lambda img_prompt:f"<|begin_of_text|>User:{img_prompt}<end_of_utterance>\nAssistant:", # noqa: E501
img_idx_to_prompt=lambda idx: "<image>",
max_model_len=8192,
max_num_seqs=2,
auto_cls=AutoModelForVision2Seq,
marks=[large_gpu_mark(min_gb=48)],
hf_output_post_proc=model_utils.idefics3_trunc_hf_output,
),
"intern_vl": VLMTestInfo(
models=[