20231088/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

[V1] Update doc and examples for H2O-VL (#13349)

Browse Source 
Signed-off-by: Roger Wang <ywang@roblox.com>
This commit is contained in:
Roger Wang 2025-02-16 02:35:54 -08:00 committed by GitHub
parent dc0f7ccf8b
commit b7d309860e
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 6 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
docs/source/models/supported_models.md
View File

@ -726,7 +726,7 @@ See [this page](#generative-models) for more information on how to use generativ
* `h2oai/h2ovl-mississippi-800m`, `h2oai/h2ovl-mississippi-2b`, etc. * `h2oai/h2ovl-mississippi-800m`, `h2oai/h2ovl-mississippi-2b`, etc.
* *
* ✅︎ * ✅︎
* \* * ✅︎\*
- * `Idefics3ForConditionalGeneration` - * `Idefics3ForConditionalGeneration`
* Idefics3 * Idefics3
* T + I * T + I
@ -869,7 +869,7 @@ See [this page](#generative-models) for more information on how to use generativ
<sup>+</sup> Multiple items can be inputted per text prompt for this modality. <sup>+</sup> Multiple items can be inputted per text prompt for this modality.
:::{note} :::{note}
H2O-VL series models will be available in V1 once we support backends other than FlashAttention. `h2oai/h2ovl-mississippi-2b` will be available in V1 once we support backends other than FlashAttention.
::: :::
:::{note} :::{note}

4
examples/offline_inference/vision_language.py
View File

@ -119,7 +119,7 @@ def run_glm4v(question: str, modality: str):
def run_h2ovl(question: str, modality: str): def run_h2ovl(question: str, modality: str):
assert modality == "image" assert modality == "image"
model_name = "h2oai/h2ovl-mississippi-2b" model_name = "h2oai/h2ovl-mississippi-800m"
llm = LLM( llm = LLM(
model=model_name, model=model_name,
@ -136,7 +136,7 @@ def run_h2ovl(question: str, modality: str):
add_generation_prompt=True) add_generation_prompt=True)
# Stop tokens for H2OVL-Mississippi # Stop tokens for H2OVL-Mississippi
# https://huggingface.co/h2oai/h2ovl-mississippi-2b # https://huggingface.co/h2oai/h2ovl-mississippi-800m
stop_token_ids = [tokenizer.eos_token_id] stop_token_ids = [tokenizer.eos_token_id]
return llm, prompt, stop_token_ids return llm, prompt, stop_token_ids

5
examples/offline_inference/vision_language_multi_image.py
View File

@ -78,14 +78,13 @@ def load_deepseek_vl2(question: str, image_urls: List[str]):
def load_h2ovl(question: str, image_urls: List[str]) -> ModelRequestData: def load_h2ovl(question: str, image_urls: List[str]) -> ModelRequestData:
model_name = "h2oai/h2ovl-mississippi-2b" model_name = "h2oai/h2ovl-mississippi-800m"
llm = LLM( llm = LLM(
model=model_name, model=model_name,
trust_remote_code=True, trust_remote_code=True,
max_model_len=8192, max_model_len=8192,
limit_mm_per_prompt={"image": len(image_urls)}, limit_mm_per_prompt={"image": len(image_urls)},
mm_processor_kwargs={"max_dynamic_patch": 4},
) )
placeholders = "\n".join(f"Image-{i}: <image>\n" placeholders = "\n".join(f"Image-{i}: <image>\n"
@ -99,7 +98,7 @@ def load_h2ovl(question: str, image_urls: List[str]) -> ModelRequestData:
add_generation_prompt=True) add_generation_prompt=True)
# Stop tokens for H2OVL-Mississippi # Stop tokens for H2OVL-Mississippi
# https://huggingface.co/h2oai/h2ovl-mississippi-2b # https://huggingface.co/h2oai/h2ovl-mississippi-800m
stop_token_ids = [tokenizer.eos_token_id] stop_token_ids = [tokenizer.eos_token_id]
return ModelRequestData( return ModelRequestData(