Instructions to use naver-hyperclovax/HyperCLOVAX-SEED-Think-32B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use naver-hyperclovax/HyperCLOVAX-SEED-Think-32B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="naver-hyperclovax/HyperCLOVAX-SEED-Think-32B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("naver-hyperclovax/HyperCLOVAX-SEED-Think-32B", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use naver-hyperclovax/HyperCLOVAX-SEED-Think-32B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "naver-hyperclovax/HyperCLOVAX-SEED-Think-32B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "naver-hyperclovax/HyperCLOVAX-SEED-Think-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/naver-hyperclovax/HyperCLOVAX-SEED-Think-32B

SGLang

How to use naver-hyperclovax/HyperCLOVAX-SEED-Think-32B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "naver-hyperclovax/HyperCLOVAX-SEED-Think-32B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "naver-hyperclovax/HyperCLOVAX-SEED-Think-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "naver-hyperclovax/HyperCLOVAX-SEED-Think-32B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "naver-hyperclovax/HyperCLOVAX-SEED-Think-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use naver-hyperclovax/HyperCLOVAX-SEED-Think-32B with Docker Model Runner:
```
docker model run hf.co/naver-hyperclovax/HyperCLOVAX-SEED-Think-32B
```

[Notice] Upcoming BC-break: model_type, chat_template.jinja

#13

by bigshanedogg - opened 8 days ago

Discussion

bigshanedogg

8 days ago

[Notice] Upcoming BC-break: model_type, class naming, and processor updates

We're planning to push an update to this repository that introduces a backward-incompatible change as part of the integration with transformers (PR #44314).

What's changing

model_type in config.json: "vlm" → "hyperclovax_vision_v2"
Updated jinja chat template
(optional) Class naming: HCX* → HyperCLOVAX*
(optional) Processor structure: split into image_processing_*.py / video_processing_*.py, unified via processing_*.py

Impact

Code relying on the current model_type, class names, or processor layout will need to be updated after this change lands.

⚠️ Temporary compatibility note

Between the moment the new changes are pushed to this repo and the moment they're merged into a released version of transformers, there may be a short window during which loading this model with trust_remote_code=True is not fully compatible.

If you run into issues during this period, you can continue using the previous version by downloading the repo at an earlier commit hash and pointing to it via cache_dir. For example:

from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="<repo-id>",
    revision="<previous-commit-hash>",
    cache_dir="<your-cache-dir>",
)

We'll keep this window as short as possible — the plan is to apply the update within roughly a day after this notice.

Why

This aligns the repo with the official transformers integration and the standard naming/structure used across other vision-language models on the Hub.

Thank you for your patience, and please feel free to leave any questions or concerns below. We'll post another update once the change is merged.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment