Instructions to use BAAI/Recon2Reason-Reasoning-4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use BAAI/Recon2Reason-Reasoning-4B with Transformers:

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

pipe = pipeline("image-text-to-text", model="BAAI/Recon2Reason-Reasoning-4B")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("BAAI/Recon2Reason-Reasoning-4B")
model = AutoModelForMultimodalLM.from_pretrained("BAAI/Recon2Reason-Reasoning-4B", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use BAAI/Recon2Reason-Reasoning-4B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "BAAI/Recon2Reason-Reasoning-4B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "BAAI/Recon2Reason-Reasoning-4B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/BAAI/Recon2Reason-Reasoning-4B

SGLang

How to use BAAI/Recon2Reason-Reasoning-4B 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 "BAAI/Recon2Reason-Reasoning-4B" \
    --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": "BAAI/Recon2Reason-Reasoning-4B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "BAAI/Recon2Reason-Reasoning-4B" \
        --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": "BAAI/Recon2Reason-Reasoning-4B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use BAAI/Recon2Reason-Reasoning-4B with Docker Model Runner:
```
docker model run hf.co/BAAI/Recon2Reason-Reasoning-4B
```

Recon2Reason-Reasoning-4B / verify_release.py

Magnefic

Add files using upload-large-folder tool

f8faf41 verified about 2 months ago

Raw History Blame Contribute Delete

3.14 kB

#!/usr/bin/env python3
"""Smoke-test a local or Hugging Face Recon2Reason Reasoning 4B repository."""

from __future__ import annotations

import argparse

import torch
from PIL import Image
from transformers import AutoProcessor, Qwen3VLForConditionalGeneration


EXPECTED_PARAMETERS = 4_437_815_808


def main() -> None:
    parser = argparse.ArgumentParser()
    parser.add_argument("model", help="Local directory or Hugging Face repository ID")
    parser.add_argument(
        "--device",
        default="cpu",
        help="Device map target used for loading, for example cpu or cuda:0",
    )
    parser.add_argument(
        "--generate",
        action="store_true",
        help="Also run a one-token generation test with an in-memory image",
    )
    args = parser.parse_args()

    processor = AutoProcessor.from_pretrained(args.model)
    model = Qwen3VLForConditionalGeneration.from_pretrained(
        args.model,
        dtype="auto",
        device_map={"": args.device},
        low_cpu_mem_usage=True,
    ).eval()

    parameter_count = sum(parameter.numel() for parameter in model.parameters())
    if parameter_count != EXPECTED_PARAMETERS:
        raise SystemExit(
            f"Parameter mismatch: expected {EXPECTED_PARAMETERS:,}, found {parameter_count:,}"
        )

    expected_tokens = {
        "<|feat|>": 151669,
        "<|/feat|>": 151670,
        "<|pos|>": 151671,
        "<|/pos|>": 151672,
        "<|feat_pad|>": 151673,
    }
    for token, expected_id in expected_tokens.items():
        actual_id = processor.tokenizer.convert_tokens_to_ids(token)
        if actual_id != expected_id:
            raise SystemExit(
                f"Token mismatch for {token}: expected {expected_id}, found {actual_id}"
            )

    print(f"Processor: {processor.__class__.__name__}")
    print(f"Model: {model.__class__.__name__}")
    print(f"Parameters: {parameter_count:,}")
    print(f"Dtype: {next(model.parameters()).dtype}")
    print("Special tokens: OK")

    if args.generate:
        messages = [
            {
                "role": "user",
                "content": [
                    {"type": "image", "image": Image.new("RGB", (224, 224), "white")},
                    {"type": "text", "text": "What color is this image?"},
                ],
            }
        ]
        inputs = processor.apply_chat_template(
            messages,
            tokenize=True,
            add_generation_prompt=True,
            return_dict=True,
            return_tensors="pt",
        ).to(args.device)
        with torch.inference_mode():
            generated_ids = model.generate(
                **inputs,
                max_new_tokens=1,
                do_sample=False,
            )
        prompt_length = inputs["input_ids"].shape[1]
        generated_text = processor.batch_decode(
            generated_ids[:, prompt_length:],
            skip_special_tokens=True,
            clean_up_tokenization_spaces=False,
        )[0]
        print(f"One-token multimodal generation: OK ({generated_text!r})")

    print("Release smoke test: PASSED")


if __name__ == "__main__":
    main()

	#!/usr/bin/env python3
	"""Smoke-test a local or Hugging Face Recon2Reason Reasoning 4B repository."""

	from __future__ import annotations

	import argparse

	import torch
	from PIL import Image
	from transformers import AutoProcessor, Qwen3VLForConditionalGeneration


	EXPECTED_PARAMETERS = 4_437_815_808


	def main() -> None:
	parser = argparse.ArgumentParser()
	parser.add_argument("model", help="Local directory or Hugging Face repository ID")
	parser.add_argument(
	"--device",
	default="cpu",
	help="Device map target used for loading, for example cpu or cuda:0",
	)
	parser.add_argument(
	"--generate",
	action="store_true",
	help="Also run a one-token generation test with an in-memory image",
	)
	args = parser.parse_args()

	processor = AutoProcessor.from_pretrained(args.model)
	model = Qwen3VLForConditionalGeneration.from_pretrained(
	args.model,
	dtype="auto",
	device_map={"": args.device},
	low_cpu_mem_usage=True,
	).eval()

	parameter_count = sum(parameter.numel() for parameter in model.parameters())
	if parameter_count != EXPECTED_PARAMETERS:
	raise SystemExit(
	f"Parameter mismatch: expected {EXPECTED_PARAMETERS:,}, found {parameter_count:,}"
	)

	expected_tokens = {
	"<\|feat\|>": 151669,
	"<\|/feat\|>": 151670,
	"<\|pos\|>": 151671,
	"<\|/pos\|>": 151672,
	"<\|feat_pad\|>": 151673,
	}
	for token, expected_id in expected_tokens.items():
	actual_id = processor.tokenizer.convert_tokens_to_ids(token)
	if actual_id != expected_id:
	raise SystemExit(
	f"Token mismatch for {token}: expected {expected_id}, found {actual_id}"
	)

	print(f"Processor: {processor.__class__.__name__}")
	print(f"Model: {model.__class__.__name__}")
	print(f"Parameters: {parameter_count:,}")
	print(f"Dtype: {next(model.parameters()).dtype}")
	print("Special tokens: OK")

	if args.generate:
	messages = [
	{
	"role": "user",
	"content": [
	{"type": "image", "image": Image.new("RGB", (224, 224), "white")},
	{"type": "text", "text": "What color is this image?"},
	],
	}
	]
	inputs = processor.apply_chat_template(
	messages,
	tokenize=True,
	add_generation_prompt=True,
	return_dict=True,
	return_tensors="pt",
	).to(args.device)
	with torch.inference_mode():
	generated_ids = model.generate(
	**inputs,
	max_new_tokens=1,
	do_sample=False,
	)
	prompt_length = inputs["input_ids"].shape[1]
	generated_text = processor.batch_decode(
	generated_ids[:, prompt_length:],
	skip_special_tokens=True,
	clean_up_tokenization_spaces=False,
	)[0]
	print(f"One-token multimodal generation: OK ({generated_text!r})")

	print("Release smoke test: PASSED")


	if __name__ == "__main__":
	main()