Instructions to use Siyuc/INFUSER-Qwen3-8B-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Siyuc/INFUSER-Qwen3-8B-base with Transformers:

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

pipe = pipeline("text-generation", model="Siyuc/INFUSER-Qwen3-8B-base")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Siyuc/INFUSER-Qwen3-8B-base")
model = AutoModelForCausalLM.from_pretrained("Siyuc/INFUSER-Qwen3-8B-base")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Siyuc/INFUSER-Qwen3-8B-base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Siyuc/INFUSER-Qwen3-8B-base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Siyuc/INFUSER-Qwen3-8B-base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Siyuc/INFUSER-Qwen3-8B-base

SGLang

How to use Siyuc/INFUSER-Qwen3-8B-base 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 "Siyuc/INFUSER-Qwen3-8B-base" \
    --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": "Siyuc/INFUSER-Qwen3-8B-base",
		"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 "Siyuc/INFUSER-Qwen3-8B-base" \
        --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": "Siyuc/INFUSER-Qwen3-8B-base",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Siyuc/INFUSER-Qwen3-8B-base with Docker Model Runner:
```
docker model run hf.co/Siyuc/INFUSER-Qwen3-8B-base
```

INFUSER-Qwen3-8B-base

This repository contains the Hugging Face export for an INFUSER-trained checkpoint based on Qwen/Qwen3-8B-Base.

Summary

Base model: Qwen/Qwen3-8B-Base
Method: INFUSER
Code repository: https://github.com/FFishy-git/INFUSER
Data repository: Siyuc/infuser-data

Evaluation

Released Checkpoint Scores

Category	Benchmark	Score
General	MMLU-Pro	67.81%
General	GPQA-Diamond	47.47%
General	SuperGPQA	38.86%
General	BBEH	12.51%
Math & physics	MATH500	84.25%
Math & physics	AIME2024	19.06%
Math & physics	AIME2025	18.02%
Math & physics	HMMT	9.64%
Math & physics	OlympiadBench (Math)	54.45%
Math & physics	OlympiadBench (Phys)	14.41%
Medical	MedQA	66.46%
Medical	MedXpertQA	14.57%
Coding	HumanEval+	78.86%
Coding	LiveCodeBench v1-5	28.47%

Comparison Summary

Category and overall means are computed over the same benchmark groups. R-Few (paper) and SPICE (paper) are self-reported values from their original papers, so missing categories are shown as -.

Category	This model	INFUSER avg	Base	R-Zero	AZR	R-Few (paper)	SPICE (paper)	General-Reasoner
General reasoning	41.66%	40.62%	34.43%	37.14%	37.61%	38.88%	38.75%	41.40%
Math & physics reasoning	33.30%	31.49%	26.08%	28.46%	30.28%	-	-	29.24%
Medical	40.52%	40.52%	39.34%	40.17%	39.89%	-	-	40.96%
Coding	53.66%	53.29%	50.59%	52.55%	53.18%	-	-	52.78%
Overall (14 benchmarks)	39.63%	38.50%	33.86%	36.05%	37.02%	-	-	37.75%

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("Siyuc/INFUSER-Qwen3-8B-base")
tokenizer = AutoTokenizer.from_pretrained("Siyuc/INFUSER-Qwen3-8B-base")

Downloads last month: 77

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for Siyuc/INFUSER-Qwen3-8B-base

Base model

Qwen/Qwen3-8B-Base

Finetuned

(441)

this model