Instructions to use jwg0830/AX-3.1-Light-sft_declare_v1_fix with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jwg0830/AX-3.1-Light-sft_declare_v1_fix with Transformers:

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

pipe = pipeline("text-generation", model="jwg0830/AX-3.1-Light-sft_declare_v1_fix")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("jwg0830/AX-3.1-Light-sft_declare_v1_fix")
model = AutoModelForCausalLM.from_pretrained("jwg0830/AX-3.1-Light-sft_declare_v1_fix", device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use jwg0830/AX-3.1-Light-sft_declare_v1_fix with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jwg0830/AX-3.1-Light-sft_declare_v1_fix"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jwg0830/AX-3.1-Light-sft_declare_v1_fix",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/jwg0830/AX-3.1-Light-sft_declare_v1_fix

SGLang

How to use jwg0830/AX-3.1-Light-sft_declare_v1_fix 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 "jwg0830/AX-3.1-Light-sft_declare_v1_fix" \
    --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": "jwg0830/AX-3.1-Light-sft_declare_v1_fix",
		"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 "jwg0830/AX-3.1-Light-sft_declare_v1_fix" \
        --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": "jwg0830/AX-3.1-Light-sft_declare_v1_fix",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use jwg0830/AX-3.1-Light-sft_declare_v1_fix with Docker Model Runner:
```
docker model run hf.co/jwg0830/AX-3.1-Light-sft_declare_v1_fix
```

AX-3.1-Light-sft_declare_v1_fix / README.md

jwg0830

declare_v1 fixed merge (single-GPU, no NaN)

163ed21 verified about 1 month ago

preview code

Raw History Blame Contribute Delete

1.94 kB

	---
	base_model: skt/A.X-3.1-Light
	library_name: transformers
	language:
	- ko
	license: other
	tags:
	- lora-merged
	- korean
	- k-ai-leaderboard
	---

	# AX-3.1-Light-sft_declare_v1 (동전)

	`skt/A.X-3.1-Light` 기반, v0.21과 동일한 5,801건 데이터로 LoRA SFT한 모델입니다.
	v0.21 대비 target을 자연스러운 선언형(`"정답은 X입니다."`)으로 바꾸고, 프롬프트
	지시문 없이도 답을 선언하도록 데이터를 재구성했습니다. 로컬 벤치마크 지표 검증용
	실험 모델입니다.

	- 베이스 모델: `skt/A.X-3.1-Light`
	- 학습 방법: LoRA(r=16, alpha=32, q/k/v/o+gate/up/down proj), lr 5e-5, 1 epoch
	- 목적: 자유 프롬프트에서의 "정답 선언율"이 실제 K-AI 점수와 상관(Spearman +0.90,
	n=5)이 있다는 로컬 관찰을 검증하기 위한 실험 모델

	> 이전 업로드(`AX-3.1-Light-sft_declare_v1`)는 병합 과정의 결함으로 가중치에 NaN이
	> 섞여 정상 동작하지 않았습니다. 본 repo가 수정된 정상 아티팩트입니다.

	## 사용 데이터셋 (v0.21과 동일)

	\| ID \| 데이터셋명 \| URL \|
	\|---\|---\|---\|
	\| 71857 \| 국어 교과 지문형 문제 데이터 \| https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=realm&dataSetSn=71857 \|
	\| 71874 \| 전문 의학지식 데이터 \| https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=realm&dataSetSn=71874 \|
	\| 71610 \| 금융, 법률 문서 기계독해 데이터 \| https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=realm&dataSetSn=71610 \|
	\| 569 \| 행정 문서 대상 기계독해 데이터 \| https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=realm&dataSetSn=569 \|
	\| 71949 \| 인과관계 기반 추론 데이터(업사이클링) \| https://www.aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=realm&dataSetSn=71949 \|