Query and Response Augmentation Cannot Help Out-of-domain Math Reasoning Generalization
Paper • 2310.05506 • Published • 1
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 "OFA-Sys/MuggleMath_7B" \
--host 0.0.0.0 \
--port 30000# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
-H "Content-Type: application/json" \
--data '{
"model": "OFA-Sys/MuggleMath_7B",
"prompt": "Once upon a time,",
"max_tokens": 512,
"temperature": 0.5
}'see our paper in: https://arxiv.org/abs/2310.05506
MuggleMATH is fully fine-tuned on the AugGSM8K and AugMATH datasets and based on the LLaMA-2 Models.
prompting template: ''' "Below is an instruction that describes a task. " "Write a response that appropriately completes the request.\n\n" "### Instruction:\n{instruction}\n\n### Response:" ''' We recommend using vllm to accelerate inference.
| GSM8K | MATH | |
|---|---|---|
| MuggleMATH-7B | 69.8 | 25.8 |
| MuggleMATH-13B | 74.3 | 30.7 |
| MuggleMATH-70B | 82.5 | 42.1 |
@misc{li2023query, title={Query and Response Augmentation Cannot Help Out-of-domain Math Reasoning Generalization}, author={Chengpeng Li and Zheng Yuan and Hongyi Yuan and Guanting Dong and Keming Lu and Jiancan Wu and Chuanqi Tan and Xiang Wang and Chang Zhou}, journal={arXiv preprint arXiv: 2310.05506}, year={2023} }
Install from pip and serve model
# Install SGLang from pip: pip install sglang# Start the SGLang server: python3 -m sglang.launch_server \ --model-path "OFA-Sys/MuggleMath_7B" \ --host 0.0.0.0 \ --port 30000# Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "OFA-Sys/MuggleMath_7B", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'