We propose Rome, an ensemble approach grounded in parallelism and diversity, to enhance solution robustness by avoiding the pitfalls of single-solution models. The integration of these methods within a unified framework not only broadens the exploration space but also increases the likelihood of achieving more accurate and optimal solutions for complex mathematical problems.
pip install -r requirements.txt
## Terminal 1
CUDA_VISIBLE_DEVICES=0 python3 -m sglang.launch_server --port 7501 --model-path meta-llama/Llama-3.1-8B-Instruct
## Terminal 2
python3 src/main.py --problem "1 + 1 = ?" --methods tot mctsr
- Add more solvers
- Add more ensemble patterns
jwu323@ucmerced.edu
This project was made possible thanks to a collaboration with
