summaryrefslogtreecommitdiff
path: root/README.md
diff options
context:
space:
mode:
authorzitian-gao <zitian.gao@outlook.com>2025-05-27 17:03:03 +0800
committerzitian-gao <zitian.gao@outlook.com>2025-05-27 17:03:03 +0800
commit56af282a6177c43da8a639fa59bca51c9b1412dc (patch)
tree49adf9d6c8f3b3b1a0763999419c082c8875742f /README.md
parentd6703f54331b1c2d37c6ffdb91dc604062eee22b (diff)
update
Diffstat (limited to 'README.md')
-rw-r--r--README.md60
1 files changed, 59 insertions, 1 deletions
diff --git a/README.md b/README.md
index 7d3b0e6..3e1f8ea 100644
--- a/README.md
+++ b/README.md
@@ -1,3 +1,61 @@
## One-shot Entropy Minimization
-<a href='https://arxiv.org/abs/2505.20282'><img src='https://img.shields.io/badge/arXiv-2505.20282-b31b1b.svg'></a> &nbsp; \ No newline at end of file
+<a href='https://arxiv.org/abs/2505.20282'><img src='https://img.shields.io/badge/arXiv-2505.20282-b31b1b.svg'></a> &nbsp;
+
+### Overview
+
+This repository provides code and instructions to reproduce the results of the **One-shot Entropy Minimization (EM)** method presented in our paper. It includes both 1-shot and multi-shot training setups, as well as evaluation using the Qwen2.5-Math benchmark.
+
+---
+
+### Reproducing One-shot EM Training (SOTA)
+
+```bash
+accelerate launch train.py --lr 2e-5 --temperature 0.5 --bsz 64
+```
+
+---
+
+### Reproducing Multi-shot EM Training
+
+```bash
+accelerate launch train.py --lr 2e-5 --temperature 0.5 --bsz 64 --data_path "dataset/numina/numina_00.parquet"
+```
+
+---
+
+### Evaluation
+
+```bash
+cd Qwen2.5-Eval/evaluation
+bash sh/eval_all_math.sh
+```
+
+---
+
+### Acknowledgements
+
+Our dataset references and builds upon the following open-source contributions:
+
+- [NuminaMath-CoT](https://huggingface.co/datasets/AI-MO/NuminaMath-CoT)
+- [DeepScaler](https://github.com/agentica-project/deepscaler)
+- [One-shot RLVR](https://github.com/ypwang61/One-Shot-RLVR/) – for data selection strategies
+- [Qwen2.5-Eval](https://github.com/QwenLM/Qwen2.5-Math/) – for evaluation benchmarks
+
+We sincerely thank the authors and maintainers of these projects for their excellent contributions to the research community!
+
+
+---
+
+### Citation
+```
+@misc{gao2025oneshotentropyminimization,
+ title={One-shot Entropy Minimization},
+ author={Zitian Gao and Lynx Chen and Joey Zhou and Bryan Dai},
+ year={2025},
+ eprint={2505.20282},
+ archivePrefix={arXiv},
+ primaryClass={cs.CL},
+ url={https://arxiv.org/abs/2505.20282},
+}
+``` \ No newline at end of file