---
license: apache-2.0
task_categories:
- text-generation
tags:
- math
- code
- on-policy-distillation
---

# A2-OPD Dataset

This dataset accompanies the paper [Rethinking Relative Policy Differences Transfer of On-Policy Distillation](https://huggingface.co/papers/2610.13205).

Code: https://github.com/KnowledgeXLab/A2-OPD

The dataset is organized by `math/` and `code/` folders and contains training and evaluation data for the A2-OPD experiments.

## Dataset Structure

| Path | Split | Used for |
|------|-------|----------|
| `math/train.parquet` | train | DeepMath-103K training |
| `math/test.parquet` | test | Training validation and offline math eval |
| `code/train.parquet` | train | Eurus-RL-Code training |
| `code/validation.parquet` | validation | Training validation during code RL |
| `code/benchmarks/Humaneval_Plus/HumanEvalPlus.jsonl` | test | Offline HumanEval+ |
| `code/benchmarks/Mbpp_Plus/MbppPlus.jsonl` | test | Offline MBPP+ |
| `code/benchmarks/LivecodeBench/test.jsonl` | test | Offline LiveCodeBench |

Training data must be in `verl` parquet format. Each row needs fields such as `data_source`, `prompt`, `ability`, `reward_model.ground_truth`, and optional `extra_info`.

## Sample Usage

```bash
# One-click download (requires huggingface_hub: `hf` or `huggingface-cli`)
hf download KnowledgeXLab/A2-OPD --repo-type dataset --local-dir data
export A2OPD_DATA_ROOT=$PWD/data   # default layout matches training/eval scripts
```