Update README.md
Browse files
README.md
CHANGED
|
@@ -13,8 +13,6 @@ configs:
|
|
| 13 |
We annotate the entire [**Open Reasoner Zero**]((https://huggingface.co/Open-Reasoner-Zero/Open-Reasoner-Zero-7B)) dataset with a **difficulty score** based on the performance of the [Qwen 2.5-MATH-7B](https://huggingface.co/Qwen/Qwen2.5-Math-7B) model. This provides an adaptive signal for curriculum construction.
|
| 14 |
Open Reasoner Zero is a curated a dataset of 57,000 reasoning-intensive problems used to train and evaluate reinforcement learning-based methods for large language models.
|
| 15 |
|
| 16 |
-
Github: https://github.com/uscnlp-lime/verl
|
| 17 |
-
|
| 18 |
## Difficulty Scoring Method
|
| 19 |
|
| 20 |
Difficulty scores are estimated using the **Qwen 2.5-MATH-7B** model with the following generation settings:
|
|
@@ -48,6 +46,7 @@ We also apply the same difficulty estimation procedure to the following datasets
|
|
| 48 |
For questions or feedback, feel free to reach out to [**Taiwei Shi**](https://maksimstw.github.io/) at [[email protected]](mailto:[email protected]).
|
| 49 |
|
| 50 |
## 📚 Citations
|
|
|
|
| 51 |
|
| 52 |
If you find our dataset useful, please cite [Efficient Reinforcement Finetuning via Adaptive Curriculum Learning](https://huggingface.co/papers/2504.05520):
|
| 53 |
|
|
|
|
| 13 |
We annotate the entire [**Open Reasoner Zero**]((https://huggingface.co/Open-Reasoner-Zero/Open-Reasoner-Zero-7B)) dataset with a **difficulty score** based on the performance of the [Qwen 2.5-MATH-7B](https://huggingface.co/Qwen/Qwen2.5-Math-7B) model. This provides an adaptive signal for curriculum construction.
|
| 14 |
Open Reasoner Zero is a curated a dataset of 57,000 reasoning-intensive problems used to train and evaluate reinforcement learning-based methods for large language models.
|
| 15 |
|
|
|
|
|
|
|
| 16 |
## Difficulty Scoring Method
|
| 17 |
|
| 18 |
Difficulty scores are estimated using the **Qwen 2.5-MATH-7B** model with the following generation settings:
|
|
|
|
| 46 |
For questions or feedback, feel free to reach out to [**Taiwei Shi**](https://maksimstw.github.io/) at [[email protected]](mailto:[email protected]).
|
| 47 |
|
| 48 |
## 📚 Citations
|
| 49 |
+
Github: https://github.com/uscnlp-lime/verl
|
| 50 |
|
| 51 |
If you find our dataset useful, please cite [Efficient Reinforcement Finetuning via Adaptive Curriculum Learning](https://huggingface.co/papers/2504.05520):
|
| 52 |
|