Update README.md
Browse files
README.md
CHANGED
|
@@ -43,10 +43,14 @@ PCL-Reasoner-V1.5 attains **90.9% on AIME 2024** and **85.7% on AIME 2025**, sig
|
|
| 43 |
|
| 44 |
We have fully open-sourced the **model weights**, **dataset**, and **training code** to foster transparency, reproducibility, and community innovation. Follow the tutorial below to deploy, evaluate, or extend PCL-Reasoner-V1.5 in your own research!
|
| 45 |
|
| 46 |
-
##
|
| 47 |
|
| 48 |
[GitHub Repository](https://github.com/PCL-Reasoner/V1.5)
|
| 49 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 50 |
|
| 51 |
## Evaluation
|
| 52 |
All results are reported using the **Avg@32 metric** (average accuracy over 32 independent sampling attempts per problem), ensuring robust and fair comparison.
|
|
|
|
| 43 |
|
| 44 |
We have fully open-sourced the **model weights**, **dataset**, and **training code** to foster transparency, reproducibility, and community innovation. Follow the tutorial below to deploy, evaluate, or extend PCL-Reasoner-V1.5 in your own research!
|
| 45 |
|
| 46 |
+
## Code
|
| 47 |
|
| 48 |
[GitHub Repository](https://github.com/PCL-Reasoner/V1.5)
|
| 49 |
|
| 50 |
+
## RL Dataset
|
| 51 |
+
|
| 52 |
+
[Huggingface Dataset](https://huggingface.co/datasets/PCL-Reasoner/V1.5-RL-Math)
|
| 53 |
+
|
| 54 |
|
| 55 |
## Evaluation
|
| 56 |
All results are reported using the **Avg@32 metric** (average accuracy over 32 independent sampling attempts per problem), ensuring robust and fair comparison.
|