Update README.md
Browse files
README.md
CHANGED
|
@@ -6,6 +6,7 @@ tags:
|
|
| 6 |
- generated_from_trainer
|
| 7 |
- trl
|
| 8 |
- grpo
|
|
|
|
| 9 |
licence: license
|
| 10 |
license: apache-2.0
|
| 11 |
language:
|
|
@@ -23,7 +24,7 @@ It has been trained using [TRL](https://github.com/huggingface/trl).
|
|
| 23 |
|
| 24 |
|
| 25 |
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
| 26 |
-
|
| 27 |
### Framework versions
|
| 28 |
|
| 29 |
- TRL: 0.19.0
|
|
@@ -31,6 +32,7 @@ This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing
|
|
| 31 |
- Pytorch: 2.6.0+cu124
|
| 32 |
- Datasets: 3.6.0
|
| 33 |
- Tokenizers: 0.21.2
|
|
|
|
| 34 |
|
| 35 |
## Citations
|
| 36 |
|
|
|
|
| 6 |
- generated_from_trainer
|
| 7 |
- trl
|
| 8 |
- grpo
|
| 9 |
+
- rpt
|
| 10 |
licence: license
|
| 11 |
license: apache-2.0
|
| 12 |
language:
|
|
|
|
| 24 |
|
| 25 |
|
| 26 |
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
|
| 27 |
+
The GRPO training followed the methodology introduced in RPT (Reinforcement Pre-Training Paper ): https://huggingface.co/papers/2506.08007
|
| 28 |
### Framework versions
|
| 29 |
|
| 30 |
- TRL: 0.19.0
|
|
|
|
| 32 |
- Pytorch: 2.6.0+cu124
|
| 33 |
- Datasets: 3.6.0
|
| 34 |
- Tokenizers: 0.21.2
|
| 35 |
+
- vllm: 0.8.5.post1
|
| 36 |
|
| 37 |
## Citations
|
| 38 |
|