|
--- |
|
license: mit |
|
license_link: https://huggingface.co/OpenPipe/Deductive-Reasoning-Qwen-32B/blob/main/LICENSE |
|
language: |
|
- en |
|
pipeline_tag: text-generation |
|
base_model: |
|
- OpenPipe/Deductive-Reasoning-Qwen-32B |
|
tags: |
|
- chat |
|
library_name: transformers |
|
--- |
|
|
|
Credits to [@DavidAU](https://huggingface.co/OpenPipe/Deductive-Reasoning-Qwen-32B/discussions/4#67ce59237c6e6ea1cc54fc22) for his findings. |
|
|
|
Used llama.cpp-b4837 for quantization. |
|
|
|
Original model: [OpenPipe/Deductive-Reasoning-Qwen-32B](https://huggingface.co/OpenPipe/Deductive-Reasoning-Qwen-32B) |
|
|
|
# Deductive-Reasoning-Qwen-32B |
|
|
|
 |
|
|
|
Deductive Reasoning Qwen 32B is a reinforcement fine-tune of [Qwen 2.5 32B Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct) to solve challenging deduction problems from the [Temporal Clue](https://github.com/bradhilton/temporal-clue) dataset, trained by [OpenPipe](https://openpipe.ai)! |
|
|
|
Here are some additional resources to check out: |
|
|
|
- [Blog Post](https://openpipe.ai/blog/using-grpo-to-beat-o1-o3-mini-and-r1-on-temporal-clue) |
|
- [Training Recipe](https://github.com/openpipe/deductive-reasoning) |
|
- [RL Experiments](https://github.com/openpipe/rl-experiments) |
|
- [Deductive Reasoning Qwen 14B](https://huggingface.co/OpenPipe/Deductive-Reasoning-Qwen-14B) |
|
|
|
If you're interested in training your own models with reinforcement learning or just chatting, feel free to [reach out](https://openpipe.ai/contact) or email Kyle directly at kyle@openpipe.ai! |
