---
license: apache-2.0
---
# TinyDeepSeek: Reproduction of DeepSeek-R1 and Beyond

<p align="center">
   📃 <a href="" target="_blank">Paper</a> • 🤗 <a href="https://huggingface.co/FreedomIntelligence/TinyDeepSeek-0.5B-base" target="_blank">TinyDeepSeek-0.5B-base</a> • 🤗 <a href="https://huggingface.co/FreedomIntelligence/TinyDeepSeek-3.3B-base" target="_blank">TinyDeepSeek-3.3B-base</a>
   <br> 🤗 <a href="https://huggingface.co/FreedomIntelligence/TinyDeepSeek-3.3B-checkpoints" target="_blank"> TinyDeepSeek-3.3B-checkpoints</a> • 🤗 <a href="https://huggingface.co/FreedomIntelligence/TinyDeepSeek-0.5B-checkpoints" target="_blank">TinyDeepSeek-0.5B-checkpoints</a>

</p>

**Innovation and Open source are the best tributes and reproductions of DeepSeek**. 

The open-source ethos, rooted in technological equity, upholds two key principles: **free access for all developers** and **the opportunity for technical contributions**. While DeepSeek exemplifies the first principle, the second principle is hindered by restrictive training strategies, unclear data sources, and the high costs of model training. These constraints limit the open-source community's capacity to contribute and impede technological progress.

To overcome these challenges, we launched a comprehensive reproduction project of DeepSeek. This initiative involves training models from scratch, replicating DeepSeek's architecture and algorithms. We will **fully open-source** the training code, datasets, and models, offering code framework, reference solution, and base models for low-cost continual exploration.



## 🌈 Update

* **[2025.03.11]** TinyDeepSeek repo is published！🎉




## Reproduction of DeepSeek-R1

### Architecture

   <details><summary>Click to expand</summary>

   ![arch](assets/arch.png)

   As shown in above Figure, the DeepSeek technical report introduces three architectural designs:

   - A. *Multi-Head Latent Attention (MLA)*

   - B. *Load Balancing Strategy without Auxiliary Loss*: Please refer [code](https://github.com/FreedomIntelligence/TinyDeepSeek/blob/main/src/modeling/tinydeepseek/modeling_tinydeepseek.py#L550) for implementation.

   - C. *Multi-Token Prediction*: Please refer [code](https://github.com/FreedomIntelligence/TinyDeepSeek/blob/main/src/modeling/tinydeepseek/modeling_tinydeepseek.py#L1875) for implementation.

   Our Detailed architectural parameter are shown in Table below.

   ![arch](assets/iarch.png)

   </details>

### Data Construction

   <details><summary>Click to expand</summary>

   ![arch](assets/data.png)

   - Pretrain: 
      - Meta & Labeled Data: *TBD*
      - Process Code: Please refer to [Link](/pretrain_data/Readme.md) for implementation.
   - SFT: 
      - [HuggingFaceTB/smoltalk](https://huggingface.co/datasets/HuggingFaceTB/smoltalk)
      - [allenai/tulu-3-sft-olmo-2-mixture](https://huggingface.co/datasets/allenai/tulu-3-sft-olmo-2-mixture)
   - RL:
      - [SynthLabsAI/Big-Math-RL-Verified](https://huggingface.co/datasets/SynthLabsAI/Big-Math-RL-Verified)

   </details>

### Model Training

```
bash examples/pretrainStage1.sh
bash examples/pretrainStage2.sh
bash examples/sft.sh

bash examples/rl.sh (TBD)
```

### Results

*TBD*

## Beyond Reproduction

### Scale Up RL to Pretrain

> **RWO: Reward Weighted Optimization**

For the training, please include the include the following flags in the training command.

- For Pretrain please prepare data item with key 'text_evaluation':{"knowledge":4, "reasoning":3, ...}.
- For SFT, please provide data file 'General.json' and reward file 'General_reward.json' in the same directory. 

```
--reward_weighted_optimization True \
--remove_unused_columns False \
```

## 📃 To do 

- [ ] Release Evaluation Results
- [ ] Release RL Training Code and Model
- [ ] Release Pretrain Data quality annotation label
- [ ] Release TinyDeepSeek Technical Report


##  Acknowledgment

- [DeepSeek-R1](https://github.com/deepseek-ai/DeepSeek-R1)


##  Citation
Please use the following citation if you intend to use our dataset for training or evaluation:

```
@misc{tinydeepseek,
  title={TinyDeepSeek},
  author={FreedomIntelligence Team},
  year = {2025},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/FreedomIntelligence/TinyDeepSeek}},
}
```