Skip to content

Revise README and fix typo #4

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Revise README and fix typo #4

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 25 additions & 28 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,63 +38,59 @@

## Introduction


We are excited to announce the official release of DeepSeek-V3.2-Exp, an experimental version of our model. As an intermediate step toward our next-generation architecture, V3.2-Exp builds upon V3.1-Terminus by introducing DeepSeek Sparse Attention—a sparse attention mechanism designed to explore and validate optimizations for training and inference efficiency in long-context scenarios.

This experimental release represents our ongoing research into more efficient transformer architectures, particularly focusing on improving computational efficiency when processing extended text sequences.

<div align="center">
<img src="cost.jpg" >
<img src="cost.jpg" alt="Cost Analysis">
</div>

- DeepSeek Sparse Attention (DSA) achieves fine-grained sparse attention for the first time, delivering substantial improvements in long-context training and inference efficiency while maintaining virtually identical model output quality.


- To rigorously evaluate the impact of introducing sparse attention, we deliberately aligned the training configurations of DeepSeek-V3.2-Exp with V3.1-Terminus. Across public benchmarks in various domains, DeepSeek-V3.2-Exp demonstrates performance on par with V3.1-Terminus.


| Benchmark | DeepSeek-V3.1-Terminus | DeepSeek-V3.2-Exp |
| :--- | :---: | :---: |
| **Reasoning Mode w/o Tool Use** | | |
| MMLU-Pro | 85.0 | 85.0 |
| GPQA-Diamond | 80.7 | 79.9 |
| Humanity's Last Exam | 21.7 | 19.8 |
| LiveCodeBench | 74.9 | 74.1 |
| AIME 2025 | 88.4 | 89.3 |
| HMMT 2025 | 86.1 | 83.6 |
| Codeforces | 2046 | 2121 |
| Aider-Polyglot | 76.1 | 74.5 |
| **Agentic Tool Use** | | |
| BrowseComp | 38.5 | 40.1 |
| BrowseComp-zh | 45.0 | 47.9 |
| SimpleQA | 96.8 | 97.1 |
| SWE Verified | 68.4 | 67.8 |
| SWE-bench Multilingual | 57.8 | 57.9 |
| Terminal-bench | 36.7 | 37.7 |


| Benchmark | DeepSeek-V3.1-Terminus | DeepSeek-V3.2-Exp |
| :------------------------------ | :--------------------: | :---------------: |
| **Reasoning Mode w/o Tool Use** | | |
| MMLU-Pro | 85.0 | 85.0 |
| GPQA-Diamond | 80.7 | 79.9 |
| Humanity's Last Exam | 21.7 | 19.8 |
| LiveCodeBench | 74.9 | 74.1 |
| AIME 2025 | 88.4 | 89.3 |
| HMMT 2025 | 86.1 | 83.6 |
| Codeforces | 2046 | 2121 |
| Aider-Polyglot | 76.1 | 74.5 |
| **Agentic Tool Use** | | |
| BrowseComp | 38.5 | 40.1 |
| BrowseComp-zh | 45.0 | 47.9 |
| SimpleQA | 96.8 | 97.1 |
| SWE Verified | 68.4 | 67.8 |
| SWE-bench Multilingual | 57.8 | 57.9 |
| Terminal-bench | 36.7 | 37.7 |

## Open-Source Kernels

For TileLang kernels with **better readability and research-purpose design**, please refer to [TileLang](https://github.com/tile-ai/tilelang/tree/main/examples/deepseek-v32).

For **high-performance CUDA kernels**, indexer logit kernels (including paged versions) are available in [DeepGEMM](https://github.com/deepseek-ai/DeepGEMM/pull/200). Sparse attention kernels are released in [FlashMLA](https://github.com/deepseek-ai/FlashMLA/pull/98).



## How to Run Locally

### HuggingFace

We provide an updated inference demo code in the [inference](https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Exp/tree/main/inference) folder to help the community quickly get started with our model and understand its architectural details.

First convert huggingface model weights to the the format required by our inference demo. Set `MP` to match your available GPU count:
First convert HuggingFace model weights to the format required by our inference demo. Set `MP` to match your available GPU count:

```bash
cd inference
export EXPERTS=256
python convert.py --hf-ckpt-path ${HF_CKPT_PATH} --save-path ${SAVE_PATH} --n-experts ${EXPERTS} --model-parallel ${MP}
```

Launch the interactive chat interface and start exploring DeepSeek's capabilities:

```bash
export CONFIG=config_671B_v3.2.json
torchrun --nproc-per-node ${MP} generate.py --ckpt-path ${SAVE_PATH} --config ${CONFIG} --interactive
Expand All @@ -117,6 +113,7 @@ docker pull lmsysorg/sglang:dsv32-a3
```

#### Launch Command

```bash
python -m sglang.launch_server --model deepseek-ai/DeepSeek-V3.2-Exp --tp 8 --dp 8 --page-size 64
```
Expand All @@ -141,4 +138,4 @@ This repository and the model weights are licensed under the [MIT License](LICEN

## Contact

If you have any questions, please raise an issue or contact us at [service@deepseek.com](service@deepseek.com).
If you have any questions, please raise an issue or contact us at [service@deepseek.com](mailto:service@deepseek.com).