DeepSeek-V2
Due to the constraints of HuggingFace, the open-source code currently experiences slower performance than our internal codebase when running on GPUs with Huggingface.
text → text · made by DeepSeek
Sold by
Nobody in the catalogue publishes a price for this yet.
Measured 7 standings
| Place | Board | Metric | Score |
|---|---|---|---|
| 4thof 134 | Arc ai2 — Epoch AI | Challenge score | 0.922 |
| 6thof 113 | Hella swag — Epoch AI | Overall accuracy | 0.871 |
| 6thof 137 | Wino grande — Epoch AI | Accuracy | 86.3 |
| 8thof 113 | Piqa — Epoch AI | Score | 83.9 |
| 11thof 88 | Bbh — Epoch AI | Average | 0.788 |
| 28thof 115 | Trivia qa — Epoch AI | EM | 0.8 |
| 37thof 217 | Mmlu — Epoch AI | EM | 0.784 |
About
DeepSeek-V2 — a text model from DeepSeek, placed 4th of 134 on Arc ai2 — Epoch AI.
It takes text and returns text. It was published in April 2024. The catalogue files it under chat. Its weights are published under the other licence, which attaches conditions the plain open licences do not. It has been measured on 7 boards, and stands best at 4th of 134 on Arc ai2 — Epoch AI.
Every current figure
- Maker
- DeepSeek
- Register
- model
- Takes
- text
- Returns
- text
- Published
- April 2024
- Parameters
- 236 billion · read from its own weights
- Licence
- other
- Boards
- 7
- Best place
- 4th of 134 — Arc ai2 — Epoch AI
Known as 1 name
deepseek-ai/DeepSeek-V2