Draft:DeepSeek AI
 | This is a draft article. It is a work in progress open to editing by anyone. Please ensure core content policies are met before publishing it as a live Wikipedia article. Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL Last edited by Citation bot (talk | contribs) 4 months ago. (Update)
Finished drafting? or |
Deepseek
[edit]Deepseek is a Artificial Intelligence Company founded in 2023. It is a Chinese company dedicated to making AGI a reality. They made headline when they dropped the DeepSeek-R1-Lite-Preview Large Language Model (LLM) with reasoning capability like the o1-preview model by OpenAI. Deepseek Chinese AI Research Lab backed by High-Flyer Hedge-Fund. It is the largest quantitative funds in China.
The CEO and Founder of Deepseek AI Research Lab is Liang Wenfeng. Before Deepseek, Liang Wenfeng was involved with High-Flyer Hedge Fund. Under his leadership, Deepseek has focused on foundational AI Technologies. Deepseek is fully funded by High-Flyer and has no plan to fundraise. Deepseek focuses on building foundational technology rather than commercial applications and has committed to open sourcing all of its models.
According to Liang Wenfeng the CEO of Deepseek states in an Interview with China Talk Media that they will not change to "Closed Source" unlike OpenAI. He also claimed that "Money has never been the problem for them, bans on shipments of advanced chips are the problem."[1]
List of Deep Seek AI Models
[edit]Deepseek offer various AI Models designed for various purposes.[1] Most notable once are as follows:
- Deepseek-LLM[2]
- This model generates human-like text and engaging in context-aware dialogues, making it great for chatbots and customer-service applications.
- Deepseek-V2.5
- Parameters: 236 Billion Parameter
- This model is suitable for General language understanding and coding.
- It is Specializes in mathematics, reasoning and coding tasks.
- It Support Context length up to 128K tokens.
- Deepseek-Coder[3]
- Specializes for autocomplete functions in coding environments.
- Deepseek Math[4]
- Specialized in mathematical tasks.
- DeepSeek VL (Vision-Language)[5]
- Designed for tasks that requires understanding both Text and Visual Information to Answer.
- It is a Multi-Modal Large Language Model.
- DeepSeek-R1-Lite-Preview
- It is the reasoning AI Models.
- It is the first "open source" reasoning model.
- It excel in complex tasks—particularly in mathematics and coding—reportedly matching or even surpassing OpenAI’s o1-preview on tough benchmarks like AIME and MATH1.
DeepSeek-R1-Lite-Preview
[edit]Deepseek-R1-Lite-Preview Model introduces "chain-of-thought" reasoning by putting a bunch of Compute. It also reveals inference Scaling Law: "Longer Reasoning Results in Better Performance".
Deepseek-R1-Lite-Preview model is available on DeepSeek Chat Interface, where users can access the models and Test it. Note that usage is currently limited to 50 messages per day. Deepseek thought has to release open-source versions of its R1 models.[6]
References
[edit]- https://www.chinatalk.media/p/deepseek-ceo-interview-with-chinas
- https://www.deepseek.com/
- https://www.datacamp.com/blog/deepseek-r1-lite-preview
- https://www.gadgets360.com/ai/news/deepseek-r1-lite-preview-china-ai-model-advanced-reasoning-rival-openai-o1-7072569
- https://analyticsindiamag.com/ai-news-updates/deepseek-launches-r1-lite-preview-outperforms-openais-o1-model/
- https://huggingface.co/deepseek-ai
- ^ Schneider, Jordan. "Deepseek: The Quiet Giant Leading China's AI Race". www.chinatalk.media. Retrieved 2024-11-28.
- ^ deepseek-ai/DeepSeek-LLM, DeepSeek, 2024-11-27, retrieved 2024-11-28
- ^ deepseek-ai/DeepSeek-Coder, DeepSeek, 2024-11-28, retrieved 2024-11-28
- ^ deepseek-ai/DeepSeek-Math, DeepSeek, 2024-11-28, retrieved 2024-11-28
- ^ deepseek-ai/DeepSeek-VL, DeepSeek, 2024-11-27, retrieved 2024-11-28
- ^ Jindal, Siddharth (2024-11-21). "DeepSeek Launches R1-Lite-Preview, Outperforms OpenAI's o1 Model". Analytics India Magazine. Retrieved 2024-11-28.