I'm thinking about using DeepSeek's open-source AI models in my current research and would like to hear from others who have done this.
- What challenges did you face during the integration process, and how did you overcome them?
- How have these models impacted your data analysis and research results?
- Are there any best practices or recommendations you can share for effectively incorporating DeepSeek's models into existing research workflows?
Your experiences will be very helpful to me and others considering this integration.
Thank you.
Hello, Ehsan. You can self-host the deep-seek models using Ollama.
1. Challenges: If the available GPU does not have sufficient VRAM, hosting the larger models will no longer be possible. In that case, smaller models are the only available option, which can create difficulties if the tasks are too complex. We can also use the model's API, which raises privacy and budget issues.
2. Qualitiativly, not that much. However, using them can significantly reduce time if the research workflow has no repetitive processes.
3. Follow the documentation. You can adjust related parameters/configurations later.
Best Regards.
- Badges
- Science topic
- Similar topics
- Comment