Results of the Cloud Native LLM Project with Kubermatic (Video and Report, AMOS Summer 2024 Project)
This project is one of seven Scrum projects with industry partners that were part of the AMOS Summer 2024 Projects. Below please find the video (you may also like the other videos) and the project summary which details the final result of the project. We run these projects every semester, so please be in touch if you would like to motivate one of your own!
Demo Video
Project Summary
| Project Name | Amos2024ss08-cloud-native-llm |
| Project Mission | TTo create a Large Language Model on Cloud Native Technologies our work was focused mainly on the following topics: 1. Automated Dataset Preparation: Develop an automated system that prepares structured datasets from CNCF landscape documentation. 2. LLM Fine-Tuning: Enhance a Large Language Model specifically tailored to the prepared dataset for precise and relevant responses. 3. Open-Source Contribution: Release and open-source the fine-tuned model and dataset preparation tools to the community, fostering collaboration and innovation. |
| Industry Partner | Kubermatic |
| Team Logo |  |
| Project Summary | DeepCNCF is a specialized Large Language Model designed to provide context-aware answers to questions related to Cloud Native technologies. To create this model, we collected data from various sources, including documentation sites on Cloud Native technologies and Stack Overflow. This data was then processed into a Q&A format. Using this curated dataset, we retrained the gemma2 model to improve its ability to answer questions specifically about Cloud Native technologies. Evaluations indicate that our best-trained model, with 9 billion parameters, significantly outperforms the original gemma2 model in this domain. |
| Project Illustration |   |
| Team Photo |  |
| Project Repository | https://github.com/amosproj/amos2024ss08-cloud-native-llm.git https://huggingface.co/Kubermatic |