LLaMA3 & TinyLlama Fine-Tuned for Function Calls
Ғылым және технология
In this video, I discuss my latest project on fine-tuning LLaMA3 and TinyLlama to natively support function calls, which is crucial for the development of AI agents in the open-source community.
I cover the following topics:
- Colab notebooks demonstrating how to run the models using helper - classes and GGFU versions
- Examples of using the models locally and with the Ollama server
- Prompt templates and usage guidelines
I also mention my plans for future improvements, such as multi-function detection, function binding, and fine-tuning models with less than 1B parameters.
This video provides an overview of my project and how you can start using these function-calling LLMs in your own projects.
Access the models, dataset, and repository through the links in the description. If you have any questions or want to contribute, feel free to open an issue on my GitHub repo: github.com/unclecode/callama
You can find models on HuggingFace: huggingface.co/unclecode
Follow me on Twitter (X) for updates on my research on function-calling for LLMs and AI agents: x.com/unclecode
I appreciate your feedback and thoughts on this project.
#OpenSourceAI #FunctionCalling #LLaMA3 #TinyLlama
Пікірлер: 10
Amazing, but is not available in LM Studio 😥😥 anyway, liked and subscribed. I should learn to use ollama.
@unclecode788
5 күн бұрын
You can load the GGFU version over there. Anyway, I'm working on a research project, and we're close to releasing what might be the smallest ever language model solely for function calls. Stay tuned!
Hi, The GGUF colab notebook cannot be accesed
@unclecode788
27 күн бұрын
It should be by now. Please check again
can it run on free google colab ram?
@unclecode788
25 күн бұрын
Yes, L4 would be good enough
Muito legal, ficou simples de compreender
@unclecode788
25 күн бұрын
de nada.
Very nice work but please make your code fullscreen, the video like this with dark screen and darkened symbosl is super hard to follow.
@unclecode788
25 күн бұрын
Haha you right, I will consider that for the next video, stay tuned