Unlock the Power of LLM: Build a HuggingFace Agent
summary
Builds a working Hugging Face agent that chains real tools: uses Hugging Face Agents and OpenAI agent, requires a Hugging Face token and OpenAI key, and boots with a few pip installs and an init call. Demonstrates image captioning, translation (NLLB backend, ~80 tested languages though NLLB claims ~200), text-to-speech, document QA and Stable Diffusion flavors, and adds a custom Google TTS tool added to improve audio. Code runs in Colab, agents work but remain experimental and brittle for complex multi-step prompts.
AI-written summary · about
chapters (13)
- 0:00intro
- 0:20preamble
- 1:01agenda
- 1:35what are agents
- 3:35what are tools
- 4:46agents and tools
- 6:17why should we care
- 7:31hands-on: setup and initialize the agent
- 14:57what other predefined tools exist
- 19:20custom tool creation
- 25:25conclusion
- 26:48further ideas
- 27:41code and links
related talks
31mHugging Face ecosystem for Local AI/ ML
14mKeynote: Building AI with Open Source & Hugging Face- 14mKeynote: Building AI with Open Source & Hugging Face
33mMachine Learning 2.0 with Hugging Face
10mHow to Make Friends and Influence Users with Huggingface
34mExploring Hugging Face: A Hub for Machine Learning Models
48mUnlocking LLM Potential: From Exploration to Integration
19mBuilding Machine Learning Applications using Hugging Face