24 lines
1.0 KiB
Markdown
24 lines
1.0 KiB
Markdown
|
# Hugging Face — reference material
|
||
|
|
|
||
|
|
*Placeholder.* This folder will cover Hugging Face: the de facto hub for open-weight AI models, datasets, and demos.
|
||
|
|
|
||
|
|
Planned topics:
|
||
|
|
|
||
|
|
- What Hugging Face is and why it matters (the "GitHub of AI models")
|
||
|
|
- Browsing the model hub: filtering by task, size, license
|
||
|
|
- Reading a model card: what to look for before you commit to a model
|
||
|
|
- Downloading and running a model with `transformers`
|
||
|
|
- The `pipeline()` shortcut for common tasks (text generation, transcription, classification, etc.)
|
||
|
|
- `sentence-transformers` for embeddings
|
||
|
|
- Where the weights actually live on your disk, and how to clean them up
|
||
|
|
- Datasets and Spaces (brief tour)
|
||
|
|
- Authentication and the few model families that need it (e.g. Llama)
|
||
|
|
|
||
|
|
## When to dip in
|
||
|
|
|
||
|
|
When your project needs a model — for transcription, summarization, image generation, embeddings, classification — and you'd rather run something locally than pay for an API.
|
||
|
|
|
||
|
|
## Prerequisites
|
||
|
|
|
||
|
|
Some [`../python/`](../python/), and ideally [`../pytorch/`](../pytorch/) if you want to peek under the hood.
|