https://github.com/oxidized-transformers/oxidized-transformers

Modular Rust transformer/LLM library using Candle
https://github.com/oxidized-transformers/oxidized-transformers

bert llama llama2 llm roberta rust transformers

Last synced: about 1 month ago
JSON representation

Modular Rust transformer/LLM library using Candle

• Host: GitHub
• URL: https://github.com/oxidized-transformers/oxidized-transformers
• Owner: oxidized-transformers
• License: other
• Created: 2023-08-08T17:41:02.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-05-05T15:41:32.000Z (7 months ago)
• Last Synced: 2024-09-30T23:33:21.019Z (about 2 months ago)
• Topics: bert, llama, llama2, llm, roberta, rust, transformers
• Language: Rust
• Homepage:
• Size: 142 KB
• Stars: 36
• Watchers: 6
• Forks: 3
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE-APACHE
  • Authors: AUTHORS

Awesome Lists containing this project

README

        
# Oxidized Transformers

Oxidized Transformers is a Rust transformers library that started out as

a port of [Curated Transformers](https://github.com/explosion/curated-transformers/tree/main).

The foundations are in place and some popular models are implemented, but

Oxidized Transformers is still too volatile to use in projects. Keep an eye

on the repo, since progress is currently fast.

## 🧰 Supported Model Architectures

Supported encoder-only models:

- ALBERT

- BERT

- RoBERTa

- XLM-RoBERTa

Supported decoder-only models:

- GPT-NeoX

- Llama 1/2

All types of models can be loaded from Huggingface Hub. Float16/bfloat16

models can use flash attention v2 on recent CUDA GPUs.