transformers
Hugging Face Transformers for loading Hub models, running pipeline inference
by k-dense-aiFree
Description
transformers is an agent skill for Hugging Face Transformers, which gives access to thousands of pretrained models for NLP, computer vision, audio and multimodal tasks. It covers loading Hub models, running pipeline inference, text generation, and fine-tuning with the Trainer on custom data. The skill targets Transformers 5.18.0, verified against its released source on 1 October 2026, with native CPU checks on Python 3.11, Torch 2.14.1, Datasets 5.0.1, Accelerate 1.15.0, PEFT 0.21.2 and Hub 1.33.0. It recommends a dedicated environment and pins these packages together. It notes that Hub 2.1.1 is newer, but Datasets 5.0.1 requires a Hub version below 2, so upgrading each package on its own can make the training stack impossible to resolve. Optional packages are needed only for particular workflows, such as Pillow for images, librosa and soundfile for audio, FFmpeg for encoded audio files, Tesseract for OCR, and bitsandbytes for quantisation. The skill advises choosing precision or quantisation after reading its model-loading notes. The separate esm SDK currently requires Transformers below 5, so it belongs in another environment.…
Highlights
- Loads models from the Hugging Face Hub
- Pipeline inference and text generation
- Trainer fine-tuning
- NLP, vision, audio and multimodal tasks
- Pinned stack targeting Transformers 5.18.0
Ratings & Reviews
No ratings yet
Be the first to rate transformers.
People also visit
transformers FAQ
What can it do?
Load Hub models, run pipelines, generate text and fine-tune with the Trainer for NLP, vision, audio and multimodal tasks.
Why pin the packages together?
The skill says upgrading each package independently can make the training stack unsatisfiable, because Datasets 5.0.1 needs Hub below 2.
Which version was checked?
Transformers 5.18.0, verified against its released source on 1 October 2026.
Does it conflict with the esm SDK?
The skill says the esm SDK needs Transformers below 5, so it belongs in a separate environment.
How do I install it?
Run npx skills add https://github.com/k-dense-ai/scientific-agent-skills --skill transformers.
Is it checked for security?
Its security audits show passes from Gen Agent Trust Hub and Socket, and a warning from Snyk.
What are the best transformers alternatives?
Popular alternatives to transformers include pytorch-patterns, deepchem and pytorch-lightning. See all transformers alternatives