Transformers

Transformers

The Hugging Face transformers library is the de-facto standard for accessing and running thousands of pre-trained models. It provides a unified API across PyTorch, TensorFlow, and JAX for text, image, audio, and multimodal inference and fine-tuning.

Key Features

• Hub integration — one-line download of 500,000+ models from huggingface.co
• Pipelines API — zero-boilerplate inference for text generation, classification, NER, translation, summarization, and more
• Trainer API — full fine-tuning loop with mixed precision, gradient accumulation, and distributed training
• Generation utilities — beam search, sampling, contrastive search, speculative decoding
• Quantization — bitsandbytes 4-bit/8-bit, GPTQ, and AWQ support built-in
• TGI / vLLM compatible — models load into all major serving runtimes

Quick Start

pip install transformers torch

from transformers import pipeline

generator = pipeline("text-generation", model="meta-llama/Llama-3.2-1B-Instruct")
result = generator("The future of AI agents is", max_new_tokens=50)
print(result[0]["generated_text"])

Install via ai-supply

npx ai-supply add huggingface-transformers

Curated mirror of the open-source Transformers project (Apache-2.0). Install upstream from the repository.