# Hugging Face Transformers v4.16.0 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v4.16.0
- **Released:** Jan 27, 2022
- **Impact Score:** 5/10
- **Breaking Changes:** Not verified
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v4.16.0/

## Executive Summary (TL;DR)
Hugging Face Transformers v4.16.0. Source highlight: The Nyströmformer model was proposed in Nyströmformer: A Nyström-Based Algorithm for Approximating Self-Attention by Yunyang Xiong, Zhanpeng Zeng, Rudrasis Chakraborty, Mingxing Tan, Glenn Fung, Yin Li, and Vikas Singh.

## New Features
- Add Nystromformer
- Add REALM
- Add FastTokenizer to REALM
- Add MAE
- Add ViLT
- Add Swin Transformer

## Improvements
- Keras metric callback by @Rocketknight1 and @merveenoyan in https://github.com/huggingface/transformers/pull/14867

## Bug Fixes
- Fix installation instructions for BART ONNX example
- Fix doc examples: ... takes no keyword arguments
- Fix `AttributeError` from `PreTrainedTokenizerFast.decoder`
- Add 'with torch.no_grad()' to ALBERT integration test forward pass
- Add ONNX support for MarianMT models
- add custom stopping criteria to human eval script

## Upgrade Command
```bash
python -m pip install 'transformers==4.16.0'
```