# Hugging Face Transformers v4.35.0 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v4.35.0
- **Released:** Nov 02, 2023
- **Impact Score:** 7/10
- **Breaking Changes:** Source flagged
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v4.35.0/

## Executive Summary (TL;DR)
Hugging Face Transformers v4.35.0. Source highlight: Distil-Whisper is a distilled version of Whisper that is 6 times faster, 49% smaller, and performs within 1% word error rate (WER) on out-of-distribution data. It was proposed in the paper Robust Knowledge Distillation via Large-Scale Pseudo Labelling.

## Breaking Changes & Deprecations
- Breaking changes * 🚨🚨🚨 [`Quantization`] Store the original dtype in the config as a private attribute 🚨🚨🚨  by @younesbelkada in #26761

## New Features
- [WhisperForCausalLM] Add WhisperForCausalLM for speculative decoding  by @patrickvonplaten in #27195
- [Whisper, Bart, MBart] Add Flash Attention 2  by @patrickvonplaten in #27203
- Add fuyu model  by @molbap in #26911
- Add Seamless M4T model  by @ylacombe in #25693
- Add `Kosmos-2` model  by @ydshieh in #24709
- Add OWLv2, bis  by @NielsRogge in #26668

## Improvements
- [Assistant Generation] Improve Encoder Decoder  by @patrickvonplaten in #26701
- Fuyu: improve image processing  by @molbap in #27007

## Bug Fixes
- Conversation pipeline fixes by @Rocketknight1 in #26795
- Add flash attention for `gpt_bigcode`  by @susnato in #26479
- [`FA2`] Fix flash attention 2 fine-tuning with Falcon  by @younesbelkada in #26852
- [Whisper, Bart, MBart] Add Flash Attention 2  by @patrickvonplaten in #27203
- 🚨🚨🚨 [`Quantization`] Store the original dtype in the config as a private attribute 🚨🚨🚨  by @younesbelkada in #26761
- [`FA-2`] Final fix for FA2 dtype  by @younesbelkada in #26846

## Upgrade Command
```bash
python -m pip install 'transformers==4.35.0'
```