# Hugging Face Transformers v4.44.0 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v4.44.0
- **Released:** Aug 06, 2024
- **Impact Score:** 7/10
- **Breaking Changes:** Source flagged
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v4.44.0/

## Executive Summary (TL;DR)
Hugging Face Transformers v4.44.0. Source highlight: Release v4.44.0: End to end compile generation!!! Gemma2 (with assisted decoding), Codestral (Mistral for code), Nemotron, Efficient SFT training, CPU Offloaded KVCache, torch export for static cache

## Breaking Changes & Deprecations
- Breaking changes: We removed the chat template **in the code**, they should all be on the hub! * 🚨 No more default chat templates #31733 by @Rocketknight1

## New Features
- Add Nemotron HF Support #31699
- Add codestral mamba2 #32080 by @molbap and @vasqu
- adds: extra_repr() to MambaRMSNorm to include hidden size / size of weights in the layer
- Adds: extra_repr for RMSNorm layers in most models
- Add check for `target_sizes is None` in `post_process_image_guided_detection` for owlv2
- Add stream messages from agent run for gradio chatbot

## Improvements
- Enhancing SFT Training Efficiency Using Packing and FlashAttention2 with Position IDs
- Yell at the user if zero-3 init wasn't performed, but expected to have been done

## Bug Fixes
- fix
- fix: Fixed an if condition that is always evaluating to true
- fix: default value reflects the runtime environment variables rather than the ones present at import time.
- Fix resize embedding with Deepspeed
- Fix float8_e4m3fn in modeling_utils
- fix: Replaced deprecated `unittest method` with the correct one

## Upgrade Command
```bash
python -m pip install 'transformers==4.44.0'
```