# Hugging Face Transformers v2.7.0 — Release & Upgrade Intelligence
- **Product:** Hugging Face Transformers (https://huggingface.co)
- **Version:** v2.7.0
- **Released:** Mar 30, 2020
- **Impact Score:** 5/10
- **Breaking Changes:** Not verified
- **Canonical Analysis:** https://changelogs.us/transformers/updates/v2.7.0/

## Executive Summary (TL;DR)
Hugging Face Transformers v2.7.0. Source highlight: T5 Model (@patrickvonplaten, @thomwolf )

## Improvements
- Remove the LM head and use the embedding matrix instead (~200MB)
- Call encoder before expanding input_ids (~1GB)
- SelfAttention only returns weights if config.output_attentions (~500MB)
- Two separate, smaller decoder attention masks (~500MB)
- drop columns that are exclusively pad_token_id from input_ids in `evaluate_cnn` example.

## Upgrade Command
```bash
python -m pip install 'transformers==2.7.0'
```