# vLLM v0.2.3 — Release & Upgrade Intelligence
- **Product:** vLLM (https://vllm.ai)
- **Version:** v0.2.3
- **Released:** Dec 03, 2023
- **Impact Score:** 5/10
- **Breaking Changes:** Not verified
- **Canonical Analysis:** https://changelogs.us/vllm/updates/v0.2.3/

## Executive Summary (TL;DR)
vLLM v0.2.3. Source highlight: Refactoring on Worker, InputMetadata, and Attention

## New Features
- Add instructions to install vllm+cu118
- Add stop_token_ids in SamplingParams.__repr__
- [DOCS] Add engine args documentation
- Added echo function to OpenAI API server.
- [Docs] Add information about using shared memory in docker
- Add profile option to latency benchmark

## Improvements
- [Docs] Update the AWQ documentation to highlight performance issue

## Bug Fixes
- Fix TP support for AWQ models
- Fix Baichuan & Baichuan 2
- Fix the code block's format in deploying_with_docker page
- [FIX] Update the doc link in README.md
- [BugFix] Fix a bug in loading safetensors
- Fix hanging in the scheduler caused by long prompts

## Upgrade Command
```bash
python -m pip install 'vllm==0.2.3'
```