# Hugging Face — No GPU left behind: Unlocking Efficiency with Co-located vLLM in TRL

- Company: Hugging Face (huggingface.co)
- Announced: 2025-06-03T00:00:00+00:00
- Category: capability-change
- Subject: Platform
- Source: https://huggingface.co/blog/vllm-colocate
- Record: https://forck.live/items/1686-no-gpu-left-behind-unlocking-efficiency-with-co-located-vllm-in-trl

TRL now supports co-located vLLM, allowing training and inference to share the same GPUs, reducing idle time and improving throughput without extra hardware.

## Evidence

Verbatim from https://huggingface.co/blog/vllm-colocate:

> This approach is what we refer to as colocation. Training and inference are co-located on the same GPUs and coordinated via the same process group, allowing them to take turns smoothly — no extra hardware needed.

---

Record: https://forck.live/items/1686-no-gpu-left-behind-unlocking-efficiency-with-co-located-vllm-in-trl
Catalogue: https://forck.live/llms.txt
Feed: https://forck.live/feed.md