# Hugging Face — PipelineRL

- Company: Hugging Face (huggingface.co)
- Announced: 2025-04-25T22:37:16+00:00
- Category: developer-tool-release
- Subject: Platform
- Source: https://huggingface.co/blog/ServiceNow/pipelinerl
- Record: https://forck.live/items/1706-pipelinerl

ServiceNow and Hugging Face open-source PipelineRL, an experimental reinforcement learning implementation that addresses the trade-off between inference throughput and on-policy data collection by using inflight weight updates during RL training. It enables high inference throughput while keeping data on-policy, achieving competitive results on reasoning benchmarks compared to Open-Reasoner-Zero when training 7B and 32B models.

## Evidence

Verbatim from https://huggingface.co/blog/ServiceNow/pipelinerl:

> We are excited to open-source PipelineRL, an experimental RL implementation that tackles a fundamental challenge in large-scale Reinforcement Learning with LLMs: the trade-off between inference throughput and on-policy data collection.

---

Record: https://forck.live/items/1706-pipelinerl
Catalogue: https://forck.live/llms.txt
Feed: https://forck.live/feed.md