# Hugging Face — vLLM V0 to V1: Correctness Before Corrections in RL

- Company: Hugging Face (huggingface.co)
- Announced: 2026-05-06T19:06:55+00:00
- Subject: Platform
- Source: https://huggingface.co/blog/ServiceNow-AI/correctness-before-corrections
- Record: https://forck.live/items/1506-vllm-v0-to-v1-correctness-before-corrections-in-rl

ServiceNow-AI describes migrating their RL training pipeline from vLLM V0 to V1, identifying and fixing four issues (logprob semantics, runtime defaults, inflight weight updates, fp32 lm_head) to achieve training parity.

## Evidence

Verbatim from https://huggingface.co/blog/ServiceNow-AI/correctness-before-corrections:

> vLLM V1 matched our vLLM V0 reference after we fixed four things: processed rollout logprobs, V1-specific runtime defaults, the inflight weight-update path, and the fp32 lm_head used for the final projection.

---

Record: https://forck.live/items/1506-vllm-v0-to-v1-correctness-before-corrections-in-rl
Catalogue: https://forck.live/llms.txt
Feed: https://forck.live/feed.md