When the rollout finishes, the training loop asks Turnstile for the episode's data and gets back training-ready sequences grouped by group_id (an arbitrary user-chosen tag for a rollout or experiment).
I’m excited to share something I’ve been working on: Turnstile, a Rust proxy for token-level rollout capture in agentic RL.
It sits between a Chat Completions harness and model backend, recording everything trainers need.
https://t.co/TiFbv26HF8
Turnstile sits in the middle. The orchestrator thinks it's talking to a normal OpenAI API. Turnstile forwards requests to the real inference backend (sglang today, vLLM and others planned), streams responses back, and silently records every token in and out along with its logprob