[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

![willccbb Avatar](https://lunarcrush.com/gi/w:24/cr:twitter::3064259332.png) will brown [@willccbb](/creator/twitter/willccbb) on x 28.6K followers
Created: 2025-07-25 13:02:57 UTC

@yacineMTB @jsuarez5341 for tabula rasa RL this is probably true

it's not true for LLMs, we already have a good universal abstraction for LLM inference via the OpenAI API spec, all serious inference engines support it. you can decouple your rollout logic from your training logic at zero cost


XXX engagements

![Engagements Line Chart](https://lunarcrush.com/gi/w:600/p:tweet::1948730692382159354/c:line.svg)

**Related Topics**
[open ai](/topic/open-ai)
[inference](/topic/inference)
[llm](/topic/llm)
[rl](/topic/rl)

[Post Link](https://x.com/willccbb/status/1948730692382159354)