Comment by janalsncm
13 hours ago
Technically speaking when you send the prefix “The capital of France is “ into an LLM it will also produce probabilities across its whole vocabulary.
13 hours ago
Technically speaking when you send the prefix “The capital of France is “ into an LLM it will also produce probabilities across its whole vocabulary.
The probability values don’t really represent confidence in modern LLMs though, especially after RLHF and RLVR.
System One says they use RLCD, Reinforcement Learning for Calibrated Decisions, which presumably has accurate probabilities as an explicit optimisation goal.
How is that different from RLVR?
RLVR generally upweights tokens along the whole thinking trace that led to a correct answer, whether each token was "correct" or not. RLVR doesn't train a model to output an 80% likelihood, it just trains it to produce correct answers, and not to produce incorrect ones.
System One hasn't said how RLCD works, but they do say it is explicitly training models to output "calibrated" probabilities, which makes it distinct from RLVR. This is how they describe it:
> System One models are trained for calibrated decisions: their probabilities are optimized against outcomes to reflect uncertainty.
… which they could provide in their APIs but are vehemently opposed to because it makes distillation much easier, and faster.
Yeah. Yet another reason why open-weight models are better. If I want to use the logits, I can.