Concept
Step accuracy
Step accuracy measures how often an agent gets each individual action right across a multi-step task, as opposed to whether it completes the whole task. Think of it like sequence-level versus token-level accuracy in a decoder: task success is the sequence match (all or…
The rest of “Step accuracy” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→