Concept
Test-time compute
Test-time compute refers to the amount of computation a model spends per query at inference time, as opposed to the compute spent during training. The term gained traction because a model's quality is no longer fixed once weights are frozen: you can spend more cycles at…
The rest of “Test-time compute” 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→