the Nemotron 3 models are trained to work with inference-time budget control
1 more from this model
The regular and medium-effort reasoning modes have the option to be used in conjunction with inference-time budget control.
general family · filed under inference & serving
Reasoning computation is constrained or allocated through a token budget, including inference-time and per-problem controls.
The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See reasoning control for the methods filed beside it.
Also called Configurable reasoning token budget, given a token budget, inference-time budget control, Inference-time reasoning budget control, per-problem budget control mechanism.
8 quotes from 7 sources, one card per model, each opening with its strongest.
the Nemotron 3 models are trained to work with inference-time budget control
The regular and medium-effort reasoning modes have the option to be used in conjunction with inference-time budget control.
To fine-tune reasoning effort while maximizing token efficiency, we implement a per-problem budget control mechanism during RL
The regular and medium-effort reasoning modes have the option to be used in conjunction with inference-time budget control.
as well as inference-time reasoning-budget control
can be turned off or given a token budget
The regular and medium-effort reasoning modes have the option to be used in conjunction with inference-time budget control.
The regular and medium-effort reasoning modes have the option to be used in conjunction with inference-time budget control.
Supports inference time reasoning budget control.