Compare Generated Sequences with Length-Normalized Log Probabilities

Sun, 06 Sep 2026 00:00:00 +0700

A language model assigns a probability to each next token, but applications often need to compare complete candidate sequences. A reranker may choose among generated answers. A decoder may keep several partial hypotheses. An evaluator may compare alternative completions under the same prompt.

The obvious approach is to multiply each candidate’s token probabilities, or equivalently add their log probabilities. That gives the probability the model assigns to the whole continuation. It also creates an important bias: every additional token contributes a probability no greater than 1, so longer sequences usually accumulate lower raw scores even when their individual tokens are highly plausible.

Reduce Transformer Inference Cost with Early Exiting

Sun, 06 Sep 2026 00:00:00 +0700

A transformer classifier normally spends the same number of layers on every input. A clear support request and an ambiguous one both pass through the entire network, even when an intermediate representation already contains enough information to classify the easy case correctly.

Early exiting changes that fixed-compute rule. It attaches prediction heads to intermediate layers and lets an input stop once a chosen exit rule considers the prediction sufficiently reliable. Easy inputs can use less computation, while harder inputs continue through deeper layers.

Model Inference on Nalar

Compare Generated Sequences with Length-Normalized Log Probabilities

Reduce Transformer Inference Cost with Early Exiting