Case Study

Performance Bottleneck in a Generative Model

A developer is profiling a text-generation model that uses a multi-layer Transformer architecture. They observe the following performance characteristics during autoregressive generation, where the time measured is dominated by the self-attention computations. Based on this data, identify the mathematical relationship between the generated sequence length and the computation time, and explain why this relationship occurs.

0

1

Updated 2025-10-08

Contributors are:

Who are from:

Tags

Ch.5 Inference - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences

Analysis in Bloom's Taxonomy

Cognitive Psychology

Psychology

Social Science

Empirical Science

Science