AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Exploring Length Generalization in Large Language Models

Cem Anil, Yuhuai Wu, Anders Johan Andreassen, Aitor Lewkowycz, Vedant Misra, Vinay Venkatesh Ramasesh, Ambrose Slone, Guy Gur-Ari, Ethan Dyer, Behnam Neyshabur

NeurIPS 2022 · Large language models › Language, safety and interpretability · approach: Pruning and circuits in LLMs · 41 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
scale finetuned LaMDA from 244m to 64bvanilla finetuning · Parity length generalizationOOD accuracyNo clear effect
add scratchpad to finetuningfinetuned LaMDA · Parity and variable assignmentOOD length accuracyNo clear effect
pad input and scratchpad to fix token count and position bias bins (ablation)scratchpad finetuning · ParityOOD length accuracyMixed
mask distractor tokens in input and scratchpad (ablation)padded scratchpad finetuning · Parityscratchpad step accuracy on OOD lengthsImproved
mask distractor tokens in input only vs scratchpad only (ablation)padded scratchpad finetuning · ParityOOD length accuracyImproved
add few-shot scratchpad prompting without finetuningLaMDA 128b · Coin-flip parityaccuracy on longer lengthsImproved
add few-shot prompting to scratchpad finetuningLaMDA · ParityOOD length accuracyImproved
add few-shot prompting to scratchpad finetuningLaMDA · Variable assignmentOOD length accuracyNo clear effect
use larger learning rates in finetuningfinetuned LaMDA · ParityOOD length accuracyImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy