AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Function Induction and Task Generalization: An Interpretability Study with Off-by-One Addition

Qinyuan Ye, Robin Jia, Xiang Ren

ICLR 2026 · Architectures and efficiency › Attention and transformers · approach: Attention head mechanisms · 3 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
add the six function induction heads (versus replacing their outputs with base-run outputs) (ablation)Gemma-2 (9B) off-by-one addition · 16-shot off-by-one addition, range [0,9]off-by-one accuracy: 86% -> 0%Improved
keep six randomly chosen heads instead of ablating them (control) (ablation)Gemma-2 (9B) · 16-shot off-by-one additionaccuracyNo clear effect
ablate FI heads with zero or mean ablation instead of instance-specific (ablation)Gemma-2 (9B) · 16-shot off-by-one additioncontrast and base accuracy: contrast 0%, base 100%No clear effect
add more in-context examples (shots)six LMs on off-by-one addition · off-by-one addition, range [0,999]off-by-one accuracyImproved
increase more shots, effect on standard-addition accuracy under contrast promptsix LMs · off-by-one additionbase (standard addition) accuracyHurt
enlarge number range from [0,9] to [0,99]/[0,999]Gemma-2 (9B) and Qwen-2.5 (7B) (digit tokenization) · off-by-one additionaccuracyHurt
enlarge number range for models with single-token numbersLlama-2, Mistral, Llama-3, Phi-4 · off-by-one additionaccuracyNo clear effect
allow c_test to equal some earlier c_i (no copy restriction)Mistral-v0.1 (7B), Gemma-2 (9B) · off-by-one additionaccuracyImproved
increase absolute offset k in off-by-k additionGemma-2 (9B), Llama-3 (8B) · 32-shot off-by-k additionaccuracyHurt
ablate FI heads on four task pairs (off-by-two, shifted MMLU, Caesar cipher, base-8) (ablation)Gemma-2 (9B) · task generalization pairscontrast accuracyHurt
identified circuit used alone versus full modelGemma-2 (9B) · circuit evaluationfaithfulness: (7.17-0.56)/(7.17-(-1.26)) = 78.4%Improved
base-8 addition: model adjustments in no-carry caseGemma-2 (9B) · Base-8 addition Case 1unnecessary adjustment rate: 93% neither; 7% (=6%+1%) adjusted unnecessarilyMixed

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