AI Paper Atlas

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

Breaking the Reversal Curse in Autoregressive Language Models via Identity Bridge

Xutao Ma, Yixiao Huang, Hanlin Zhu, Somayeh Sojoudi

ICML 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks

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 identity bridge data (A->A) to forward-only training dataLlama-3.2-1B-Instruct finetuning · Husband-Wife and Parent-Child reversalreversal test accuracy: near-zero -> 50%Improved
add identity bridge data to one-layer transformer trainingone-layer transformer · symbolic reversal task N=10reversal test MRRImproved
phrase identity bridge directly (IDN) instead of rewriting to OCR form (ablation)Identity Bridge on Llama-3.2-1B · Husband-Wife reversalreversal test accuracyHurt
repeat forward relation data k=6 times in OCR format (ablation)Identity Bridge OCR format · Husband-Wife reversalreversal test accuracy: 50% (OCR-6, highest)Improved
use shorter entity names (one-token numbers versus three-token names) (ablation)Identity Bridge OCR format · Husband-Wife reversalreversal test accuracy: 10% -> nearly 100%Improved
use moderate weight decay (0.20 to 0.40) rather than 0.15 or 0.45 (ablation)Identity Bridge OCR format · Husband-Wife reversalreversal test accuracy: 0.12 (wd 0.15), 0.55 (wd 0.30), 0.33 (wd 0.45)Mixed
train on forward relation only with factorized one-layer transformer (ablation)one-layer transformer (theory) · symbolic reversal taskreversal margin: margin = 0Hurt
model learns a shortcut copying the name tokenIdentity Bridge on Llama-3.2-1B · Husband-Wife shortcut testshortcut test accuracy: near 100% early, around 60% for IDNHurt

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