Pith. sign in

REVIEW 3 cited by

One-layer transformers fail to solve the induction heads task

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2408.14332 v1 pith:SYSI26BT submitted 2024-08-26 cs.LG stat.ML

One-layer transformers fail to solve the induction heads task

classification cs.LG stat.ML
keywords headsinductionone-layersizesolvetasktransformerargument
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

A simple communication complexity argument proves that no one-layer transformer can solve the induction heads task unless its size is exponentially larger than the size sufficient for a two-layer transformer.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Indexing: the Beginning and the End

    cs.LG 2026-07 conditional novelty 7.0

    Causal-complexity bounds show RNNs, SSMs, and masked linear attention need ω(1) layers for right-hand indexing, while a one-layer softmax transformer solves it; when the index is first, a one-layer RNN suffices.

  2. Eigenvalues as a Metric for Memory Dynamics in Sequence Models

    cs.LG 2025-10 conditional novelty 6.0

    Eigenvalue spectra of attention and SSM dynamics show consistent signatures of memory retention and selective forgetting that align with task requirements.

  3. Fast attention mechanisms: a tale of parallelism

    cs.LG 2025-09 conditional novelty 6.0

    ANNA, a hashing-based sub-quadratic attention mechanism, provably preserves standard attention's MPC expressiveness and can simulate low-rank attention, while being simulable by MPC with near-linear machines.