Pith. sign in

REVIEW 8 cited by

The Impact of Positional Encoding on Length Generalization in Transformers

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

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 2305.19466 v2 pith:P3T3FEVR submitted 2023-05-31 cs.CL cs.AIcs.LG

Amirhossein Kazemnejad , Inkit Padhi , Karthikeyan Natesan Ramamurthy , Payel Das , Siva Reddy This is my paper · ORCID
classification cs.CLcs.AIcs.LG
keywords encodinggeneralizationlengthpositionaltransformersnopepositionrelative
verification ladder T0 review T1 audit T2 compute T3 formal

Signed reviews

No signed human review yet.

0 comments
read the original abstract

Length generalization, the ability to generalize from small training context sizes to larger ones, is a critical challenge in the development of Transformer-based language models. Positional encoding (PE) has been identified as a major factor influencing length generalization, but the exact impact of different PE schemes on extrapolation in downstream tasks remains unclear. In this paper, we conduct a systematic empirical study comparing the length generalization performance of decoder-only Transformers with five different position encoding approaches including Absolute Position Embedding (APE), T5's Relative PE, ALiBi, and Rotary, in addition to Transformers without positional encoding (NoPE). Our evaluation encompasses a battery of reasoning and mathematical tasks. Our findings reveal that the most commonly used positional encoding methods, such as ALiBi, Rotary, and APE, are not well suited for length generalization in downstream tasks. More importantly, NoPE outperforms other explicit positional encoding methods while requiring no additional computation. We theoretically demonstrate that NoPE can represent both absolute and relative PEs, but when trained with SGD, it mostly resembles T5's relative PE attention patterns. Finally, we find that scratchpad is not always helpful to solve length generalization and its format highly impacts the model's performance. Overall, our work suggests that explicit position embeddings are not essential for decoder-only Transformers to generalize well to longer sequences.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 8 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. OpenAlex reports about 33 citations worldwide. Full citation record

  1. Disentangling the Expressivity of RoPE

    cs.LG 2026-08 conditional novelty 7.0 of 10

    Periodic RoPE schedules yield exactly LTL with past operators and modular predicates, while conventional non-periodic RoPE provides only a bounded, precision-limited fixed-offset look-back mechanism.

  2. Anti-Periodic Positional Encoding: M\"obius Boundary Conditions Make In-Context Retrieval Reliable

    cs.CL 2026-07 conditional novelty 7.0 of 10

    An anti-periodic rotary positional encoding (M\u00f6bius RoPE) on 25% of heads collapses seed-to-seed variance in needle-in-a-haystack retrieval at matched perplexity.

  3. Addressable Memory for Video World Models

    cs.CV 2026-08 conditional novelty 6.0 of 10

    Video world models can recall revisited scenes far beyond their training horizon by storing compressed memory at fixed in-distribution positions and averaging keys in a rotation-free space.

  4. Input Pathways Shape Few-Shot, Not Zero-Shot, Binding in Tiny Transformers: A Fully-Enumerable Study

    cs.LG 2026-07 accept novelty 6.0 of 10

    In information-matched tiny transformers, zero-shot compositional binding fails for every route, while few-shot efficiency is governed by input-pathway sharing and code readability.

  5. RIS-Kernel: A Model-Agnostic Architecture for Long-Context LLM Inference via Sparse Attention

    cs.LG 2026-07 reject novelty 5.0 of 10

    RIS-Kernel claims O(N log N) sparse attention via ensembled random masks for long-context CPU inference, but the complexity claim conflicts with the paper's fixed-density mask design and the accuracy evidence is stati...

  6. Intra-Layer Recurrence in Transformers for Language Modeling

    cs.CL 2025-05 conditional novelty 5.0 of 10

    Selectively reusing early transformer layers inside a forward pass improves language modeling perplexity without adding parameters, but at extra compute cost and with evidence only from small-scale runs.

  7. HoPE: Hyperbolic Rotary Positional Encoding for Stable Long-Range Dependency Modeling in Large Language Models

    cs.CL 2025-09 reject novelty 4.0 of 10

    HoPE replaces RoPE's sine/cosine rotations with hyperbolic functions plus an exponential damping term to enforce monotonic attention decay, but the claimed consistent superiority and the 'RoPE as special case' theorem...

  8. Softplus Attention with Re-weighting Boosts Length Extrapolation in Large Language Models

    cs.CL 2025-01 conditional novelty 4.0 of 10

    A two-stage softplus-based attention mechanism with re-weighting (LSSAR) is reported to keep validation loss nearly flat when a 124M-parameter GPT-2 is tested at up to 16x its 1024-token training length.

Pith tools