Pith. sign in

REVIEW 7 cited by

Unlocking State-Tracking in Linear RNNs Through Negative Eigenvalues

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 2411.12537 v5 pith:GL7UX6XB submitted 2024-11-19 cs.LG cs.CLcs.FL

classification cs.LGcs.CLcs.FL
keywords lrnnsmatricesstate-trackingparitysolvedeltaneteigenvaluesmamba
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

Linear Recurrent Neural Networks (LRNNs) such as Mamba, RWKV, GLA, mLSTM, and DeltaNet have emerged as efficient alternatives to Transformers for long sequences. However, both Transformers and LRNNs struggle to perform state-tracking, which may impair performance in tasks such as code evaluation. In one forward pass, current architectures are unable to solve even parity, the simplest state-tracking task, which non-linear RNNs can handle effectively. Recently, Sarrof et al. (2024) demonstrated that the failure of LRNNs like Mamba to solve parity stems from restricting the value range of their diagonal state-transition matrices to $[0, 1]$ and that incorporating negative values can resolve this issue. We extend this result to non-diagonal LRNNs such as DeltaNet. We prove that finite precision LRNNs with state-transition matrices having only positive eigenvalues cannot solve parity, while non-triangular matrices are needed to count modulo $3$. Notably, we also prove that LRNNs can learn any regular language when their state-transition matrices are products of identity minus vector outer product matrices, each with eigenvalues in the range $[-1, 1]$. Our experiments confirm that extending the eigenvalue range of Mamba and DeltaNet to include negative values not only enables them to solve parity but consistently improves their performance on state-tracking tasks. We also show that state-tracking enabled LRNNs can be pretrained stably and efficiently at scale (1.3B parameters), achieving competitive performance on language modeling and showing promise on code and math tasks.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 7 Pith papers

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

  1. SHiPPO: Recurrent Memory with Transported Polynomial Projections

    cs.LG 2026-07 conditional novelty 7.0 of 10

    SHiPPO transports HiPPO coefficient memories via right actions into Sylvester dynamics, and diagnostics show this recovers order-sensitive memory changes that high-rank writes cannot.

  2. The Capability Convergence Hypothesis: Capability from Access Structure, Not Scale

    cs.AI 2026-07 conditional novelty 6.0 of 10

    Long-context capability converges to hybrid architectures pairing O(1) state with a scalable index; pure scaling of SSMs or Transformers alone hits information, horizon, or circuit walls.

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

    cs.LG 2025-10 conditional novelty 6.0 of 10

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

  4. Rethinking Addressing in Language Models via Contexualized Equivariant Positional Encoding

    cs.CL 2025-01 conditional novelty 6.0 of 10

    TAPE makes positional embeddings content-aware and equivariant, improving Transformer performance on arithmetic and long-context tasks and extending representational power to NC1-complete algorithms.

  5. Understanding Transformer from the Perspective of Associative Memory

    cs.LG 2025-05 conditional novelty 4.0 of 10

    Frames the Transformer as associative memory, derives retrieval SNR for linear, softmax, ReLU, and SoLU kernels, and proposes DeltaFormer, a softmax-plus-delta-rule variant claimed to exceed TC0 expressivity.

  6. State Propagation Also Satisfies: A Complex-Valued State-Space Model for Deterministic State Tracking

    cs.AI 2026-08 reject novelty 3.0 of 10

    A small complex-valued recurrent model achieves 100% accuracy on three length-16 state-tracking puzzles, but the claim that state propagation alone is sufficient is not tested against any baseline.

  7. ARWKV: Pretrain is not what we need, an RNN-Attention-Based Language Model Born from Transformer

    cs.CL 2025-01 reject novelty 3.0 of 10

    A three-stage distillation pipeline converts Qwen 2.5 into a pure RWKV-7 RNN, but the resulting 7B models lose substantial benchmark accuracy.

Pith tools