REVIEW 17 cited by
Understanding Transformer from the Perspective of Associative Memory
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
Understanding Transformer from the Perspective of Associative Memory
read the original abstract
In this paper, we share our reflections and insights on understanding Transformer architectures through the lens of associative memory--a classic psychological concept inspired by human cognition. We start with the basics of associative memory (think simple linear attention) and then dive into two dimensions: Memory Capacity: How much can a Transformer really remember, and how well? We introduce retrieval SNR to measure this and use a kernel perspective to mathematically reveal why Softmax Attention is so effective. We also show how FFNs can be seen as a type of associative memory, leading to insights on their design and potential improvements. Memory Update: How do these memories learn and evolve? We present a unified framework for understanding how different Transformer variants (like DeltaNet and Softmax Attention) update their "knowledge base". This leads us to tackle two provocative questions: 1. Are Transformers fundamentally limited in what they can express, and can we break these barriers? 2. If a Transformer had infinite context, would it become infinitely intelligent? We want to demystify Transformer architecture, offering a clearer understanding of existing designs. This exploration aims to provide fresh insights and spark new avenues for Transformer innovation.
Forward citations
Cited by 17 Pith papers
-
Test-Time Training with KV Binding Is Secretly Linear Attention
Test-time training with KV binding reduces to learned linear attention.
-
Scaling Latent Reasoning via Looped Language Models
Looped language models with latent iterative computation and entropy-regularized depth allocation achieve performance matching up to 12B standard LLMs through superior knowledge manipulation.
-
Raven: High-Recall Sequence Modeling with Sparse Memory Routing
Raven is a linear-time sequence model whose sparse, input-dependent routing writes tokens into dedicated memory slots, preserving long-context recall and extrapolating 16x beyond training length.
-
FlowNar: Scalable Streaming Narration for Long-Form Videos
FlowNar achieves bounded memory and 3x higher throughput for streaming narration on Ego4D, EgoExo4D, and EpicKitchens100 by combining dynamic historical context removal with a Cross Linear Attentive Memory module.
-
More Expressive Feedforward Layers: Part I. Token-Adaptive Mixing of Activations
Mixture of Activations mixes activation functions token-adaptively in FFNs via lightweight gates, strictly more expressive than fixed or learnable activations, and yields lower pretraining loss from 0.12B to 2B models.
-
UniPool: A Globally Shared Expert Pool for Mixture-of-Experts
A shared global expert pool in MoE improves validation loss over per-layer experts and allows sublinear expert-parameter growth with depth.
-
Stateful Token Reduction for Long-Video Hybrid VLMs
For hybrid Mamba–Transformer video models, keeping 25% of visual tokens with a query-based progressive schedule gives 3.8–4.2x prefilling speedups with near-baseline accuracy; the paper attributes this to stateful com...
-
Neural Attention Search Linear: Towards Adaptive Token-Level Hybrid Attention Models
NAtS-L learns per-chunk routing between Gated DeltaNet and softmax attention, improving long-context retrieval and length extrapolation over fixed hybrids.
-
Kimi Linear: An Expressive, Efficient Attention Architecture
Kimi Linear hybridizes linear attention with a new KDA module to beat full attention on tasks while slashing KV cache by 75% and speeding decoding up to 6x.
-
Short window attention enables long-term memorization
Short sliding windows in hybrid attention-xLSTM models boost long-context performance by encouraging long-term memory use, and stochastic window sizing improves both short and long tasks.
-
PriorEye: Geospatial Visual Priors for End-to-End Autonomous Driving
PriorEye augments end-to-end driving models with a dual-memory architecture that stores and gates geospatial visual priors to improve performance and robustness to sensor corruption on NAVSIM-v2.
-
MDN: Parallelizing Stepwise Momentum for Delta Linear Attention
MDN parallelizes stepwise momentum for delta linear attention using geometric reordering and dynamical systems analysis, yielding performance gains over Mamba2 and GDN on 400M and 1.3B models.
-
Attention Residuals
Attention Residuals replaces fixed residual summation with input-dependent softmax attention over preceding layers, and a blocked variant is shown to improve uniformity and downstream performance in a 48B-parameter mo...
-
Distributed Dynamic Associative Memory via Online Convex Optimization
Tree-based distributed online gradient descent gives sublinear static regret and path-length-dependent dynamic regret for heterogeneous multi-agent associative memory with communication delays.
-
Scaling Latent Reasoning via Looped Language Models
A 1.4B and a 2.6B looped (weight-tied, recurrent-depth) language model trained on 7.7T tokens match or exceed several 4B–8B transformer baselines on selected reasoning benchmarks.
-
Distributed Associative Memory via Online Convex Optimization
A distributed online convex optimization protocol for associative memory achieves sublinear regret guarantees and outperforms baselines in experiments.
-
When Good Enough Is Optimal: Multiplication-Only Matrix Inversion Approximation for Quantized Gated DeltaNet
A multiplication-only truncated Neumann approximation for matrix inversion in quantized Gated DeltaNet linear attention delivers up to 5x kernel speedup and 20% decode overhead reduction while preserving accuracy on Q...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.