paper-with-me

홈 › Papers

Can Mamba Always Enjoy the "Free Lunch"?

2024-10-04 · Ruifeng Ren, Zhicong Li, Yong liu

Transformers have been the cornerstone of current Large Language Models (LLMs); however, its linear growth in overhead during inference with respect to sequence length poses challenges for modeling long sequences. In this context, Mamba has gradually attracted attention due to its constant-level size during inference and existing empirical results have shown that it can perform comparably to Transformers in sequence modeling while offering significant savings. However, one may ask that, can Mamba always enjoy the ``free lunch"? In this paper, we focus on analyzing the expressive ability of Mamba from a theoretical standpoint. First, inspired by the connection between Mamba and linear attention, we investigate potential shortcomings of the Mamba when performing the COPY operation. Our results indicate that Mamba with constant size may encounter bottlenecks when handling COPY, while it can achieve perfect performance when the size scales linearly with sequence length. Based on this observation, we analyze Mamba's ability to tackle DP problems when equipped with Chain of Thought (CoT). Our findings suggest that to solve arbitrary DP problems, the total cost of Mamba is comparable to standard and efficient Transformers. However, similar to efficient Transformers, when facing DP problems with favorable properties such as locality, Mamba can provide savings in overhead. Our results contribute to a deeper understanding of Mamba.

📄 PDF Abstract BibTeX arXiv:2410.03810

Code (0)

등록된 구현이 없습니다.

Tasks

Mamba

Methods 이 논문이 사용한 방법론

Softmax The Softmax output function transforms a previous layer's output into a vector of probabilities. It is commonly used for multiclass classification. Given an input vector $x$…
Attention 설명 없음
Mamba Foundation models, now powering most of the exciting applications in deep learning, are almost universally based on the Transformer architecture and its core attention module.…
Focus 설명 없음

Similar Papers 제목 키워드 기반

On Fast Convergence of Proximal Algorithms for SQRT-Lasso Optimization: Don't Worry About Its Nonsmooth Loss Function

2016-05-25 · Xingguo Li, Haoming Jiang, Jarvis Haupt, Raman Arora 외

Many machine learning techniques sacrifice convenient computational structures to gain estimation robustness and modeling flexibility. However, by exploring the modeling structures, we find these "sacrifices" do not alwa…

regression

A No Free Lunch Theorem for Human-AI Collaboration

2024-11-21 · Kenny Peng, Nikhil Garg, Jon Kleinberg

The gold standard in human-AI collaboration is complementarity -- when combined performance exceeds both the human and algorithm alone. We investigate this challenge in binary classification settings where the goal is to…

Binary Classification

No Free Lunch: Non-Asymptotic Analysis of Prediction-Powered Inference

2025-05-26 · Pranav Mani, Peng Xu, Zachary C. Lipton, Michael Oberst

Prediction-Powered Inference (PPI) is a popular strategy for combining gold-standard and possibly noisy pseudo-labels to perform statistical estimation. Prior work has shown an asymptotic "free lunch" for PPI++, an adapt…

Free Lunch for Optimisation under the Universal Distribution

2016-08-16 · Tom Everitt, Tor Lattimore, Marcus Hutter

Function optimisation is a major challenge in computer science. The No Free Lunch theorems state that if all functions with the same histogram are assumed to be equally probable then no algorithm outperforms any other in…

The Implications of the No-Free-Lunch Theorems for Meta-induction

2021-03-22 · David H. Wolpert

The important recent book by G. Schurz appreciates that the no-free-lunch theorems (NFL) have major implications for the problem of (meta) induction. Here I review the NFL theorems, emphasizing that they do not only conc…