paper-with-me

Papers

Transformers from Compressed Representations

2025-10-26 · Juan C. Leon Alcazar, Mattia Soldan, Mohammad Saatialsoruji, Alejandro Pardo, Hani Itani, Juan Camilo Perez, Bernard Ghanem arxiv

Compressed file formats are the corner stone of efficient data storage and transmission, yet their potential for representation learning remains largely underexplored. We introduce TEMPEST (TransformErs froM comPressed rEpreSenTations), a method that exploits the inherent byte-stream structure of compressed files to design an effective tokenization and encoding strategy. By leveraging this compact encoding, a standard transformer can directly learn semantic representations from compressed data streams, bypassing the need for raw byte-level processing or full media decoding. Our proposal substantially reduces the number of tokens required for semantic classification, thereby lowering both computational complexity and memory usage. Through extensive experiments across diverse datasets, coding schemes, and modalities, we show that TEMPEST achieves accuracy competitive wit the state-of-the-art while delivering efficiency gains in memory and compute.

📄 PDF Abstract BibTeX arXiv:2510.23665

Code (0)

등록된 구현이 없습니다.

Tasks

Representation Learning

Similar Papers 제목 키워드 기반

Quantization-Aware and Tensor-Compressed Training of Transformers for Natural Language Understanding

2023-06-01 · Zi Yang, Samridhi Choudhary, Siegfried Kunzmann, Zheng Zhang

Fine-tuned transformer models have shown superior performances in many natural language tasks. However, the large model size prohibits deploying high-performance transformer models on resource-constrained devices. This p…

Natural Language UnderstandingQuantization

Full Contextual Attention for Multi-resolution Transformers in Semantic Segmentation

2022-12-15 · Loic Themyr, Clement Rambour, Nicolas Thome, Toby Collins 외

Transformers have proved to be very effective for visual recognition tasks. In particular, vision transformers construct compressed global representations through self-attention and learnable class tokens. Multi-resoluti…

Semantic Segmentation

Training Transformers for KV Cache Compressibility

2026-05-07 · Yoav Gelberg, Yam Eitan, Michael Bronstein, Yarin Gal 외 arxiv

Long-context language modeling is increasingly constrained by the Key-Value (KV) cache, whose memory and decode-time access costs scale linearly with the prefix length. This bottleneck has motivated a range of context-co…

Question Answering

Temporal Latent Bottleneck: Synthesis of Fast and Slow Processing Mechanisms in Sequence Learning

2022-05-30 · Aniket Didolkar, Kshitij Gupta, Anirudh Goyal, Nitesh B. Gundavarapu 외

Recurrent neural networks have a strong inductive bias towards learning temporally compressed representations, as the entire history of a sequence is represented by a single vector. By contrast, Transformers have little …

Decision MakingInductive BiasSequential Decision Making

SCOUT: Toward Sub-Quadratic Attention via Segment Compression for Optimized Utility in Transformers

2025-08-31 · Aref Jafari, Yuhe Fan, Benyamin Jamialahmadi, Parsa Farinneya 외 arxiv

Transformers have demonstrated strong performance across a wide range of sequence modeling tasks, but their quadratic attention complexity limits scalability to long sequences. Linear models such as Mamba and sliding-win…