FTCG, a Fine-to-Coarse Multi-Granularity grouping migratory compression scheme, employs Hierarchical Super- features to achieve progressive similarity quantification from fine to coarse granularity, thereby prioritizing the aggregation of highly similar data blocks to enhance local redundancy elimination.
This work presents an adaptive progressive compression framework that improves retrieval efficiency for two common targets, namely error-bound and peak Signal-to-Noise ratios and proposes coefficient decomposition, a novel method that exploits the commonly overlooked spatial correlations among decorrelated data.
Modern lossless compression enforces an acute dichotomy: industrial streaming codecs (e.g., Zstandard) prioritize throughput (10 to 1000 MiB/s) at the expense of ratio, while context-mixing algorithms achieve superior density at serial speeds (0.1 to 1 MiB/s). We present GPX, an adaptive multi-tier compression architec...
Delta compression is a promising technique for reducing the storage footprint of similar data. As the first step in delta compression, similarity detection has a significant impact on both the delta compression ratio (DCR) and the compression throughput. Unfortunately, existing approaches often make a trade-off between...
Model compression strategies are widely employed to reduce memory footprint and network complexity, particularly for devices with constrained computational, memory, and energy resources. Prior works that rely on simultaneous conversion from floating-point high-precision (FP32) to integer low-precision (INT4) representa...
A. Chaudhry, Osama Ahmad, Zubair Khalid et al.· 0 citations
This work proposes HyperZip, an efficient and scalable LLM-based compression framework that leverages diffusion-based LLMs (dLLMs) with Multi-Token Prediction (MTP) to accelerate LLM-based data compression processes and identifies a trade-off in diffusion-based compression.
The key-value (KV) cache becomes a major memory bottleneck in long-context LLM inference, placing substantial pressure on memory capacity and bandwidth. To mitigate this bottleneck, vector quantization (VQ) has emerged as a promising approach for aggressive KV cache compression. However, existing VQ methods degrade sub...
