⚡ Modern Compression - matmat · Scour

Depth Registers Unlock W4A4 on SwiGLU: A Reader/Generator Decomposition 📼Cassette Combinators

NIM4-ASR: Towards Efficient, Robust, and Customizable Real-Time LLM-Based ASR 🎙️Whisper

Speculative Decoding for Autoregressive Video Generation 🧠Neural Codecs

EvoComp: Learning Visual Token Compression for Multimodal Large Language Models via Semantic-Guided Evolutionary Labeling 🗜️LZW Variants

From Image to Music Language: A Two-Stage Structure Decoding Approach for Complex Polyphonic OMR 🎧Learned Audio

On the Robustness of LLM-Based Dense Retrievers: A Systematic Analysis of Generalizability and Stability 🔍Information Retrieval

Micro Language Models Enable Instant Responses 🌊Streaming Compression

UniEditBench: A Unified and Cost-Effective Benchmark for Image and Video Editing via Distilled MLLMs 🎨LUT Compression

Variance Is Not Importance: Structural Analysis of Transformer Compressibility Across Model Scales 🧠Learned Compression

Universally Empowering Zeroth-Order Optimization via Adaptive Layer-wise Sampling 🧮Kolmogorov Bounds

DistortBench: Benchmarking Vision Language Models on Image Distortion Identification 📊Rate-Distortion Theory

ReaLB: Real-Time Load Balancing for Multimodal MoE Inference ⚡LZ4 Streaming

LBLLM: Lightweight Binarization of Large Language Models via Three-Stage Distillation 🔢Binary Embeddings

Dynamic Parameter Scheduling in Soft-Hard BPGD for Lossy Source Coding 📊Rate-Distortion Theory

Block-encodings as programming abstractions: The Eclipse Qrisp BlockEncoding Interface ⚛️Quantum Codecs

Self-Distillation as a Performance Recovery Mechanism for LLMs: Counteracting Compression and Catastrophic Forgetting 💻Local LLMs

No more posts from matmat's subscribed feeds.

Scour all 24382 feeds Learn more about Feeds

Log in to enable infinite scrolling