A pure C inference engine runs the 2.78 trillion parameter Kimi K3 model on just 8GB RAM by streaming weights from disk, leveraging MoE sparsity (only 3.7% active per token) and computing directly on compressed formats, achieving correct output at 32.69 seconds per token on consumer hardware.
A new arXiv paper introduces Information Fidelity to measure how LLM compression of 10-Q MD&A sections and earnings calls changes downstream investment decisions, identifying decontextualization and model dependency as key failure modes and proposing Agentic Context Compression (ACC) to audit and reduce decision flips.
Google’s newly released TurboQuant algorithm compresses KV‑Cache from 16‑bit to 3‑bit, slashing memory usage to one‑sixth while preserving zero accuracy loss, dramatically accelerating large‑language‑model inference on GPUs and reshaping the memory market.
Google Research’s TurboQuant algorithm compresses large‑language‑model KV caches from 32‑bit to 3‑bit, achieving a six‑fold reduction in memory usage and an eight‑fold inference speedup on H100 GPUs while preserving 100 % accuracy, and it also improves vector search performance without requiring large codebooks.
