World Wires · story 107579 · corroborated · 1 source(s)

CompressionBench: Evaluating Coding Agents on LLM Compression

The study introduces CompressionBench to evaluate whether autonomous coding agents can successfully perform LLM compression tasks.

Open in the desk

Coverage

• CompressionBench: Evaluating Coding Agents on LLM Compression arXiv: new preprints (security, AI, physics) · tier C · clear net · 2026-11-04 18:14 UTC

What this site indexes