AI RESEARCH

You Had One Job: Per-Task Quantization Using LLMs' Hidden Representations

arXiv CS.CL

ArXi:2511.06516v3 Announce Type: replace Many LLM applications require only narrow capabilities, yet standard post-