Articles by anima-core
33

Post-transformer inference: 224× compression of Llama-70B with improved accuracy (zenodo.org)