arXiv:2604.18128v1 Announce Type: new
Abstract: We study post-training W4A4 quantization in a controlled 300M-parameter SwiGLU decoder-only language model trained on 5B tokens of FineWeb-Edu, and ask which input-activation sites dominate the error. Na…
arXiv:2604.18170v1 Announce Type: new
Abstract: LLMs edit text and code by autoregressively regenerating the full output, even when most tokens appear verbatim in the input. We study Copy-as-Decode, a decoding-layer mechanism that recasts edit generat…
arXiv:2604.12376v1 Announce Type: cross
Abstract: When LLM conversations grow beyond the context window, old content must be evicted — but how does the model recover it when needed? We propose cooperative paging: evicted segments are replaced with mi…