4 papers · 1 filter
Copy-as-Decode: Grammar-Constrained Parallel Prefill for LLM Editing
Ziyang Liu
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 th…
Depth Registers Unlock W4A4 on SwiGLU: A Reader/Generator Decomposition
Ziyang Liu
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…
Cooperative Memory Paging with Keyword Bookmarks for Long-Horizon LLM Conversations
Ziyang Liu
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…
FinDeepResearch: Evaluating Deep Research Agents in Rigorous Financial Analysis
Fengbin Zhu, Xiang Yao Ng, Ziyang Liu +19
Deep Research (DR) agents, powered by advanced Large Language Models (LLMs), have recently garnered increasing attention for their capability in conducting complex research tasks.…