The paper studies long-context extension of Gated DeltaNet, a linear-attention architecture, through the spectral properties of its state transition matrix…
VC-Attention is a training-free low-bit attention framework for video diffusion transformers that tackles both accuracy loss from value outliers and…
HypoEvolve is a multi-agent framework that uses a generational genetic algorithm to coordinate specialized LLM agents that propose, critique, and…
LimiX-2 introduces the Contextual Mechanism Networks (CMNs) paradigm, pretrained with Context-Conditional Masked Modeling, which shifts tabular in-context learning from predicting…
Fathom addresses the bottleneck of scanning offloaded KV caches during long agentic sessions by letting each query decide how many…
The paper proposes XConf, a confidence-estimation method that grounds an LLM’s calibrated confidence in a stored record of its own…
Edge0 is a streaming MoE inference engine that serves large mixture-of-experts models from SSD on consumer hardware by predicting each…
Agora is a shared-memory system for autonomous research agents that records every hypothesis, result, and verification as an immutable Git…
The authors identify ‘Value Flattening,’ a failure mode in PPO critics for LLM RL where true state values shift sharply…
HyQuant is a hybrid-precision quantization framework that addresses accuracy loss in low-bit quantization of LLM attention by selectively keeping high…
The paper presents Auto-RecSys, an autonomous research-agent system for running large-scale recommender-system experimentation, using distributed asynchronous execution to run multiple…
SearchAtlas is a framework that converts raw search trajectories produced by LLM agents into structured evidence graphs showing how evidence…
