Linear attention mechanisms deliver significant advantages for Large Language Models (LLMs) by providing linear computational complexity, enabling efficient processing of ultra-long sequences (e.g., 1M context). However, existing Sequence Parallelism (SP) methods, essential for distributing these workloads across devices, become the primary performance bottleneck due to substantial communication overhead. In this paper, we introduce ZeCO (Zero Communication Overhead) sequence parallelism for linear attention models, a new SP method designed to overcome these limitations and achieve practically end-to-end near-linear scalability for long sequence training. For example, training a model with a 1M sequence length across 64 devices using ZeCO takes roughly the same time as training with an 16k sequence on a single device. At the heart of ZeCO lies All-Scan, a novel collective communication primitive. All-Scan provides each SP rank with precisely the initial operator state it requires while maintaining a minimal communication footprint, effectively eliminating communication overhead. Theoretically, we prove the optimaity of ZeCO, showing that it introduces only negligible time and space overhead. Empirically, we compare the communication costs of different sequence parallelism strategies and demonstrate that All-Scan achieves the fastest communication in SP scenarios. Specifically, on 256 GPUs with an 8M sequence length, ZeCO achieves a 60\% speedup compared to the current state-of-the-art (SOTA) SP method. We believe ZeCO establishes a clear path toward efficiently training next-generation LLMs on previously intractable sequence lengths.

Lysergic acid diethylamide (LSD) has shown remarkable potential in modulating brain functional organization and dynamics. However, the exact mechanisms underlying its effects remain unclear. In this study, we employed a data-driven approach to analyze recurrent functional connectivity patterns in resting-state fMRI data and developed a parameterized feedback inhibition model to characterize excitatory/inhibitory (E/I) balance. The findings demonstrate that LSD enhances global brain synchrony and dynamic complexity. This enhanced synchrony likely stems from LSD's preferential stabilization of a globally synchronized yet functionally non-modular brain state - a pattern showing higher occurrence probability and acts as an "attractor" that recruits transitions from cognitive control networks. Crucially, these phenomena appear underpinned by LSD-induced convergence of excitatory/inhibitory balance across cortical hierarchies, particularly through Sensorimotor (SOM) suppression coupled with transmodal potentiation, where the Sensorimotor cortices emerge as potential regulatory hubs driving this neurochemical rebalancing. These convergent effects are consistent with the emergence of a brain state characterized by weakened sensory anchoring and enhanced cognitive flexibility, where the typical separation between concrete perception and abstract cognition becomes blurred. This neurophysiological remodeling therefore suggests a potential mechanism that could contribute to LSD's hallucinatory effects and its therapeutic potential in mental disorders characterized by rigid thought patterns.