The paper introduces Parallax, a scalable and numerically stable parameterized Local Linear Attention mechanism that significantly improves LLM performance and efficiency compared to existing methods…

The paper demonstrates that the location and nature of state encoding in sequence models are not fixed architectural traits but are highly dependent on the specific task, showing that the encoding pro…

The paper proposes AsymCache, a computation-latency-aware KV cache management system that optimizes LLM inference by aligning cache eviction decisions with GPU attention kernel performance, significan…

The paper proposes moving the query instead of the KV-cache during cross-instance attention, demonstrating that this approach is significantly cheaper than moving the cache, especially on modern GPU f…

ElasticMem introduces a novel framework that treats memory as an elastic latent resource, allowing LLM agents to adaptively manage and inject variable-budget memories for improved performance in long-…

The paper proposes SISA (SSM-Informed Softmax Attention), a novel hybrid attention mechanism that integrates state-space model (SSM) importance signals directly into the attention score, achieving sta…

The paper proposes Cross-Layer Sparse Attention (CLSA) to significantly improve the efficiency and accuracy of long-context LLMs by jointly optimizing KV-cache sharing and the routing index across dec…

The paper introduces Sparse Memory-Efficient Training (SMET), a method that stabilizes and optimizes Dynamic Sparse Training (DST) for large language models, enabling stable and memory-efficient spars…

This paper proposes a hybrid scheduler that dynamically switches between exclusive batching and mixed batching for LLM inference, achieving superior throughput, especially on bandwidth-constrained GPU…

This paper investigates how different types of compressed reasoning data (Explicit, Composed, Implicit CoT) affect LLM performance during post-training, finding that the choice of compression and subs…

The paper analyzes the distinct computational roles of positional versus symbolic attention heads in Transformers, demonstrating that symbolic mechanisms generalize more reliably to longer sequences t…

The paper introduces Eyettention II, a novel deep-learning model that can generate realistic, detailed scanpaths—including fixation location, within-word landing position, and duration—to address the…

Physical AI inference (batch-1 decode) is primarily memory-bandwidth-bound, but the observed latency gap between fast and slow GPUs is not solely due to memory bandwidth, as launch-side overheads beco…

PARCEL introduces a novel visual tokenization architecture that combines spatial pooling anchors with conditioned elastic queries, efficiently reducing the computational cost of large Vision-Language…

Soft-NBCE introduces soft entropy-weighted chunk fusion to overcome the semantic fragmentation caused by hard chunk selection in long-context LLMs, significantly improving performance on multi-hop ben…

The paper reframes context distillation as a latent memory management problem, proposing a modular framework using LoRA adapters and a Self-Gating mechanism for efficient, selective memory retrieval a…

Moment-KV introduces a novel momentum-based technique to compress the Key-Value (KV) cache during the decoding phase of LLM generation, significantly improving fidelity in long-generation tasks.

This paper proposes Supervised Memory Training (SMT), a method for training nonlinear RNNs that sidesteps recurrent credit propagation entirely.