HydraHead: Head-level hybridization of full and linear attention for long-context efficiency
Researchers introduce HydraHead, an architecture that hybridizes Full Attention (FA) and Linear Attention (LA) at the head level rather than the conventional layer level, motivated by interpretability findings showing functional heterogeneity among heads within the same layer. An interpretability-driven selection strategy preserves FA only for retrieval-critical heads, achieving a 7:1 LA-to-FA ratio while matching the long-context performance of a 3:1 layer-wise hybrid. Trained on only 15B tokens, HydraHead achieves over 69% improvement over the baseline at 512K context length, approaching Qwen3.5's performance despite that model having a native 256K context window. The work suggests head-level hybridization is a significantly underexplored and high-potential design axis for efficient long-context models.
Related guides (2)
Related events (8)
DashAttention: Differentiable and Adaptive Sparse Hierarchical Attention for Long-Context LLMs
DashAttention introduces a two-stage hierarchical sparse attention mechanism that replaces the fixed top-k block selection used in methods like NSA and InfLLMv2 with an adaptive α-entmax transformation, allowing a variable number of KV blocks to be selected per query. The approach keeps the full hierarchy differentiable by using the first-stage selection as a prior for second-stage softmax attention. Experiments show comparable accuracy to full attention at 75% sparsity with a better Pareto frontier than competing methods, and a Triton GPU implementation achieves meaningful speedup over FlashAttention-3 at inference time.
ConSA: Learned FA/SWA allocation for efficient hybrid attention in LLMs
ConSA is a framework that learns optimal assignments between full attention and sliding-window attention layers under a user-specified sparsity target, using L0 regularization and augmented Lagrangian constraints. Evaluated on 0.6B and 1.7B parameter models, learned allocations consistently outperform hand-crafted rule-based baselines, with KV-head-wise granularity outperforming layer-wise. A consistent structural pattern emerges: SWA concentrates in bottom layers while FA clusters in contiguous middle-layer blocks, diverging from the evenly interleaved patterns used in existing hybrid architectures.
Falcon-H1: A Family of Hybrid-Head Language Models Redefining Efficiency and Performance
TII UAE has released Falcon-H1, a new family of hybrid-head language models combining attention and state-space mechanisms to improve efficiency and performance. The models are published on Hugging Face and represent TII's latest iteration in the Falcon series. The hybrid architecture targets better inference economics and competitive benchmark results relative to model size.
A Failed Experiment: Infini-Attention, and Why We Should Keep Trying?
A Hugging Face blog post documents an attempt to implement and validate Infini-Attention, a technique proposed to extend transformer context length by combining local and compressed global memory. The experiment reportedly failed to reproduce the claimed benefits, raising questions about the reproducibility and practical viability of the approach. The post frames the failure as instructive and argues for continued experimentation with long-context architectures.
QK-Restore: Fixing long-context recall degradation caused by CoT fine-tuning in hybrid LLMs
Researchers find that chain-of-thought supervised fine-tuning systematically degrades long-context recall in hybrid linear-attention models (HypeNet, Jet-Nemotron), with Needle-In-A-Haystack performance collapsing dramatically—e.g., HypeNet-9B dropping from 67.2% to 9.4% at 256K context. The root cause is identified as CoT-SFT biasing attention gradients toward short-range patterns, corrupting the query-key projections responsible for long-range routing. The paper proposes QK-Restore, a training-free fix that restores only W_Q and W_K from the pre-SFT checkpoint, recovering long-context capability while preserving reasoning gains.
Attention Expansion mechanism improves keyphrase extraction from long documents without full-context LLMs
Researchers propose an 'attention expansion' mechanism that augments pre-trained language model token representations with information from out-of-context chunks using static word embeddings, enabling more effective keyphrase extraction from long documents. The approach avoids the computational cost of full-document attention or LLM-based inference while expanding the effective contextual scope of PLM-based models. Evaluated across five PLM backbones and five benchmark corpora, the method consistently improves F1 scores over state-of-the-art baselines in both scientific and news domains.
Introducing HELMET: Holistically Evaluating Long-context Language Models
HELMET is a new benchmark designed to holistically evaluate long-context language models across diverse real-world tasks rather than synthetic needle-in-a-haystack tests. The benchmark covers multiple task categories including retrieval, reasoning, summarization, and code, aiming to provide more reliable and comprehensive assessment of long-context capabilities. It is introduced via the Hugging Face blog, suggesting an open release with associated tooling for the community.
Language Models Need Sleep: Periodic Context Consolidation via Fast Weights and SSM Blocks
This paper proposes a sleep-like consolidation mechanism for transformer-based LLMs to address the quadratic scaling of attention with context length. During 'sleep' phases, the model performs N offline recurrent passes over accumulated context, updating fast weights in state-space model (SSM) blocks via a learned local rule, then clears the KV cache. The approach is evaluated on synthetic tasks (cellular automata, multi-hop graph retrieval) and math reasoning, where standard transformers and SSM-attention hybrids fail, with performance scaling with sleep duration N.