Is BA-Att superseded?

BA-Att (Long-context / context-window extension): current frontier — recent, not yet superseded in the knowledge base. 0 paper(s) critique it, 0 beat it on benchmarks — not ranked as a superseded baseline. Sub-problem: cluster led by StreamingLLM. Newer alternatives in the same sub-problem include BA-Att, CSAttention, TCA-Attention, Dynamic Hierarchical Sparse Attention (DHSA).

Long-context / context-window extension

BA-Att

Efficient Long-Context Modeling in Diffusion Language Models via Block Approximate Sparse Attention

On the frontierfirst seen May 19, 2026

Current frontier — recent, not yet superseded in the knowledge base

0 papers critique it · 0 beat it on benchmarks

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.

BA-AttEfficient Long-Context Modeling in Diffusion Language Models via Block Approximate Sparse Attention
May 19, 2026
CSAttentionCSAttention: Centroid-Scoring Attention for Accelerating LLM Inference
Mar 30, 2026
TCA-AttentionTraining-free Context-adaptive Attention for Efficient Long Context Modeling
Dec 10, 2025
Dynamic Hierarchical Sparse Attention (DHSA)Long-Context Modeling with Dynamic Hierarchical Sparse Attention for On-Device LLMs
Oct 28, 2025