AI News

MiniMax Sparse Attention (MSA): a Two-Branch Block-Sparse Attention Trained on a 109B-Parameter MoE With a 3T-Token Budget

Low Severity Global
Date Occurred Jun 17, 2026 07:44 UTC
Event Type AI News
Source AI News
Recorded Jun 17, 2026
Full Description

<p>MiniMax released MSA, a sparse attention built on Grouped Query Attention. A lightweight Index Branch selects Top-k key-value blocks per query and GQA group; the Main Branch attends only to those b

Original Source
https://www.marktechpost.com/2026/06/17/minimax-sparse-attention-msa-a-two-branch-block-sparse-attention-trained-on-a-109b-parameter-moe-with-a-3t-token-budget/
Event Metadata
  • ID #9182
  • Type AI News
  • Region Global
  • Severity Low
  • Indexed Jun 17, 2026
Quick Actions
Back to Events View on Globe Read Original Article

Related Events

Google Cloud generative AI automates council planning operations
Jun 17, 2026 Low Ukraine
MiniMax Sparse Attention (MSA): a Two-Branch Block-Sparse Attention Trained on a 109B-Parameter MoE With a 3T-Token Budget
Jun 17, 2026 Low
OpenAI’s Deployment Simulation Extends Pre-Deployment Risk Assessment to Agentic Coding Through Simulated Tool Calls
Jun 17, 2026 Low
OpenAI’s Deployment Simulation Extends Pre-Deployment Risk Assessment to Agentic Coding Through Simulated Tool Calls
Jun 17, 2026 Low
How to Build Memory-Efficient Transformers with xFormers Using Packed Sequences, GQA, ALiBi, SwiGLU, and Causal Attention
Jun 17, 2026 Low
How to Build Memory-Efficient Transformers with xFormers Using Packed Sequences, GQA, ALiBi, SwiGLU, and Causal Attention
Jun 17, 2026 Low