Two papers from the Cannistraci team accepted at NeurIPS 2026.
Advancing original research in semi-structured dynamic sparse training and channel-wise structural learning.
Read moreRESEARCH & NEWS
Research & News
Explore our papers, models and engineering work.
SparseMind and Moffett advance joint validation at CCF HPC 2026.
Connecting original sparse learning with execution on target hardware.
Read moreSparseMind supports the Boston network science and AI satellite as a Gold Sponsor.
Supporting Network Science Informs AI in Boston on June 1, 2026, and the exchange between network science and AI communities.
Read moreJoint evaluation of sparse models and chips
SparseMind × Moffett · Compiler-level evaluation and model quality.
Read moreJoint evaluation of a sparse Qwen3-8B model
SparseMind × H3C · Structured sparse training, model quality and deployment evaluation.
Read morePhysics-informed sparse modeling for GaN devices
SparseMind × MRDI · Research methods, acceptance targets and next steps.
Read more8B: Brain-inspired learning. Open to the world.
Open weights for a sparse Qwen3-8B model, bringing original research to hands-on evaluation and engineering.
Read moreHow should next-generation AI compute?
From connectivity to system efficiency, and from capability to accessibility.
Read more