39
burn-msa
v0.2.4 ExperimentalMiniMax Sparse Attention (MSA) - block-sparse attention with per-GQA-group top-k block selection for Burn
AGPL-3.0 Edition 2021
#sparse#burn#transformer#attention#gqa
Quick Verdict
- ✓Actively maintained (updated 64d ago)
- !Pre-1.0: API may have breaking changes
Security
Checking security advisories...
Downloads
169
Dependents
0
Releases
8
Size
78KB
Deep Insights
⚠️
Notable yanked versions
25% of releases have been yanked. Check version history before pinning.
Health Breakdown
Maintenance 10/25
Recency, release consistency, active ratio
Quality 10/25
Yanked ratio, deps, size, maturity, features
Community 3/20
Reverse deps, ownership, ecosystem
Popularity 3/15
Downloads, momentum, growth trend
Documentation 13/15
Docs, repo, license, metadata
Download Trend
Daily downloads · last 90 days
0/day avg
Version Adoption
v0.2.3
25%
v0.2.1
20%
v0.2.4
20%
v0.1.2
18%
v0.2.2
18%
Release Timeline
8 releases2 yankedsince 2026
J
F
M
A
M
J
J
A
S
O
N
D
2026
8
LessMore|Yanked
Feature Flags
default =["std"]
std*cudaserdecubecl
README
Loading README...
Maintainers
Dependencies
10
direct dependencies
Dependents
0
crates depend on burn-msa