MSA: Memory Sparse Attention Files

Trainable latent-memory framework for 100M-token contexts

This is an exact mirror of the MSA: Memory Sparse Attention project, hosted at https://github.com/EverMind-AI/MSA. SourceForge is not affiliated with MSA: Memory Sparse Attention.

The interactive file manager requires Javascript. Please enable it or use sftp or scp.
You may still browse the files here.

Name	Modified	Size
MSA source code.tar.gz	2026-05-06	976.1 kB
MSA source code.zip	2026-05-06	988.1 kB
Totals: 2 Items		2.0 MB

Other Useful Business Software

Go from Code to Production URL in Seconds

Cloud Run deploys apps in any language instantly. Scales to zero. Pay only when code runs.

Skip the Kubernetes configs. Cloud Run handles HTTPS, scaling, and infrastructure automatically. Two million requests free per month.

Start Free

Veeam Data Platform v13.1

Move workloads across hypervisors and clouds with no vendor lock-in. Try VDP free today.

Try Veeam Data Platform today. Experience the unified platform that's secure by design, portable by default, and proven to recover clean, fast, and anywhere.

Try Now

Recommended Projects

FlashAttention
Fast and memory-efficient exact attention
LuatOS
Powerful embedded Lua Engine for IoT devices, with many components
Kimi K3 in C
A 2.78-trillion-parameter Kimi K3 running inference on a single CPU
AirLLM
AirLLM 70B inference with single 4GB GPU
x-transformers
A simple but complete full-attention transformer