@ladybugmem

Building Agent Memory. Sponsoring @lbugdb open source project.

 San Francisco Bay Area
ladybugmem.ai
Joined January 2026
  • Tweets65
  • Following3
  • Followers80
12Media
Icebug 13.5 * SIGMOD 2025 parallel k-core from @lethalfirefox * ParallelLeidenView faster by caching computed neighbor lists selectively.
1
1
4
54
Don't forget to star ⭐ the repo. 75 stars = brew install icebug.
1
15
github.com/Ladybug-Memory/ic…

Release Release 13.5 · Ladybug-Memory/icebug

What's Changed perf(leiden): flat sort-combine aggregation in CoarsenedGraphView + ParallelLeidenView by @adsharma in #77 perf(leiden): adaptive eager neighbor cache in CoarsenedGraphView by @...

github.com
1
10
networkx goes columnar! #Python #DataScience
1
1
1
1
310
github.com/Ladybug-Memory/nx…

GitHub - Ladybug-Memory/nx-arrow

Contribute to Ladybug-Memory/nx-arrow development by creating an account on GitHub.

github.com
1
1
34
github.com/Ladybug-Memory/nx…

GitHub - Ladybug-Memory/nx-ladybug: LadybugDB storage provider for nx-arrow columnar graphs

LadybugDB storage provider for nx-arrow columnar graphs - Ladybug-Memory/nx-ladybug

github.com
1
20
nx-pyarrow As much as we'd like people to use icebug instead of networkx given the performance advantages, networkx isn't going away anytime soon.
2
1
60
It's on @pypi now. Use uv pip install nx-arrow
4
Previous results: xitter.cf/ladybugmem/status/2101…
Ladybug Memory@ladybugmemSep 19
Why should you care about icebug?
1
3
github.com/networkx/networkx…

pyarrow backend · networkx networkx · Discussion #8933

https://github.com/Ladybug-Memory/networkx/tree/pyarrow This branch contains a new backend that consumes 6.5x less memory vs the default backend. Trade off: faster for immutable graphs, but slower ...

github.com
1
3
Kudos to the DuckDB team for doing this! You can do this with @lbugdb as well. Also don't forget to query icedisk datasets on @huggingface with the recently released duckdb extension. duckdb.org/2026/09/25/huggin…

DuckDB and Hugging Face: Querying Datasets Directly

Hugging Face hosts hundreds of thousands of datasets, and DuckDB can read them directly, exactly where they are and without downloading anything, over the DuckDB hf:// protocol. This post looks at...

duckdb.org
1
2
76
Replying to @duckdb
@duckdb extension to read icebug format Graphs
1
1
2
407
github.com/Ladybug-Memory/du…

GitHub - Ladybug-Memory/duckdb-icedisk-extension: DuckDB extension: query icebug-disk CSR graphs...

DuckDB extension: query icebug-disk CSR graphs (Parquet indices/indptr) as native node/edge tables - Ladybug-Memory/duckdb-icedisk-extension

github.com
1
17
Why should you care about icebug?
1
2
3
102
⭐️ the repo: github.com/Ladybug-memory/ic… and tell all your friends about it.

GitHub - Ladybug-Memory/icebug: Icebug is a fork of networkit emphasizing zero-copy columnar storage

Icebug is a fork of networkit emphasizing zero-copy columnar storage - Ladybug-Memory/icebug

github.com
1
2
50
github.com/Ladybug-Memory/ic…

GitHub - Ladybug-Memory/icebug-benchmark-pagerank

Contribute to Ladybug-Memory/icebug-benchmark-pagerank development by creating an account on GitHub.

github.com
1
18
Icebug 13.3 is out! New FastRP algorithm implementation and minor bug fixes and improvements. Star it on github if you'd like to brew install icebug.
1
1
1
219
github.com/Ladybug-memory/ic…

GitHub - Ladybug-Memory/icebug: Icebug is a fork of networkit emphasizing zero-copy columnar storage

Icebug is a fork of networkit emphasizing zero-copy columnar storage - Ladybug-Memory/icebug

github.com
28
PSA: If you write a Graph Retrieval blog post and point at a repo with a closed source database, people can't try it easily. Consider using an open source embedded database. Porting is super easy. Link in Replies. Lot more engagement when people can try.
1
2
73
github.com/LadybugDB/ladybug…

GitHub - LadybugDB/ladybug-skill: An agent skill to interface with LadybugDB

An agent skill to interface with LadybugDB. Contribute to LadybugDB/ladybug-skill development by creating an account on GitHub.

github.com
1
28
Load more