I struggled getting a #Testcontainers container to talk back to a (mock) server in a #gitlab CI job. I finally got it working, and wrote down the solution (using `hostname -i`) in this post: www.aryweb.nl/2025/09/02/r...
How can you use the `relatedness` aggregation from #solr in #elasticsearch or #opensearch: www.aryweb.nl/2025/06/24/r.... This lets you score related terms in the same way with a painless script.
Wrote some notes for my later self to match on SPLADE/doc2query like related keywords that are weighted and use those weights for ranking with #elasticsearch www.aryweb.nl/2025/06/19/e...
Does this also mean it's possible have an SQS target in a different region (but in the same account)? I tried that last week but had to use SNS instead.
Our web performance got worse by switching of Cloudflare DNS proxy and using AWS Cloudfront. Using Cloudfront we're able to tweak caching better, but surprised with the initial results...
dev.to/arian/cloudf...
Spent an hour today figuring out why docker didn't work and reinstalling it... Apparently this was it 😓
bsky.app/profile/just...
I'm working on the finishing touch of the talk I'll be giving tomorrow with Coen Baars at the #elasticon Amsterdam event about re-ranking techniques. Looking forward to it!
A plot showing that reranking improves recall as we increase the number of reranked docs, but with increasing docs we diminishing returns and eventually a performance dip.
Mat is not on 🦋—posting on his behalf!
It's time to revisit common assumptions in IR! Embeddings have improved drastically, but mainstream IR evals have stagnated since MSMARCO + BEIR.
We ask: on private or tricky IR tasks, are rerankers better? Surely, reranking many docs is best?