In Part 1, @max.io shared a guide to building autocomplete in Elasticsearch & OpenSearch.
In Part 2, he shows you how to scale it to 6M docs and runs side-by-side load tests with a large hybrid search index.
Dive in 👉 bonsai.io/blog/how-to-...
#autocomplete #elasticsearch #opensearch
Having an official incident report on the web is a great battle scar.
Nice one. Maybe we could try experiment runs could use some histograms or violins or something for comparison, instead of the average NDCG.
The end of Whiplash except it's the Mute City theme from F-Zero.
Against my better judgement, I gave this LinkedIn scam bot the phone number of an SMS scam bot. Will they find love?
Any embedding model for the legal domain worth its salt would designate semicolon as a special token
Being a chess nerd I get especially triggered when I see chess AI slop. Hero images like this pop up on linkedin often.
Who wore it best?
Long awaited, the ROC Data Science meetup returns! If you're in Rochester, come to the meetup reboot tomorrow night! Some of us will give some predictions for 2025, and @valerie-carey.bsky.social will get the inertia back for future events!
2 years ago I tried my darndest to run CLIP models in ONNX-RT without python. The issue has been the lack of the Multi-Head-Attention operator. This has also held back Llama ONNX and others. We might get a fix in v1.18 github.com/onnx/onnx/pu...
👋 where my search nerds at?
It's going to really change everything. Really hope the huggingface replication goes well and solidifies the science.
