Andrew Kane

Pushed a version of less copying to the [hnsw-less-copy branch](https://github.com/pgvector/pgvector/compare/hnsw-less-copy). Initial results for 100k, 1536-dimension random vectors on my local machine (using 4 processes and 64MB for `maintenance_work_mem`): 189 sec...

Merged the branch in the commit above. There still may be some room to tune the memory context, but this should get most of the benefit from my testing.

Hi @VladlenPopolitov, I've looked into a few approaches for this: - simple filtering - [hnsw-filtering branch](https://github.com/pgvector/pgvector/compare/hnsw-filtering) - [HQANN](https://arxiv.org/abs/2207.07940) - [hqann2 branch](https://github.com/pgvector/pgvector/compare/hqann2) - [ACORN](https://arxiv.org/abs/2403.04871) but feel free to share if you...

Hi @jesseduffield, thanks for the PR (including tests)! Looks really great overall. I think we should have `nil` be the default (and use `Nokogiri::HTML`), and have `false` use `Nokogiri::HTML4`.

Will update in a follow-up commit. Thanks for the PR.

I'm seeing the same error with Python when trying to read the tokenizer from [Xenova/speecht5_tts](https://huggingface.co/Xenova/speecht5_tts). ```sh wget https://huggingface.co/Xenova/speecht5_tts/resolve/main/tokenizer.json ``` ```python from tokenizers import Tokenizer Tokenizer.from_file("tokenizer.json") ``` ```text thread '' panicked...

Thanks @greglandrum, here's a repro on GitHub Actions: https://github.com/ankane/rdkit-repro/actions/runs/10627243854/job/29460076854 ``` -- The C compiler identification is AppleClang 15.0.0.15000309 -- The CXX compiler identification is AppleClang 15.0.0.15000309 -- Detecting C compiler...

Yeah, the error doesn't occur with 2024.03.5 ([CI run](https://github.com/ankane/rdkit-repro/actions/runs/10628096391/job/29462305937)). `CMakeCache.txt` shows that Boost is found: ```txt Boost_DIR:PATH=/opt/homebrew/lib/cmake/Boost-1.86.0 ``` However, the include directory isn't included in the compilation command: ```sh cd...

Adding `include_directories(${Boost_INCLUDE_DIRS})` in `CMakeLists.txt` fixes it.

Hi @jasonxu123, thanks for reporting! Will think about how to address this for 1.0, but for now, you can hack around it with something like: ```yml groups: product: products: "where...