Fix bugs in global kmeans #13809
Fix bugs in global kmeans #13809
Conversation
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
67b923b to
8a31f17
Compare
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
This comment was marked as outdated.
|
⚪ Test history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
| const auto cluster = std::string_view{row}.substr(sizeof(ui16) + sizeof(ui32)); | ||
| Y_DEBUG_ABORT_UNLESS(cluster == TSerializedCellVec{row}.GetCells().at(0).AsBuf()); |
There was a problem hiding this comment.
Это что за аццкие преобразования?
There was a problem hiding this comment.
Ну это корректный способ (но более медленный так как нужно парсить строку): нам приходит строка в случае kmeans sample стадии в виде TSerializedCellVec с одной колонкой внутри.
Она приходит в таком виде потому sample написан обобщенно, чтобы можно было если нужно использовать не для векторного индекса.
А остальные стадии на вход уже ожидают данные в правильном формате (просто строка с embedding-ом)
There was a problem hiding this comment.
Можно это вынести куда-то в функции типа Parse/Serialize, чтобы они были рядом и подчеркивалась их симметричность/связь?
There was a problem hiding this comment.
Ну Serialize общий, грубо говоря он зависит от SampleRequest , а deserialize тут конкретный для вектор индекса который знает какой message он отправляет
There was a problem hiding this comment.
Кажется просто не понимаю как должно выглядеть то что ты предлагаешь
|
⚪ Test history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
|
⚪ Test history | Ya make output | Test bloat
⚪ Test history | Ya make output | Test bloat | Test bloat
🟢
*please be aware that the difference is based on comparing your commit and the last completed build from the post-commit, check comparation |
Changelog category
Additional information