fix(catalog): call embedding with max 32 size batch #60
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Yougigun
pushed a commit
that referenced
this pull request
Aug 2, 2024
🤖 I have created a release *beep* *boop* --- ## [0.8.1-alpha](v0.8.0-alpha...v0.8.1-alpha) (2024-08-01) ### Bug Fixes * **catalog:** call embedding with max 32 size batch ([#60](#60)) ([e6b25ec](e6b25ec)) * **catalog:** max 3 catalog per namespace ([#58](#58)) ([786790c](786790c)) * **catalog:** when delete catalog and file, also delete the artifact ([#61](#61)) ([cf6ecc3](cf6ecc3)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please).
Yougigun
pushed a commit
that referenced
this pull request
Sep 13, 2024
🤖 I have created a release *beep* *boop* --- ## [0.14.0-alpha](v0.13.2-alpha...v0.14.0-alpha) (2024-09-13) ### Features * add CreateRepositoryTag endpoint ([#8](#8)) ([61bc325](61bc325)) * aggregate tag list with database info ([#7](#7)) ([6cc2d8d](6cc2d8d)) * **artifact:** add minIO retry and file deletion ([#89](#89)) ([8f391f0](8f391f0)) * **artifact:** update protogen-go ([c8d41a7](c8d41a7)) * **artifact:** use retrievable to decide if chunk can be return ([#54](#54)) ([bde3e85](bde3e85)) * **catalog:** add file catalog api ([#73](#73)) ([c30317f](c30317f)) * **catalog:** check the user tier for catalog limit ([#70](#70)) ([d35a96f](d35a96f)) * **catalog:** implement conversation and message api ([#77](#77)) ([e02b1f1](e02b1f1)) * **catalog:** order asc in create time ([#80](#80)) ([98348e9](98348e9)) * **catalog:** rename endpoint from kb to catalog ([#56](#56)) ([c8e543d](c8e543d)) * **catalog:** sort the chunk ([#74](#74)) ([c434cbd](c434cbd)) * **catalog:** support concurrent text to embedding process ([#85](#85)) ([12d313c](12d313c)) * **catalog:** support different file-to-embedding process ([#69](#69)) ([7f40dc1](7f40dc1)) * **catalog:** support more file type to uplaod ([#67](#67)) ([2d3c705](2d3c705)) * **catalog:** support originalData return ([#87](#87)) ([eb0c7fd](eb0c7fd)) * **catalog:** support question answering ([#71](#71)) ([a540c93](a540c93)) * **catalog:** support xlsx ([#79](#79)) ([f1e2505](f1e2505)) * **catalog:** update the pipeline that ask endpoint use ([#83](#83)) ([b5bbc75](b5bbc75)) * **catalog:** update the proto-go ([#82](#82)) ([94fa708](94fa708)) * **catelog:** make topK default 5 ([#62](#62)) ([02259e1](02259e1)) * create artifact database if it does not exist ([#4](#4)) ([787a4ad](787a4ad)) * fetch repository tags from registry ([#6](#6)) ([3568735](3568735)) * **kb:** add acl and update pipeline ([#49](#49)) ([bf7feea](bf7feea)) * **kb:** add check and clear message in openfga init ([#52](#52)) ([0a11838](0a11838)) * **kb:** add file metadata ([#37](#37)) ([51113ce](51113ce)) * **kb:** add owner_id parameter in knowledge base endpoint ([#27](#27)) ([e85020b](e85020b)) * **kb:** add resource limit ([#43](#43)) ([5f61f44](5f61f44)) * **kb:** add some kb metadata ([#36](#36)) ([0e42ff4](0e42ff4)) * **KB:** chunk catalog api ([#39](#39)) ([71a3996](71a3996)) * **kb:** dealing with failure in file-to-embedding process ([#47](#47)) ([0448a69](0448a69)) * **KB:** file-to-embedding worker pool ([#32](#32)) ([5409db3](5409db3)) * **kb:** knowldge base crud endpoints ([#19](#19)) ([0e9f32b](0e9f32b)) * **kb:** knowledge base repository done ([#18](#18)) ([f6aeaae](f6aeaae)) * **kb:** retrieval test api ([#41](#41)) ([8b94cc9](8b94cc9)) * **kb:** support knowledge base file related api ([#23](#23)) ([3912028](3912028)) * **kb:** update preset's pipeline ([2751023](2751023)) * **kb:** using preset's pipeline for file-to-embedding worker ([#45](#45)) ([8c57ad1](8c57ad1)) * **registry:** support image deletion ([#29](#29)) ([fe818da](fe818da)) * use camelCase for HTTP body ([#22](#22)) ([5d0fc2f](5d0fc2f)) * use dind in Dockerfile ([#14](#14)) ([d95aa68](d95aa68)) ### Bug Fixes * add /bin/sh to Dockerfile ([#10](#10)) ([7df1dd3](7df1dd3)) * **artifact:** fix ctx issue ([#91](#91)) ([1c61d69](1c61d69)) * **artifact:** improve catalog deletion slow issue ([#93](#93)) ([0bbf3f2](0bbf3f2)) * **artifact:** use correct version of proto ([#84](#84)) ([cba7e06](cba7e06)) * **catalog:** add requester when calling pipeline ([#72](#72)) ([0880d91](0880d91)) * **catalog:** call embedding with max 32 size batch ([#60](#60)) ([e6b25ec](e6b25ec)) * **catalog:** fix the catalog permission issue ([#75](#75)) ([2db2ce3](2db2ce3)) * **catalog:** fix topk to topK ([#65](#65)) ([5ff89fe](5ff89fe)) * **catalog:** list file api's page token ([#64](#64)) ([ac56be0](ac56be0)) * **catalog:** max 3 catalog per namespace ([#58](#58)) ([786790c](786790c)) * **catalog:** when delete catalog and file, also delete the artifact ([#61](#61)) ([cf6ecc3](cf6ecc3)) * expose private API on private port ([#9](#9)) ([9ef4b03](9ef4b03)) * **kb:** empty similar chunks ([#55](#55)) ([d1d5345](d1d5345)) * **kb:** fix db migration error ([#53](#53)) ([7cc65b7](7cc65b7)) * **kb:** fixed some bugs in file-to-embedding process ([#35](#35)) ([703bb0b](703bb0b)) * **kb:** get owner uid ([#26](#26)) ([b1d8ac5](b1d8ac5)) * **kb:** issue of chunking ([#34](#34)) ([66307c7](66307c7)) * **kb:** similar chunk search by prompt text ([#46](#46)) ([265f101](265f101)) * **kb:** use correct kb uid in chunk similarity search ([#44](#44)) ([e76aafa](e76aafa)) * return pagination in tag list endpoint ([#17](#17)) ([72bc47a](72bc47a)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Because
pipleine has limit size of batch, it causes calling embedding pipeline failed.
This commit
limits the max batch size to 32