Commit Graph
2106 Commits
Author SHA1 Message Date
Timothy Jaeryang Baek 7f75acff96 chore: format 2025-06-08 22:08:25 +04:00
Timothy Jaeryang Baek 3a0696e1b0 refac 2025-06-08 21:21:51 +04:00
Timothy Jaeryang Baek d8d8380a78 refac/fix: multi-replica stop task (response) 2025-06-08 21:20:30 +04:00
Timothy Jaeryang Baek 0c57980e72 refac: tasks 2025-06-08 20:58:31 +04:00
Timothy Jaeryang Baek 4fe45d4430 refac/security: python code format endpoint 2025-06-08 20:26:07 +04:00
Timothy Jaeryang Baek 0cd400f5ee refac: docling picture describe params 2025-06-08 20:02:14 +04:00
Tim Jaeryang Baek 6cb519ca0e Merge pull request #14774 from rragundez/images-from-db
fix: Store and load code interpreter generated images from a central location (DB and/or cloud storage)
2025-06-08 19:09:28 +04:00
Tim Jaeryang Baek 6bf393a480 Merge pull request #14787 from vaclcer/vaclavs-custom-docling
feat: Customize Docling's "Describe Pictures" feature
2025-06-08 19:02:36 +04:00
Tim Jaeryang Baek ac4513a773 Merge pull request #14783 from mrexodia/local-dev-fixes
fix: Improve local development setup
2025-06-08 18:59:38 +04:00
Tim Jaeryang Baek 50d9a2ac58 Merge pull request #14781 from lucyknada/patch-2
fix: fix #14752 and add manual transcription retrieval
2025-06-08 18:40:28 +04:00
Vaclav Cerny 99f05561f8 Add configuration options for picture description modes and update related components 2025-06-08 16:30:26 +02:00
Duncan Ogilvie 7010393a41 Change default CORS_ALLOW_ORIGIN to '*'
The local development setup defaults do not actually work currently.
2025-06-08 15:16:51 +02:00
lucy b0965a8184 fixes #14752 and adds manual transcription option 2025-06-08 14:26:24 +02:00
Rodrigo Agundez 41e28a9f6f Use standardized procedure to save and load images 2025-06-08 16:26:58 +08:00
Rodrigo Agundez 168997a58e Return tuple consistently and strip potential left string 2025-06-08 16:26:08 +08:00
Timothy Jaeryang Baek 5e35aab292 chore: format 2025-06-05 01:12:28 +04:00
Timothy Jaeryang Baek f17bfb3adb refac 2025-06-05 01:11:31 +04:00
Timothy Jaeryang Baek 00e4391a6f refac 2025-06-05 01:05:33 +04:00
Timothy Jaeryang Baek 93236cead0 refac 2025-06-05 00:46:39 +04:00
Timothy Jaeryang Baek ab36b8aeae refac: embeddings endpoint 2025-06-05 00:37:31 +04:00
Tim Jaeryang Baek 7c4f261aa2 Merge pull request #14616 from Davixk/feat/new-perplexity-options
feat: add Perplexity AI model and search context usage configuration options
2025-06-05 00:28:00 +04:00
Tim Jaeryang Baek 14e158fde9 Merge pull request #14667 from hdnh2006/main
feat: OpenAI-compatible `/api/embeddings` endpoint with provider-agnostic OpenWebUI architecture
2025-06-05 00:25:56 +04:00
henry 04da2b05f6 convert embedding function name to be more consistence 2025-06-04 18:24:27 +02:00
Vaclav Cerny 9772c18b20 fix(loader): remove deprecated picture description configuration 2025-06-04 17:21:44 +02:00
henry cc12e9e1a3 Ollama embeddings adapted to pydantic 2025-06-04 17:06:38 +02:00
henry 3ddebefca2 openai embeddings function modified 2025-06-04 16:13:53 +02:00
henry 8f6c3f46d6 payload and response modifed for compatibility 2025-06-04 16:11:40 +02:00
henry e0769c6a1f new embedding.py added for handling openai and ollama embedding 2025-06-04 16:09:39 +02:00
Henry 415114504f Merge branch 'open-webui:main' into main 2025-06-04 15:37:33 +02:00
Vaclav Cerny 0b3719cc65 This commit fixes do_picture_description feature when using Docling as an external document parser.
Merge branch 'dev' into vaclavs-picture-description
2025-06-04 14:29:41 +02:00
Vaclav Cerny c71236ba07 feat(loader): enhance picture description prompt for improved detail and clarity 2025-06-04 14:25:31 +02:00
Vaclav Cerny c4278f4784 fix description vs classification mismatch 2025-06-04 14:13:00 +02:00
Timothy Jaeryang Baek bbafeca495 refac: reserved __user__ param format 2025-06-04 15:53:07 +04:00
Timothy Jaeryang BaekandRodrigo Agundez 9964ad0a5b refac: auth cache dir
Co-Authored-By: Rodrigo Agundez <rragundez@users.noreply.github.com>
2025-06-04 15:21:08 +04:00
Vaclav Cerny 8644e81a1c feat(loader): add picture description configuration for DoclingLoader 2025-06-04 12:34:39 +02:00
Timothy Jaeryang Baek d2b5a1614a refac 2025-06-03 21:52:25 +04:00
Timothy Jaeryang Baek 185249623b feat: follow ups backend integration 2025-06-03 18:47:49 +04:00
Timothy Jaeryang Baek 9e49fbc8bf feat: follow ups 2025-06-03 18:07:29 +04:00
Timothy Jaeryang Baek f8b941fb96 refac 2025-06-03 17:24:31 +04:00
Timothy Jaeryang Baek 4d364e2967 refac: remove msg from known type 2025-06-03 16:27:28 +04:00
Dave 77b357c73b fix: update label for search context usage to clarify its purpose 2025-06-03 00:27:07 +02:00
Dave 96e9bfe0e5 feat: add Perplexity model and search context usage configuration options 2025-06-03 00:19:08 +02:00
Tim Jaeryang Baek 3c32d2cada Merge pull request #14539 from PVBLIC-F/refac/mistral
perf mistral.py Enhance for Overall Speed and Efficiency
2025-06-02 23:52:59 +04:00
Tim Jaeryang Baek 0ebe35c571 Merge pull request #14532 from PVBLIC-F/refac/pinecone
perf pinecone.py Improve Performance and Maintainability Using Current Best Practices
2025-06-02 23:12:48 +04:00
Timothy Jaeryang Baek 2c15f8e676 refac 2025-05-31 15:07:28 +04:00
Timothy Jaeryang Baek 4e82c44f3e refac 2025-05-31 15:04:38 +04:00
Timothy Jaeryang Baek e41e375aab refac: role update ui 2025-05-31 15:00:27 +04:00
PVBLIC Foundation cf3635ba25 Update mistral.py
1. Intelligent Error Handling
Added _is_retryable_error() method to distinguish retryable vs non-retryable errors
Prevents unnecessary retries on client errors (4xx) that won't succeed
Caps retry delay at 30 seconds to prevent excessive waiting
2. Optimized Timeout Configuration
Upload: Capped at 2 minutes (was using full 5-minute timeout)
URL requests: 30 seconds (should be fast)
OCR processing: Full timeout (can take time)
Cleanup: 30 seconds (should be quick)
3. Enhanced Connection Pool
Increased connection limits: 20 total, 10 per host
Longer DNS cache TTL (10 minutes vs 5 minutes)
Increased keepalive timeout (60s vs 30s)
Added async DNS resolver for better performance
Granular timeout controls (connect, read, total)
4. Concurrency Control for Batch Processing
Added semaphore-based concurrency control (default: 5 concurrent)
Prevents API overwhelming while maintaining throughput
Configurable concurrency limit per workload
5. Memory Efficient Result Processing
Early exit for empty content validation
Better error metadata for debugging
Added content length tracking
Streamlined page processing logic
6. General Performance Improvements
Better error logging with truncated responses
Optimized metadata creation
Improved debug logging efficiency
2025-05-30 20:06:29 -07:00
PVBLIC Foundation 66bde32623 Update pinecone.py 2025-05-30 18:47:23 -07:00
PVBLIC Foundation 4ecf2a8685 Update pinecone.py
May 2025 Latest Pinecone Best Practices
2025-05-30 09:33:57 -07:00