perf3ct
3a4bb47cc1
feat(llm): embeddings work and are created when launching for the first ever time
2025-06-05 21:03:15 +00:00
perf3ct
bb8a374ab8
feat(llm): transition from initializing LLM providers, to creating them on demand
2025-06-05 19:27:45 +00:00
perf3ct
c1b10d70b8
feat(llm): also add functions to clear/unregister embedding providers
2025-06-05 18:59:32 +00:00
perf3ct
49e123f399
feat(llm): create endpoints for starting/stopping embeddings
2025-06-05 18:47:25 +00:00
perf3ct
a084805762
Merge branch 'develop' into feat/llm-change-to-single-provider
2025-06-05 18:26:40 +00:00
perf3ct
63722a28a2
feat(llm): also add embeddings options for embedding creation
2025-06-04 22:30:16 +00:00
perf3ct
fe15a0378a
fix(llm): have the model_selection_stage use the instance of the aiServiceManager
2025-06-04 20:23:06 +00:00
perf3ct
a20e36f4ee
feat(llm): change from using precedence list to using a sing specified provider for either chat and/or embeddings
2025-06-04 20:13:13 +00:00
Elian Doran
9bfadd7799
Merge branch 'develop' into dateNote
2025-06-04 22:54:10 +03:00
Elian Doran
4475568d19
fix(server): migration not working due to change in becca loader
2025-06-04 22:36:51 +03:00
Elian Doran
86689896a1
test(server): don't do automatic backup if migrating database
2025-06-04 21:44:27 +03:00
Elian Doran
df7f0d4099
Merge pull request #2110 from TriliumNext/feat/llm-integration-part3
...
LLM Integration, part 3
2025-06-04 11:41:49 +03:00
Elian Doran
6563601667
Merge pull request #2123 from FliegendeWurst/shortcuts-i18n
...
feat(i18n): description for all keyboard shortcuts
2025-06-04 11:39:47 +03:00
Elian Doran
8445ece231
Merge pull request #2106 from TriliumNext/fix/llm-becca-sync
...
fix(llm): Fix Note Embeddings not being synced correctly and causing sync loops
2025-06-04 11:38:49 +03:00
Jin
48ad381a50
refactor: 💡 Add typing
2025-06-04 10:33:40 +02:00
Elian Doran
555c596a0f
Merge branch 'develop' into shortcuts-i18n
2025-06-04 11:04:41 +03:00
Jin
f650cca652
refactor: 💡 replace substr with substring
2025-06-04 09:47:05 +02:00
FliegendeWurst
12d018fdfc
feat(i18n): description for all keyboard shortcuts
2025-06-03 19:10:12 +02:00
Elian Doran
cecbe5862c
fix(etapi): some endpoints failing due to async ( closes #2105 )
2025-06-03 19:46:38 +03:00
perf3ct
b76166b0d5
fix(llm): always fetch the embedding model
2025-06-03 05:13:32 +00:00
perf3ct
d4d55b20a8
fix(llm): get rid of a lot of log.info() statements that were spammy
2025-06-03 03:00:15 +00:00
perf3ct
ab3758c9b3
refactor(llm): resolve issue with headers being sent after request was sent
2025-06-02 23:54:38 +00:00
perf3ct
e7e04b7ccd
refactor(llm): streamline chat response handling by simplifying content accumulation and removing unnecessary thinking content processing
2025-06-02 23:25:15 +00:00
perf3ct
7b498cf384
refactor(llm): update chat saving logic to prevent race conditions between client and server
2025-06-02 22:30:59 +00:00
perf3ct
ed64a5b4f7
refactor(llm): simplify chat handling by removing session store and directly integrating chat storage service
2025-06-02 22:09:59 +00:00
perf3ct
d8d5318ace
refactor(llm): remove deprecated validateEmbeddingProviders method and update session handling in chat services
2025-06-02 21:55:52 +00:00
perf3ct
3a55735cd5
refactor(llm): implement new configuration methods for provider order and validation, enhancing error handling and deprecating legacy functions
2025-06-02 21:49:35 +00:00
perf3ct
5a5a69ebb8
refactor(llm): streamline model selection by utilizing the new configuration system and enhance error handling
2025-06-02 21:44:43 +00:00
perf3ct
ce7c4a31a1
refactor(llm): enhance configuration handling to avoid default assumptions and improve error handling
2025-06-02 21:43:36 +00:00
perf3ct
45175b6af3
refactor(llm): integrate new configuration system for provider management and model selection
2025-06-02 21:36:19 +00:00
perf3ct
00ce765e96
Merge branch 'develop' into feat/llm-integration-part3
2025-06-02 20:24:18 +00:00
perf3ct
934efab533
refactor(llm): update handling of note embeddings and clean up LoadResults class
2025-06-02 19:10:47 +00:00
perf3ct
cc0795f812
fix(llm): also add note_embeddings to froca and consistency tests
2025-06-02 15:49:45 +00:00
perf3ct
dcab4caee3
feat(llm): redo chat storage, part 3
2025-06-02 15:12:08 +00:00
Elian Doran
d9249e3760
fix(server): webclipper not working due to import issue
2025-06-02 16:34:41 +03:00
perf3ct
f6af617f6b
feat(llm): redo chat storage, part 2
2025-06-02 02:38:21 +00:00
perf3ct
35f78aede9
feat(llm): redo chat storage, part 1
2025-06-02 00:56:19 +00:00
perf3ct
3fae664877
feat(llm): add utils for excluding notes from LLM
2025-06-01 02:51:36 +00:00
perf3ct
ba98bd9097
Merge branch 'develop' into feat/llm-integration-part2
2025-05-31 10:48:00 -07:00
Elian Doran
adc9172681
Merge branch 'develop' of https://github.com/TriliumNext/Notes into develop
2025-05-30 21:33:28 +03:00
Elian Doran
8994f537dc
fix(migration): use bundleable migrations
...
This fixes TypeScript migrations, which were not being run correctly
even in previous versions.
On the ESBuild version, neither SQL migrations worked due to the fact
that they were not being bundled.
2025-05-30 21:32:35 +03:00
Elian Doran
b93e44a38f
fix(llm): button shown even if disabled
2025-05-30 17:39:38 +03:00
perf3ct
a7906d6b99
refactor(llm): improve search tools error handling and parameter validation with clearer guidance
2025-05-30 00:23:18 +00:00
perf3ct
910c5039f4
feat(llm): add LLM guidance system for failed tool executions and validations
2025-05-30 00:15:40 +00:00
perf3ct
f04e56137b
refactor(llm): improve type safety in tool calling stage and simplify tool call handling
2025-05-29 22:05:38 +00:00
perf3ct
7c63652105
feat(llm): improve type safety and error handling in tool call transformations
2025-05-29 22:03:23 +00:00
perf3ct
2f303b1ae9
feat(llm): also update OpenAI tool usage prompts
2025-05-29 21:58:03 +00:00
perf3ct
8f8b9d9e3b
feat(llm): really try to coax ollama to run tools
2025-05-29 21:24:04 +00:00
perf3ct
87859aec1c
fix(llm): reduce the use of "any" in the tool_calling_stage and update prompt for tool calling
2025-05-29 21:15:05 +00:00
perf3ct
ba59d6b3c1
feat(llm): reduce the use of "any" in the tool calling stage
2025-05-29 21:07:12 +00:00
