| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
|
|
|
|
|
| |
* Use the Ollama generate endpoint instead of chat
Ollama has two API endpoints for text generation. There is a chat endpoint for interactive and interactive chat like generation of text and there is a generate endpoint that is used one one-shot prompts, such as summarization tasks and similar things.
Karakeep used the chat endpoint that resulted in odd summaries. This commit makes karakeep use the generate endpoint instead, which results in better and more compact summaries.
* format
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
client (#2231)
* Add OPENAI_PROXY_URL configuration and support for proxy in OpenAIInferenceClient
* docs: add OPENAI_PROXY_URL configuration for proxy support in OpenAI API requests
* format
---------
Co-authored-by: Mohamed Bassem <me@mbassem.com>
|
| |
|
|
|
|
|
|
|
|
| |
(#2000)
* fix: update OpenAI API to use max_completion_tokens instead of max_tokens
The OpenAI API has deprecated max_tokens in favor of max_completion_tokens
for newer models. This change updates both text and image model calls.
* feat: add support for max_completion_tokens in OpenAI inference configuration
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Add configurable fetch timeout for Ollama client
* Worker service needs access to the .env file
* repair typescript types
* Update customFetch.ts
* update the config docs
---------
Co-authored-by: sbarbett <shane@barbetta.me>
Co-authored-by: Mohamed Bassem <me@mbassem.com>
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* support embeddings generation in inference.ts
(cherry picked from commit 9ae8773ad13ed87af8f72f167bdd56e02ea66f15)
* make AI worker generate embeddings for text bookmark
* make AI worker generate embeddings for text bookmark
* fix unintentional change -- inference image model
* support embeddings for PDF bookmarks
* Upgrade drizzle-kit
Existing version is not working with the upgraded version of drizzle-orm.
I removed the "driver" to the match the new schema of the Config.
Quoting from their Config:
* `driver` - optional param that is responsible for explicitly providing a driver to use when accessing a database
* *Possible values*: `aws-data-api`, `d1-http`, `expo`, `turso`, `pglite`
* If you don't use AWS Data API, D1, Turso or Expo - ypu don't need this driver. You can check a driver strategy choice here: https://orm.
* fix formatting and lint
* add comments about truncate content
* Revert "Upgrade drizzle-kit"
This reverts commit 08a02c8df4ea403de65986ed1265940c6c994a20.
* revert keep alive field in Ollama
* change the interface to accept multiple inputs
* docs
---------
Co-authored-by: Mohamed Bassem <me@mbassem.com>
|
| |
|
| |
successfull -> successful
|
| | |
|
| |
|
