klbr/khoj
1
0
Fork 0
mirror of https://github.com/khoaliber/khoj.git synced 2026-03-02 13:18:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

Commit Graph

  • 7f18bc0840 Add default context for gemini 2 flash. 2x it for small, commercial models Debanjum 2025-04-04 16:14:20 +05:30
  • 47a081c7bd Allow text tool to give agent ability to terminate research Debanjum 2025-04-04 13:46:12 +05:30
  • 38dd02afbf Make ordering of fields expected by research planner consistent Debanjum 2025-04-04 09:42:13 +05:30
  • 443c5a4420 Consistently wrap queries in online search prompt in double quotes Debanjum 2025-04-03 18:11:39 +05:30
  • ae8fb6f9ac Default temperature of Gemini models to 1.0 to try avoid repetition Debanjum 2025-04-03 17:54:26 +05:30
  • e9928d3c50 Eval more model, control randomization & auto read webpage via workflow Debanjum 2025-04-03 18:00:25 +05:30
  • 911e1bf981 Use gemini 2.0 flash as evaluator. Set seed for it to reduce eval variance. Debanjum 2025-04-04 00:08:48 +05:30
  • 0dcb2544d7 Use embedded postgres instead of postgres server for eval workflow Debanjum 2025-04-03 22:30:53 +05:30
  • abfdf7b1fb Temporarily remove the OTP feature on the settings page while we fix our number sabaimran 2025-04-03 19:41:18 -05:00
  • 1ef8c37c3a Implement better template for feature request issue (#1132) Artem Yurchenko 2025-04-03 12:11:04 +02:00
  • 17997d26c5 Make generated code block extractor less strict to improve code tool Debanjum 2025-04-03 14:07:15 +05:30
  • 66e9ddb6be Support OpenAI (API compatible) models and Firecrawl in eval workflow Debanjum 2025-04-03 14:03:29 +05:30
  • ef5b19479f Improve setup of openai compatible speech to text, image models Debanjum 2025-04-03 13:44:19 +05:30
  • c2b9006a7a Improve Gemini Response Reliability (#1148) Debanjum 2025-04-03 13:42:35 +05:30
  • f77e871cc8 Improve agent creation safety checker with response schema, better prompt Debanjum 2025-04-03 02:49:40 +05:30
  • aab010723c Make Gemini response adhere to the order of the schema property definitions Debanjum 2025-04-03 00:04:57 +05:30
  • ae9ca58ab9 Specify min, max items expected in ai response via schema enforcement Debanjum 2025-03-31 00:46:11 +05:30
  • 0eb2d17771 Warn and drop invalid messages when format messages for gemini Debanjum 2025-03-30 23:45:40 +05:30
  • dd957bedd3 Remove pgserver in repo from git tracking sabaimran 2025-03-31 15:07:24 -05:00
  • d53b740197 Improve online search and allow server to skip auto webpage read sabaimran 2025-03-31 13:52:48 -05:00
  • 177560655d Fix and Improve Online Search and Webpage Read (#1147) Debanjum 2025-04-01 00:09:46 +05:30
  • d62dd4ef61 Support Firecrawl as a online search provider Debanjum 2025-03-31 17:04:12 +05:30
  • 3939e995e4 Fallback to enabled, lower priority online search providers on error Debanjum 2025-03-31 17:03:31 +05:30
  • 9b7442f28f Truncate online query to Serper if query exceeds max supported length Debanjum 2025-03-31 01:16:06 +05:30
  • db7eba56f6 Fix webpage read and improve web search with Jina Debanjum 2025-03-30 20:17:42 +05:30
  • db68372b81 Update code sandbox prompts to allow network access when using E2B Debanjum 2025-03-30 09:47:25 +05:30
  • 5b8c2989d6 Add hover text on button to unshare a conversation on web app Debanjum 2025-03-30 09:45:52 +05:30
  • 85d627ceb0 Simplify docs to self-host with pip since can use embedded DB now Debanjum 2025-03-29 19:04:56 +05:30
  • 713ba06a8d Release Khoj version 1.38.0 Debanjum 2025-03-29 18:30:06 +05:30
  • e9132d4fee Support attaching programming language file types to web app for chat Debanjum 2025-03-29 01:20:43 +05:30
  • bdb6e33108 Install pgserver only when `pip install khoj[local]' is enabled Debanjum 2025-03-29 00:18:21 +05:30
  • 5ee513707e Use embedded postgres db to simplify self-hosted setup (#1141) Debanjum 2025-03-29 00:03:55 +05:30
  • 56b63f95ea Suggest Google image gen model, new Anthropic chat models on first run Debanjum 2025-03-28 19:13:51 +05:30
  • 1e34de69e9 Fix spelling in Automations Docs (#1140) Ikko Eltociear Ashimine 2025-03-29 02:37:06 +09:00
  • 72986c905a Fix default agent creation to allow chat on first run Debanjum 2025-03-28 19:04:10 +05:30
  • 03de2803f0 Fallback to default agent for chat when unset in get conversation API Debanjum 2025-03-28 00:49:33 +05:30
  • a387f638cd Enforce json schema on more chat actors to improve schema compliance Debanjum 2025-03-27 18:33:08 +05:30
  • ccd9de7792 Improve safety settings for Gemini chat models Debanjum 2025-03-27 13:48:45 +05:30
  • 2ec5cf3ae7 Normalize type of chat messages arg sent to Anthropic completion funcs Debanjum 2025-03-25 18:27:01 +05:30
  • 4085c9b991 Fix infer webpage url step actor to request upto specified max urls Debanjum 2025-03-26 10:54:41 +05:30
  • c337c53452 Fix to use agent chat model for research model planning Debanjum 2025-03-25 15:29:56 +05:30
  • df090e5226 Enable unsharing of a public conversation (#1135) Debanjum 2025-03-25 14:24:01 +05:30
  • 9dfa7757c5 Unshare public conversations from the title pane on web app Debanjum 2025-03-24 17:38:37 +05:30
  • d9c758bcd2 Create API endpoint to unshare a public conversation Debanjum 2025-03-24 17:37:08 +05:30
  • e3f6d241dd Normalize chat messages sent to gemini funcs to work with prompt tracer Debanjum 2025-03-25 13:32:07 +05:30
  • 7976aa30f8 Terminate research if query or tool is empty Debanjum 2025-03-24 19:38:45 +05:30
  • 39aa48738f Set effort for openai reasoning models to pick tool in research mode Debanjum 2025-03-24 07:02:57 +05:30
  • b4929905b2 Add costs of ai prompt cache read, write. Use for calls to Anthropic Debanjum 2025-03-24 11:14:30 +05:30
  • d4b0ef5e93 Fix ability to disable code and internet providers in eval workflow Debanjum 2025-03-24 07:04:25 +05:30
  • a8285deed7 Release Khoj version 1.37.2 sabaimran 2025-03-23 11:38:25 -07:00
  • b7ac8771de Update a few pieces of documentation around data sources. sabaimran 2025-03-23 11:36:20 -07:00
  • 12e7409da9 Release Khoj version 1.37.1 sabaimran 2025-03-23 11:10:34 -07:00
  • 985f1672ed Remove eval lists from git tracking sabaimran 2025-03-23 10:59:32 -07:00
  • d1df9586ca Standardize AI model response temperature across provider specific ranges Debanjum 2025-03-23 18:04:21 +05:30
  • 55ae0eda7a Upgrade package dependencies nextjs for web app and torch on server Debanjum 2025-03-23 16:55:06 +05:30
  • 8409e64ff0 Clean AI model API providers documentation Debanjum 2025-03-23 15:49:22 +05:30
  • 86a51d84ca Access Claude and Gemini via GCP Vertex AI (#1134) Debanjum 2025-03-23 16:26:02 +05:30
  • 16ffebf765 Document how to configure using AI models via GCP Vertex AI Debanjum 2025-03-23 15:48:31 +05:30
  • 7153d27528 Cache Google AI API client for reuse Debanjum 2025-03-23 14:33:54 +05:30
  • da33c7d83c Support access to Gemini models via GCP Vertex AI Debanjum 2025-03-23 14:22:50 +05:30
  • 603c4bf2df Support access to Anthropic models via GCP Vertex AI Debanjum 2025-03-22 12:36:05 +05:30
  • 8bebcd5f81 Support longer API key field in DB to store GCP service account keyfile Debanjum 2025-03-23 10:39:27 +05:30
  • f2b438145f Upgrade sentence-transformers. Avoid transformers v4.50.0 as problematic Debanjum 2025-03-23 08:38:29 +05:30
  • 510cbed61c Make google auth package dependency explicit to simplify code Debanjum 2025-03-23 08:31:41 +05:30
  • 5fff05add3 Set seed for Google Gemini models using KHOJ_LLM_SEED env variable Debanjum 2025-03-22 08:59:31 +05:30
  • 6cc5a10b09 Disable SimpleQA eval on release as saturated & low signal for usecase Debanjum 2025-03-22 08:05:12 +05:30
  • 45015dae27 Limit to json enforcement via json object with DeepInfra hosted models Debanjum 2025-03-21 12:00:00 +05:30
  • dc473015fe Set default model, sandbox to display in eval workflow summary on release Debanjum 2025-03-20 14:44:56 +05:30
  • 80d864ada7 Release Khoj version 1.37.0 Debanjum 2025-03-20 14:06:57 +05:30
  • 0c53106b30 Fix passing inline images to vision models Debanjum 2025-03-20 09:26:04 +05:30
  • 1ce1d2f5ab Deduplicate, clean code for S3 images uploads Debanjum 2025-03-20 08:56:42 +05:30
  • f15a95dccf Show Khoj agent in agent dropdown by default on mobile in web app home Debanjum 2025-03-20 12:24:11 +05:30
  • 9a0b126f12 Allow chat input on web app while Khoj responds to speed interactions Debanjum 2025-03-19 23:00:27 +05:30
  • e68428dd24 Support enforcing json schema in supported AI model APIs (#1133) Debanjum 2025-03-19 22:59:23 +05:30
  • a5627ef787 Use json schema to enforce generate online queries format Debanjum 2025-03-19 14:31:39 +05:30
  • 2c53eb9de1 Use json schema to enforce research mode tool pick format Debanjum 2025-03-19 01:07:35 +05:30
  • 6980014838 Support constraining Gemini model output to specified response schema Debanjum 2025-03-19 16:10:24 +05:30
  • ac4b36b9fd Support constraining OpenAI model output to specified response schema Debanjum 2025-03-19 17:46:26 +05:30
  • 4a4d225455 Only enforce json output in supported AI model APIs Debanjum 2025-03-19 19:28:54 +05:30
  • d74c3a1db4 Simplify OpenAI reasoning model specific arguments to OpenAI API Debanjum 2025-03-19 17:37:52 +05:30
  • 9b6d626a09 Fix to store e2b code execution text output file content as string Debanjum 2025-03-19 14:59:48 +05:30
  • a7e261a191 Implement better bug issue template (#1129) Artem Yurchenko 2025-03-18 16:23:57 +01:00
  • 931f555cf8 Configure max allowed iterations in research mode via env var Debanjum 2025-03-18 17:01:14 +05:30
  • 2ab8e711d3 Fix Gemini models to output valid json when configured Debanjum 2025-03-18 16:53:28 +05:30
  • ce60cb9779 Remove max-w 80vw, which was smushing AI responses sabaimran 2025-03-13 13:44:05 -07:00
  • a3c4347c11 Add a one-click action to export all conversations. Add a self-service delete account action to the settings page sabaimran 2025-03-12 23:54:02 -07:00
  • 79816d2b9b Upgrade package dependencies of server, clients and docs Debanjum 2025-03-11 23:59:17 +05:30
  • 7bb6facdea Add support for Google Imagen AI models for image generation Debanjum 2025-03-11 22:52:20 +05:30
  • bd06fcd9be Stop using old google generativeai package to raise, catch exceptions Debanjum 2025-03-11 23:38:27 +05:30
  • bdfa6400ef Upgrade to new Gemini package to interface with Google AI Debanjum 2025-03-11 06:03:06 +05:30
  • 2790ba3121 Update default temperature for calls to Gemini models to 0.6 from 0.2 Debanjum 2025-03-11 06:07:13 +05:30
  • 50f71be03d Support Claude 3.7 and use its extended thinking in research mode Debanjum 2025-03-11 01:54:12 +05:30
  • 69048a859f Fix E2B tool description prompt to mention plotly package available Debanjum 2025-03-11 02:16:12 +05:30
  • 9751adb1a2 Improve Code Tool, Sandbox and Eval (#1120) Debanjum 2025-03-09 18:49:59 +05:30
  • c133d11556 Improvements based on code feedback Debanjum 2025-02-28 14:37:54 +05:30
  • 94ca458639 Set default chat model to KHOJ_CHAT_MODEL env var if set Debanjum 2025-02-16 02:58:27 +05:30
  • 7b2d0fdddc Improve code gen chat actor to output code in inline md code blocks Debanjum 2025-02-15 18:29:35 +05:30
  • 8305fddb14 Default to non-zero temperature for all queries to Gemini models. Debanjum 2025-02-15 13:55:08 +05:30
  • 45fb85f1df Add E2B as an optional code sandbox provider Debanjum 2025-02-15 03:11:15 +05:30
  • b4183c7333 Default to gemini 2.0 flash instead of 1.5 flash on Gemini setup Debanjum 2025-02-12 21:42:23 +05:30