klbr/khoj
1
0
Fork 0
mirror of https://github.com/khoaliber/khoj.git synced 2026-03-08 21:29:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Commit Graph

  • 22cd638add Fix handling unset openai_base_url to run eval with openai chat models Debanjum 2025-05-18 18:43:40 -07:00
  • cf55582852 Retry on empty response or error in chat completion by llm over api Debanjum 2025-05-18 14:32:31 -07:00
  • 7827d317b4 Widen vision support for chat models served via openai compatible api Debanjum 2025-05-11 11:13:27 -06:00
  • 4f3fdaf19d Increase khoj api response timeout on evals call. Handle no decision Debanjum 2025-05-18 14:42:27 -07:00
  • 31dcc44c20 Output tokens >> reasoning tokens to avoid early response termination. Debanjum 2025-05-18 14:40:57 -07:00
  • 73e28666b5 Fix to set default chat model for all user tiers via env var Debanjum 2025-05-18 09:49:30 -07:00
  • 06dcd4426d Improve Research Mode Context Management (#1179) Debanjum 2025-05-17 17:38:31 -07:00
  • fd591c6e6c Upgrade tenacity to respect min time for exponential backoff Debanjum 2025-05-17 14:21:47 -07:00
  • 988bde651c Make researcher aware of no. of web, doc queries allowed per iteration Debanjum 2025-05-13 12:00:06 -06:00
  • 417ab42206 Track gemini 2.0 flash lite cost. Reduce max prompt size for 4o-mini Debanjum 2025-05-13 13:03:35 -06:00
  • e125e299a7 Ensure time to first token logged only once per chat response Debanjum 2025-05-13 12:14:19 -06:00
  • 2694734d22 Update truncation logic to handle multi-part message content Debanjum 2025-05-10 18:41:16 -06:00
  • a337d9e4b8 Structure research iteration msgs for more granular context management Debanjum 2025-05-10 15:44:50 -06:00
  • 0f53a67837 Prompt web page reader to extract quantitative data as is from pages Debanjum 2025-05-11 15:29:38 -06:00
  • 99a2305246 Improve tool chat history constructor and fix its usage during research. Debanjum 2025-05-10 16:15:12 -06:00
  • 8050173ee1 Timeout calls to khoj api in evals to continue to next question Debanjum 2025-05-17 17:25:31 -07:00
  • 442c7b6153 Retry running code on more request exception Debanjum 2025-05-17 17:07:31 -07:00
  • 10a5d68a2c Improve retry, increase timeouts of gemini api calls Debanjum 2025-05-17 02:40:47 -07:00
  • 20f08ca564 Reduce timeouts on calling local and online llms via openai api Debanjum 2025-05-17 16:24:05 -07:00
  • e0352cd8e1 Handle unset ttft in metadata of failed chat response. Fixes evals. Debanjum 2025-05-17 02:36:10 -07:00
  • 673a15b6eb Upgrade hf hub package to include hf_xet for faster downloads Debanjum 2025-05-17 02:37:20 -07:00
  • d867dca310 Fix send_message_to_model_wrapper by using sync is_user_subscribed check Debanjum 2025-05-17 02:29:29 -07:00
  • a4ab498aec Update README for better contributions (#1170) Sajjad Baloch 2025-05-12 20:51:01 +05:00
  • 2feed544a6 Add Gemini 2.0 flash back to default gemini chat models list Debanjum 2025-05-11 19:05:09 -06:00
  • 2e290ea690 Pass conversation history to generate non-streaming chat model responses Debanjum 2025-05-07 19:16:21 -06:00
  • 8787586e7e Dedupe code to format messages before sending to appropriate chat model Debanjum 2025-05-07 19:09:44 -06:00
  • e94bf00e1e Add cancellation support to research mode via asyncio.Event Debanjum 2025-04-07 21:03:05 +05:30
  • 1572781946 Parse and show reasoning model thoughts (#1172) Debanjum 2025-05-02 20:29:38 -06:00
  • 2cd7302966 Parse Grok reasoning model thoughts returned by API Debanjum 2025-05-02 19:57:49 -06:00
  • 8cadb0dbc0 Parse Anthropic reasoning model thoughts returned by API Debanjum 2025-05-02 19:08:57 -06:00
  • ae4e352b42 Fix formatting to use Deepseek reasoner for completion via OpenAI API Debanjum 2025-05-02 18:38:41 -06:00
  • 61a50efcc3 Parse DeepSeek reasoning model thoughts served via OpenAI compatible API Debanjum 2025-05-02 18:36:24 -06:00
  • 16f3c85dde Handle thinking by reasoning models. Show in train of thought on web client Debanjum 2025-05-02 06:41:50 -06:00
  • d10dcc83d4 Only enable reasoning by qwen3 models in deepthought mode Debanjum 2025-05-02 10:49:42 -06:00
  • 6eaf54eb7a Parse Qwen3 reasoning model thoughts served via OpenAI compatible API Debanjum 2025-05-02 10:33:42 -06:00
  • 7b9f2c21c7 Parse thoughts from thinking models served via OpenAI compatible API Debanjum 2025-05-02 10:29:36 -06:00
  • 6843db1647 Use conversation specific chat model to respond to free tier users Debanjum 2025-05-02 17:24:40 -06:00
  • 5b5efe463d Remove inline base64 images from webpages read with Firecrawl Debanjum 2025-05-02 14:06:24 -06:00
  • 559b323475 Support attaching jupyter/ipython notebooks from the web app to chat Debanjum 2025-04-26 13:29:06 +02:00
  • dab6977fed add number 1 repo of day badge sabaimran 2025-04-23 16:49:12 -07:00
  • 964a784acf Release Khoj version 1.41.0 Debanjum 2025-04-23 19:01:27 +05:30
  • 23dae72420 Update default models: Gemini models to 2.5 series, Gpt 4o to 4.1 Debanjum 2025-04-23 18:36:12 +05:30
  • d84a0f6e2c Use latest node base image to build web app for khoj docker image Debanjum 2025-04-23 17:43:30 +05:30
  • dd46bcabc2 Track gpt-4.1 model costs. Set prompt size of new gemini, openai models Debanjum 2025-04-23 17:31:29 +05:30
  • 87262d15bb Save conversation to DB in the background, as an asyncio task Debanjum 2025-04-22 17:15:53 +05:30
  • f929ff8438 Simplify AI Chat Response Streaming (#1167) Debanjum 2025-04-21 14:28:02 +05:30
  • a4b5842ac3 Remove ThreadedGenerator class, previously used to stream chat response Debanjum 2025-04-21 14:15:06 +05:30
  • 763fa2fa79 Refactor Offline chat response to stream async, with separate thread Debanjum 2025-04-20 03:42:04 +05:30
  • 932a9615ef Refactor Anthropic chat response to stream async, no separate thread Debanjum 2025-04-20 03:18:32 +05:30
  • a557031447 Refactor Gemini chat response to stream async, no separate thread Debanjum 2025-04-20 02:54:56 +05:30
  • 0751f2ea30 Refactor Openai chat response to stream async, no separate thread Debanjum 2025-04-19 21:36:45 +05:30
  • c93c0d982e Create async get anthropic, openai client funcs, move to reusable package Debanjum 2025-03-24 15:14:19 +05:30
  • 973aded6c5 Fix system prompt to make openai reasoning models md format response Debanjum 2025-04-20 20:09:24 +05:30
  • 21d19163ba Just pass user rather than whole request object to doc search func Debanjum 2025-04-19 20:49:10 +05:30
  • b2390fa977 Allow attaching typescript files to chat on web app Debanjum 2025-04-11 09:51:25 +05:30
  • 4d331e5ad2 Bump documentation dependencies Debanjum 2025-04-19 18:37:48 +05:30
  • d6aafef464 Fix formatting of FAQ section in README.md Debanjum 2025-04-19 18:31:16 +05:30
  • 8f9090940b Resolve datetime utcnow deprecation warnings (#1164) Debanjum 2025-04-19 18:26:52 +05:30
  • 5441793a10 Allow AI model switching based on User Tier (#1151) Debanjum 2025-04-19 18:14:37 +05:30
  • ab29ffd799 Fix web app packaging for pypi since upgrade to python 3.11.12 in CI Debanjum 2025-04-19 17:53:00 +05:30
  • 79fc911633 Enable free tier users to switch between free tier AI models Debanjum 2025-04-01 11:54:09 +05:30
  • 30570e3e06 Track Price tier for each Chat, Speech, Image, Voice AI model in DB Debanjum 2025-04-01 11:41:47 +05:30
  • fdaf51f0ea Fix formatting in readme and documentation Debanjum 2025-04-18 22:39:27 +05:30
  • fee1d3682b Resolve datetime deprecation warnings Emmanuel Ferdman 2025-04-18 10:41:16 -07:00
  • eb1406bcb4 Support deepthought in research mode with new Gemini 2.5 reasoning model Debanjum 2025-04-18 14:19:45 +05:30
  • f95173bb0a Support deepthought in research mode with new Grok 3 reasoning model Debanjum 2025-04-18 14:57:45 +05:30
  • 9c70a0f3f5 Support recently released Openai reasoning models Debanjum 2025-04-18 14:21:07 +05:30
  • 2f8283935a Warn and drop empty messages when format messages for Anthropic Debanjum 2025-04-13 06:49:50 +05:30
  • 51e19c6199 Simplify KHOJ_DOMAIN states. All production deployments should set it. Debanjum 2025-04-13 06:47:42 +05:30
  • e072530471 Deduplicate images generated using the e2b code tool Debanjum 2025-04-12 17:58:50 +05:30
  • dc398d30f8 Add FAQ section to Readme and Troubleshooting Tips to setup GCP Vertex (#1158) RIKIN BRIGHT 2025-04-15 03:49:44 +01:00
  • 6a30da3e9e Fix default state for tools in the agent settings for the chat sidebar sabaimran 2025-04-11 11:12:22 -07:00
  • 2470eea421 Release Khoj version 1.40.0 Debanjum 2025-04-11 18:10:56 +05:30
  • d0a933b072 Add email based rate limiting to email login API endpoint Debanjum 2025-04-06 15:14:06 +05:30
  • fe308c2911 Handle scenario where no valid otps for selected users on admin panel Debanjum 2025-04-11 17:21:16 +05:30
  • 02a6ce9f14 Upgrade server django dependencies Debanjum 2025-04-11 17:48:31 +05:30
  • d84a0abb7f Fix and Improve Chat sidebar and component setup on Web App (#1157) Debanjum 2025-04-11 16:12:03 +05:30
  • 50b0b8a6e7 Fix typo in Development documentation (#1159) Dmitry 2025-04-11 13:35:40 +03:00
  • f609a2d050 Fix typos in admin panel documentation (#1156) Darya 2025-04-11 04:30:17 +03:00
  • 2935ea52cf Set chatSidebar prompt, Setting name fields to empty str if value null Debanjum 2025-04-09 10:23:23 +05:30
  • aea7b90fec Track if agent modified in chatSidebar to simplify code, fix looping Debanjum 2025-04-09 09:53:07 +05:30
  • e9ee9004fb Suppress spurious dark mode hydration warnings on the web app Debanjum 2025-04-09 09:02:57 +05:30
  • 9ab5ead3ca Set key for chatMessage parent to get UX efficiently updated by react Debanjum 2025-04-09 09:01:55 +05:30
  • 1ad7314fe6 Let only root next.js layout handle html, body tags, not child layouts Debanjum 2025-04-09 09:01:38 +05:30
  • 33665dee50 Dynamically set default agent chat model to server > user > first chat model Debanjum 2025-04-09 18:13:45 +05:30
  • 1eb092010c Fix handling unset response_schema being passed to gemini models Debanjum 2025-04-10 19:44:34 +05:30
  • 5b248e8515 Fix fallback to default agent if none set for conversation Debanjum 2025-04-09 18:16:44 +05:30
  • 4012a6372f Fix pgvector docker image pull by using postgres 15 tag Debanjum 2025-04-08 17:56:47 +05:30
  • 19618605a5 Upgrade documentation packages Debanjum 2025-04-07 20:07:10 +05:30
  • 3fc1435cd1 Fix login to local admin panel without need to set KHOJ_DEBUG Debanjum 2025-04-07 17:08:23 +05:30
  • ed70d2254e Suppress spurious RequestAborted ASGI errors on the admin panel Debanjum 2025-04-07 18:45:32 +05:30
  • 353a4aa597 Upgrade pgvector to use the new official docker image Debanjum 2025-04-07 18:17:27 +05:30
  • 50508d97f9 Disable telemetry only via KHOJ_TELEMETRY_DISABLE environment variable Debanjum 2025-04-07 18:14:47 +05:30
  • 2ae8c62547 Repopulate the client API-key generation section sabaimran 2025-04-07 06:54:47 -07:00
  • 6872817d41 Remove request to set default chat model during interactive init Debanjum 2025-04-07 14:38:55 +05:30
  • 0c257c044e Handle unset response_schema being passed to gemini models Debanjum 2025-04-05 23:11:32 +05:30
  • c1912f8ca7 Default eval to use 10 iterations for research mode Debanjum 2025-04-05 10:09:58 +05:30
  • 645c2bc546 Improve Khoj is ready message Debanjum 2025-04-05 10:09:28 +05:30
  • 6e61ec64a4 Release Khoj version 1.39.0 Debanjum 2025-04-04 20:35:50 +05:30
  • 751215a701 Improve response quality with Gemini. Improve evaluation harness (#1150) Debanjum 2025-04-04 20:17:36 +05:30