Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Research Mode [Part 2]: Improve Prompts, Edit Chat Messages. Set LLM Seed for Reproducibility #954

Merged
merged 36 commits into from
Nov 2, 2024
Merged

Research Mode [Part 2]: Improve Prompts, Edit Chat Messages. Set LLM Seed for Reproducibility #954

merged 36 commits into from
Nov 2, 2024

Conversation

debanjum
Copy link
Member

@debanjum debanjum commented Nov 1, 2024 •
edited
Loading

  • Improve chat actors and their prompts for research mode.
  • Add documentation to enable the code tool when self-hosting Khoj
  • Edit Chat Messages
    • Store Turn Id in each chat message.
    • Expose API to delete chat message.
    • Expose delete chat message button to turn delete chat message from web app
  • Set LLM Generation Seed for Reproducible Debugging and Testing
    • Setting seed for LLM generation is supported by Llama.cpp and OpenAI models.
      This can (somewhat) restrain LLM output
    • Getting fixed responses for fixed inputs helps test, debug longer reasoning chains like used in advanced reasoning

debanjum and others added 30 commits October 30, 2024 14:00
@debanjum
Make cursor in chat input take on selected agent color
2ac840e
@debanjum
Defer turning cursor color to selected agents color for later
358a6ce 
Capability exists but idea needs to be investigated further
@debanjum
Improve handling embedding model config from admin interface
d44e68b 
- Allow server to start if loading embedding model fails with an error.
  This allows fixing the embedding model config via admin panel.

  Previously server failed to start if embedding model was configured
  incorrectly. This prevented fixing the model config via admin panel.

- Convert boolean string in config json to actual booleans when passed
  via admin panel as json before passing to model, query configs

- Only create default model if no search model configured by admin.
  Return first created search model if its been configured by admin.
@debanjum
Support setting seed for reproducible LLM response generation
b3a6301 
Anthropic models do not support seed. But offline, gemini and openai
models do. Use these to debug and test Khoj via KHOJ_LLM_SEED env var
@debanjum
Handle add/delete file filter operation on non-existent conversation
f64f5b3
@debanjum
Store turn id with each chat message. Expose API to delete chat turn
ba15686 
Each chat turn is a user query, khoj response message pair
@debanjum
Add ability to delete messages from the web app
ca5a683
@debanjum
Resolve train of thought component needs unique key id error on web app
cb90abc
@debanjum
Fix deleting new messages generated after conversation load
e8e6ead
@debanjum
Put train of thought ui before Khoj response on web app
e17dc9f
@sabaimran
Only show trash can when turnId is present
a137606
@sabaimran
Do not exit if/else loop in research loop when notes not found
559601d
@debanjum
Only add /research prefix in research mode if not already in user query
5b15176
@debanjum
Json dump contents in prompt tracer to make structure discernable
89597ae
@debanjum
Improve research planner prompt to reduce looping
52163fe
@debanjum
Improve online chat actor prompt for research and normal mode
302bd51 
- Match the online query generator prompt to match the formatting of
  extract questions
- Separate iteration results by newline
- Improve webpage and online tool descriptions
@sabaimran
Merge branch 'master' of github.com:khoj-ai/khoj into improve-debug-r…
1924180 
…easoning-and-other-misc-fixes
@sabaimran
Remove conversation command always in query, filter out inferred quer…
21858ac 
…ies that were not with selected tool when going through tool selection iterations
@sabaimran
Use bottom anchor for the commandbar popover
149cbe1
@sabaimran
Only include inferred-queries in chat history when present
3ea94ac
@sabaimran
Set usage limits on the research mode
0145b2a
@sabaimran
Add experimental notice to research mode tooltip
33d36ee
@sabaimran
Handle case where infer_webpage_url returns no valid urls
1fc280d
@sabaimran
Add template for a code sandbox to the docker-compose configuration
ffa7f95
@sabaimran
Add documentation for python code execution capability
23a49b6
@sabaimran
Standardize rate limits to 1/6 ratio
b3dad1f
@sabaimran
Merge branch 'master' of github.com:khoj-ai/khoj into improve-debug-r…
2b35790 
…easoning-and-other-misc-fixes
@debanjum
Simplify logic to get default search model. Remove unused import
ac21b10
@debanjum
Use standard per minute rate limits across user types
9c7b36d
@sabaimran
Clarify description of the code evaluation environment: not for docum…
b79a9ec 
…ent creation
@debanjum debanjum force-pushed the improve-debug-reasoning-and-other-misc-fixes branch from fbc753c to 14e4530 Compare November 2, 2024 01:12
@debanjum debanjum merged commit cff8e02 into master Nov 2, 2024
8 of 9 checks passed
@debanjum debanjum deleted the improve-debug-reasoning-and-other-misc-fixes branch November 2, 2024 01:17
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@debanjum @sabaimran