55fddbc37b
Format Python code with psf/black push
2 years ago
3952842856
Add setting top_k amount to the query command
...
Count LLM tokens for deep compose
Update presentation of cost in the console
Update readme
Set a prompthelper at 500 tokens output up from the default 256 I believe. ~4 characters per token so the discord limit is around 500 tokens.
2 years ago
5a1bd48d94
Fix regression in full discord server indexing
...
Add back response mode, only on vector index queries
Added the llmpredictor to queries
Fix docker build
Update requirements
2 years ago
73390b262c
Format Python code with psf/black push
2 years ago
319ae34318
composability improvements
2 years ago
fd2ab04d93
More composability
2 years ago
258a87945c
composability
2 years ago
c93bcfa59a
update readme
2 years ago
464f332915
csv, powerpoint, image, mp3 support
2 years ago
3ed55b556f
Support youtube videos
2 years ago
b60bcfd68d
reset indexes command
2 years ago
22ba80c469
Multi-index support
2 years ago
c7e6a447d3
Role checks
2 years ago
e63c6227dd
initial link loading
2 years ago
5d406b666b
remove .cleanup() for temp_path (it gets auto cleaned up its ok
2 years ago
842c7241a3
Removed some imports
2 years ago
c6a6245e00
Added server index backup and loading
2 years ago
dd9cb0ce4c
Add option for response mode on queries
2 years ago
37a20a7e37
Added discord channel and whole server indexing
...
A channel index does 1000 messages
Whole server index does 300 messages per channel
Might need to add more restrictions since all the index commands are expensive
2 years ago
64ce627bb7
Update filetypes
2 years ago
0555cf8b33
Some async changes
2 years ago
9052bc2d80
Added gpt-index commands
2 years ago
github-actions
Rene Teigen
Kaveen Kumarasinghe