Cohee
f1bc217e79
Expressions: Add WebLLM extension classification
2024-12-21 01:14:50 +02:00
Cohee
66a862f797
Lint the use of data-i18n attributes
2024-12-01 15:10:22 +02:00
Cohee
be9f34ab8a
Fix data-i18n removing options
2024-06-17 11:23:59 +03:00
steve green
5e44403346
fixup translates ( #2382 )
...
* Update i18n.js
* fixup
* update i18n CI
* fix trigger?
* i18n changes
* Update zh-cn.json
* add missing keys
* Revert "i18n changes"
This reverts commit ebe0ede6e1ac923c8bd614a845836b1762918301@qq.com >
2024-06-16 14:39:46 +03:00
steve green
3ede4aafbe
Update zh-cn.json ( #2376 )
...
* Update zh-cn.json
* html changes
* `div`->`span`
* fixes
* fixes
* more fixes
* Revert last divs
---------
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com >
2024-06-14 22:46:40 +03:00
Cohee
dc8530049f
Reference implementation: Set expressions with function calling
2024-05-25 15:38:32 +03:00
deffcolony
0252b21901
move expression buttons above Sprite set
...
+moves the buttons above the Sprite set so user does not have to scroll all the way down to find the buttons now it will be directly visiable
2024-05-09 16:33:04 +02:00
Cohee
bc9c70556e
Clean-up mentions of /public/
2024-04-21 23:53:46 +03:00
Cohee
0ff5d0b5f1
Substitute macro is LLM prompt
2024-04-14 19:53:40 +03:00
Cohee
3e60919289
Specify LLM prompt in case JSON schema is not supported
2024-04-14 17:13:54 +03:00
kingbri
bea63a2efe
Expressions: Remove TalkingHead from choices
...
This is a sub-choice under extras.
Signed-off-by: kingbri <bdashore3@proton.me >
2024-04-12 01:59:44 -04:00
kingbri
6b656bf380
Expressions: Classify using LLM
...
Rather than using a separate BERT model to classify the last message,
use the LLM itself to get the classified expression label as a JSON
and set that as the current sprite. Doing this should take more information
into consideration and cut down on extra processing.
This is made possible by the use of constrained generation with JSON
schemas. Only available to TabbyAPI since it's the only backend that
supports the use of JSON schemas, but there can hopefully be a way
to use this with other backends as well.
Intercepts the generation and sets top_k = 1 (for greedy sampling)
and the json_schema to an emotion enum. Doing this also prevents
reingestion of the entire context every time a message is sent and
then asked to be classified, which doesn't compromise the chat
experience.
Signed-off-by: kingbri <bdashore3@proton.me >
2024-04-12 01:55:16 -04:00
Cohee
6290dff3d9
#2065 Add option to translate text before classification
2024-04-11 22:09:05 +03:00
Wolfsblvt
2ffb44b4e1
Add extension setting for fallback expression
2024-04-07 06:12:52 +02:00
Cohee
b24509ef43
Visual touchup of custom expressions
2023-09-14 21:41:30 +03:00
Cohee
7553efc308
Custom char expressions
2023-09-14 21:30:02 +03:00
Cohee
77a28c7131
Don't hide expressions menu if no chat open
2023-09-13 15:19:10 +03:00
Cohee
7f55d108cf
Don't use talkinghead with local classification model
2023-09-11 13:01:45 +03:00
Cohee
967a084aad
(WIP) Local emotion classification pipeline
2023-09-09 15:14:16 +03:00
Cohee
409eef6e46
Extract expressions HTML to templates
2023-08-25 23:48:59 +03:00
