KimJammer
|
c7afcadcb2
|
Activate multimodality, Refactor llmWrapper
llmWrappers are now subclasses of abstractLLMWrapper, so different llms can be used seamlessly.
|
2024-06-11 23:16:21 -07:00 |
|
KimJammer
|
a786696082
|
Auto send more data on frontend connection
|
2024-05-26 16:34:48 -07:00 |
|
KimJammer
|
2a39603a12
|
Finish Memory Fetching
Also fixed blacklist bug filtering out words that contain blacklisted words.
|
2024-05-22 23:00:36 -07:00 |
|
KimJammer
|
f1bb771dcb
|
Start work on Memories
Implement backend & frontend for ChromaDB persistent memories and RAG.
Also added live-customizable prompt section.
|
2024-05-20 23:35:35 -07:00 |
|
KimJammer
|
b7894c2785
|
Implement Vtube Studio Plugin
|
2024-05-08 20:26:17 -04:00 |
|
KimJammer
|
228314b7e2
|
Implement song playback and chat history deleting
|
2024-04-13 20:16:54 -04:00 |
|
KimJammer
|
1f93c1e192
|
Environment upgrade & Modules rework
Python to 3.11
Torch to 2.2.2
Deepspeed updated to match new environment
Modules rework so that its easier to inject text into the prompt
|
2024-04-09 11:15:03 -04:00 |
|
KimJammer
|
3c3404d301
|
Better Termination Handling, New LLM Prompter
Program *should* terminate more gracefully with terminate flag.
New LLM prompter using the completions endpoint for greater control
|
2024-03-17 19:18:30 -04:00 |
|
KimJammer
|
f5454222a1
|
Major refactor
Switched back to multithreading, but smartly this time.
|
2024-03-10 01:09:26 -05:00 |
|
KimJammer
|
d9a1bffe30
|
Final version before refactor to single thread
|
2024-03-09 22:25:07 -05:00 |
|
KimJammer
|
951753d33e
|
Proof of Concept Frontend communication
|
2024-03-02 23:59:33 -05:00 |
|