 oobabooga
|
cacbcda208
Two new options: truncation length and ban eos token
|
2 éve |
 catalpaaa
|
78bbc66fc4
allow custom stopping strings in all modes (#903)
|
2 éve |
 IggoOnCode
|
09d8119e3c
Add CPU LoRA training (#938)
|
2 éve |
|
oobabooga
|
bd04ff27ad
Make the bos token optional
|
2 éve |
|
oobabooga
|
0f1627eff1
Don't treat Intruct mode histories as regular histories
|
2 éve |
 MarkovInequality
|
992663fa20
Added xformers support to Llama (#950)
|
2 éve |
|
oobabooga
|
ea6e77df72
Make the code more like PEP8 for readability (#862)
|
2 éve |
 SDS
|
378d21e80c
Add LLaMA-Precise preset (#767)
|
2 éve |
|
oobabooga
|
e722c240af
Add Instruct mode
|
2 éve |
|
oobabooga
|
65d8a24a6d
Show profile pictures in the Character tab
|
2 éve |
|
oobabooga
|
b24147c7ca
Document --pre_layer
|
2 éve |
|
oobabooga
|
4c9ed09270
Update settings template
|
2 éve |
 OWKenobi
|
dcf61a8897
"character greeting" displayed and editable on the fly (#743)
|
2 éve |
|
oobabooga
|
b0890a7925
Add shared.is_chat() function
|
2 éve |
|
oobabooga
|
b857f4655b
Update shared.py
|
2 éve |
|
oobabooga
|
2c52310642
Add --threads flag for llama.cpp
|
2 éve |
|
oobabooga
|
1d1d9e40cd
Add seed to settings
|
2 éve |
|
oobabooga
|
d4a9b5ea97
Remove redundant preset (see the plot in #587)
|
2 éve |
|
oobabooga
|
55755e27b9
Don't hardcode prompts in the settings dict/json
|
2 éve |
|
oobabooga
|
1cb9246160
Adapt to the new model names
|
2 éve |
|
oobabooga
|
010b259dde
Update documentation
|
2 éve |
|
oobabooga
|
036163a751
Change description
|
2 éve |
|
oobabooga
|
005f552ea3
Some simplifications
|
2 éve |
|
oobabooga
|
fde92048af
Merge branch 'main' into catalpaaa-lora-and-model-dir
|
2 éve |
|
oobabooga
|
49c10c5570
Add support for the latest GPTQ models with group-size (#530)
|
2 éve |
 catalpaaa
|
b37c54edcf
lora-dir, model-dir and login auth
|
2 éve |
|
oobabooga
|
c5ebcc5f7e
Change the default names (#518)
|
2 éve |
|
oobabooga
|
7618f3fe8c
Add -gptq-preload for 4-bit offloading (#460)
|
2 éve |
|
oobabooga
|
ddb62470e9
--no-cache and --gpu-memory in MiB for fine VRAM control
|
2 éve |
|
oobabooga
|
a78b6508fc
Make custom LoRAs work by default #385
|
2 éve |
