mmarch/text-generation-webui
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Mixtral Instruct: detect prompt format for llama.cpp loader

Browse source 
Workaround until the tokenizer.chat_template kv field gets implemented
This commit is contained in:
oobabooga 2023-12-15 06:58:29 -08:00
parent 3bbf6c601d
commit a060908d6c
2 changed files with 2 additions and 11 deletions
Download patch file Download diff file Expand all files Collapse all files

2
models/config.yaml
View file

@ -174,7 +174,7 @@
instruction_template: 'OpenChat'
.*codellama.*instruct:
instruction_template: 'Llama-v2'
.*mistral.*instruct:
.*(mistral|mixtral).*instruct:
instruction_template: 'Mistral'
.*mistral.*openorca:
instruction_template: 'ChatML'