Skip to content

LoRA instruction tuning script for Japanese Language models

License

Notifications You must be signed in to change notification settings

to-aoki/lora_finetuning

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

92 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

LoRA instruction tuning script for Japanese Language models.

Tested

(maybe work on x86 too)

models
rinna gptneox
cyberagent opencalm
meta llama2
line-corporation gptneox
stabilityai stablelm
llm-jp/FugakuLLM
deepseek coder
google gemma
Swallow MS 7b v.01
Swallow MX 8x7b NVE v.01
Phi3-mini

How to work

finetune

nohup python3 lora_finetune.py --model_name="HUGGINGFACE_MODEL_NAME" > logging.out &

test (example line-corporation/japanese-large-lm-3.6b dolly-15k-ja 1epoch)

python3 lora_test.py --slow_tokenizer
:
赤、緑、青
日本で2番目に高い山は、北岳(3,193m)である。
紫式部は、物語や詩文に多くの女性的な要素を取り入れているのに対し、清少納言は、男性的で論理的な文章を多く残している

merge

python3 lora_merge.py 
:
ls final_merged_checkpoint/
config.json             special_tokens_map.json  tokenizer_config.json
generation_config.json  spiece.model
pytorch_model.bin       tokenizer.json

webui

python3 app.py 

img.png

About

LoRA instruction tuning script for Japanese Language models

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages