This Repo is not for use sLLM, just Research and Developing sLLM logics
all files doesn't include model like safetensor type file or json
just only code and jupyter notebook's output results
if you try some fine-tuning in sLLM using GPU(Nvidia-A100), you can follow this 3steps.
load dataset and tokenizer and model
load dataset need library "datasets"
if you install "datasets" yet. you can follow this prompt "pip3 install datasets"
you must make 3 arguments.
- training argument
- tokenizer argument
- evaluate argument
Train function use like this model_trainer.train()