[Optimized version coming soon]
pip install -r requirements.txt
Once the requirements are installed, download the eval datasets i.e the "dataset" folder from https://github.com/AGI-Edgerunners/LLM-Adapters into the LLM-Adapters directory.
./run_commonsense.sh
Is configured to run Gemma-2B models on Commonesense-15K dataset.
Evaluation is done by running,
python3 multi_dataset_eval.py
First, download the MetaMathQA dataset into the data/train
directory. Then download the MetaMathQA-40K dataset
cd ./data/train
wget https://huggingface.co/datasets/meta-math/MetaMathQA-40K/resolve/main/MetaMathQA-40K.json
To run experiments on Pythia models,
./run_pythia.sh
For other models, run,
./run_math.sh
which is currently configured to run Gemma-2B with SVFT.
run_math.sh
also contains an example to run evaluation on GSM-8K and MetaMath-40K.
For the vision experiments, see the ReadMe file in the vision experiments folder
@misc{lingam2024svft,
title={SVFT: Parameter-Efficient Fine-Tuning with Singular Vectors},
author={Vijay Lingam and Atula Tejaswi and Aditya Vavre and Aneesh Shetty and Gautham Krishna Gudur and Joydeep Ghosh and Alex Dimakis and Eunsol Choi and Aleksandar Bojchevski and Sujay Sanghavi},
year={2024},
eprint={2405.19597},
archivePrefix={arXiv},
primaryClass={cs.LG}
}