This is my environment I used for the ICML project; there can be some redundant modules :/
conda env create -f environment.yml
Datasets are in "LLMForEM/data/". So you do not need to download data on your own!
To run base setting on the abt-buy dataset,
CUDA_VISIBLE_DEVICES=0 python main.py --data abt-buy --strategy base
To run similarity baseline on the abt-buy dataset,
CUDA_VISIBLE_DEVICES=0 python main.py --data abt-buy --strategy similarity
For Debug mode, use the "--debug" argument. This will reduce the number of data samples to 50, to speed up coding :)
CUDA_VISIBLE_DEVICES=0 python main.py --data abt-buy --strategy similarity --debug