Sanity check of running code as give doesn't work even with corrected pip installs

[brando90]

To reproduce do the following:

Install all dependencies

# -- Create conda env for evaporate
conda create --name evaporate python=3.10
# conda create --name evaporate python=3.8
conda activate evaporate

# -- Evaporate code
cd ~
# cd $AFS
# echo $AFS
git clone git@github.com:brando90/evaporate.git
# ln -s /afs/cs.stanford.edu/u/brando9/evaporate $HOME/evaporate
cd ~/evaporate
pip install -e .

# -- Install missing dependencies not in setup.py
pip install tqdm openai manifest-ml beautifulsoup4 pandas cvxpy sklearn scikit-learn snorkel snorkel-metal tensorboardX

# -- Weak supervision code
cd ~/evaporate/metal-evap
git submodule init
git submodule update
pip install -e .

# -- Manifest (to install from source, which helps you modify the set of supported models. Otherwise, ``setup.py`` installs ``manifest-ml``)
cd ~
# cd $AFS
# echo $AFS
git clone git@github.com:HazyResearch/manifest.git
# ln -s /afs/cs.stanford.edu/u/brando9/manifest $HOME/manifest
cd ~/manifest
pip install -e .

# -- Git lfs install is a command used to initialize Git Large File Storage (LFS) on your machine.
git lfs install
cd ~/evaporate/
# cd ~/data
git clone https://huggingface.co/datasets/hazyresearch/evaporate
# get data in python
# from datasets import load_dataset
# dataset = load_dataset("hazyresearch/evaporate")

Run file with your open ai key

# keys="PLACEHOLDER" # INSERT YOUR API KEY(S) HERE
keys=$(cat ~/data/openai_api_key.txt)
# echo $keys

cd ~/evaporate/evaporate
conda activate evaporate
# conda activate maf

# evaporate code closed ie
python ~/evaporate/evaporate/run_profiler.py \
    --data_lake fda_510ks \
    --do_end_to_end False \
    --num_attr_to_cascade 50 \
    --num_top_k_scripts 10 \
    --train_size 10 \
    --combiner_mode ws \
    --use_dynamic_backoff True \
    --KEYS ${keys}

Run run_profiler.py

Error:

(evaporate) brando9@ampere1:~/evaporate$ python ~/evaporate/evaporate/run_profiler.py     --data_lake fda_510ks     --do_end_to_end False     --num_attr_to_cascade 50     --num_top_k_scripts 10     --train_size 10     --combiner_mode ws     --use_dynamic_backoff True     --KEYS ${keys}
Traceback (most recent call last):
  File "/lfs/ampere1/0/brando9/evaporate/evaporate/run_profiler.py", line 14, in <module>
    from profiler import run_profiler
  File "/afs/cs.stanford.edu/u/brando9/evaporate/evaporate/profiler.py", line 33, in <module>
    from run_ws import run_ws
ModuleNotFoundError: No module named 'run_ws'

May we get help?

---

Related issue: https://github.com/HazyResearch/evaporate/issues/21

[brando90]

I have my own fork trying to fix this basic import statement but code still doesn't work yet. Would be nice to release a version of the code that works out of the box at least for reproducing the original results before trying it on our own custom data.