hmwk-1.txt

# vs code create folder

# check python versions
python -v # make sure 3.9-3.12
pip -v

# create venv
python -m virtualenv venv 
source venv/bin/activate

# install prodigy, using Prodigy Key as an environmental variable
export PRODIGY_KEY = xxxx-xxxx-xxxx-xxxx
python -m pip install prodigy -f https://$PRODIGY_KEY@download.prodi.gy
python -m prodigy stats

# ngrok website install: https://ngrok.com/download
python -m pip install ngrok
python -m ngrok config add-authtoken <token>

# add canvas files into data/
python ner.manual cooking_manual blank:en data/unlabeled.jsonl --label DISH,INGREDIENT,EQUIPMENT
# optional step if you want to share local app to someone
(New terminal) $ ngrok http 8080

# train
python -m prodigy train —-ner cooking_manual —output ./output

# correct other unlabeled
python -m ner.correct cooking_correct ./output/model-best ./data/unlabeled.jsonl --label DISH,INGREDIENT,EQUIPMENT

# combine datasets
python -m prodigy db.merge cooking_manual,cooking_correct cooking_train

# check how label progressing
python -m prodigy train-curve —-ner cooking_train --label-stats

# retrain
python -m prodigy train —ner cooking_train —output ./output

# zip file
zip ./output/model-best

# install setuptools and wheel
pip install -U pip setuptools wheel

# upload to HF, huggingface.co/settings/tokens
huggingface-cli login
python -m spacy package ./en_ner_cooking ./output --build wheel
cd ./output/en_ner_cooking-0.0.0/dist
python -m spacy huggingface-hub push en_ner_cooking-0.0.0-py3-none-any.whl