predict_eval_script.sh

for BST_EXP_MDL in <best_model_folder/best_ckpt>; do
    for test_set in all senseval2 senseval3 semeval2007 semeval2013 semeval2015; do
        echo Testing model ${BST_EXP_MDL} on ${test_set} test-set
        CKPT_DIR="${BST_EXP_MDL%%/*}"
        echo ${CKPT_DIR}
        CUDA_VISIBLE_DEVICES="" python predict.py --processor checkpoints/${CKPT_DIR}/processor_config.json \
            --model checkpoints/${BST_EXP_MDL} \
            --model_input data/preprocessed/${test_set}/${test_set}.json \
            --model_output output.${CKPT_DIR}.txt;
        CUDA_VISIBLE_DEVICES="" python evaluate.py --processor checkpoints/${CKPT_DIR}/processor_config.json \
            --model checkpoints/${BST_EXP_MDL} \
            --model_input data/preprocessed/${test_set}/${test_set}.json \
            --model_output output.${CKPT_DIR}.gold.txt \
            --evaluation_input data/original/${test_set}/${test_set}.gold.key.txt
    done
    echo "------------------------------------------------------------------------------------"
done