search

yashsmehta / personality-prediction

Experiments for automated personality detection using Language Models and psycholinguistic features on various famous personality datasets including the Essays dataset (Big-Five)
MIT License
186 stars 51 forks source link

Predicting unseen text #23

Closed edersoncorbari closed 1 year ago

edersoncorbari commented 1 year ago

Prepared for finetune models: MLP_LM and SVM_LM.

Tests performed with the essays dataset:

python LM_extractor.py -dataset_type 'essays' -token_length 512 -batch_size 32 -embed 'bert-base' -op_dir 'pkl_data'

python finetune_models/MLP_LM.py -save_model 'yes' python finetune_models/SVM_LM.py -save_model 'yes'

python utils/predictor.py -finetune_model 'mlp_lm' python utils/predictor.py -finetune_model 'svm_lm'

With the kaggle dataset:

python LM_extractor.py -dataset_type 'kaggle' -token_length 512 -batch_size 32 -embed 'bert-base' -op_dir 'pkl_data'

python finetune_models/MLP_LM.py -dataset 'kaggle' -save_model 'yes' python finetune_models/SVM_LM.py -dataset 'kaggle' -save_model 'yes'

python utils/predictor.py -dataset 'kaggle' -finetune_model 'mlp_lm' python utils/predictor.py -dataset 'kaggle' -finetune_model 'svm_lm'