FNU Hardeep, Harish Babu Manogaran, Mridul Khurana, Shri Sarvesh Venkatachala Moorthy and Upasana Sivaramakrishnan
Fall 2022 CS 5824/ECE 5424 Advanced Machine Learning:
Course Project - Virginia Tech
transformers=4.14.1
torch=1.8
vocab=0.0.5
numpy
tqdm
sklearn
pickle
pandas
The Following datasets are available at data/
data/
└───meld/
│ │ dev_sent_emo.csv
│ │ test_sent_emo.csv
| | train_sent_emo.csv
│
└───emorynlp/
│ │ emorynlp_dev_final.csv
│ │ emorynlp_test_final.csv
| | emorynlp_train_final.csv
│
└───iemocap/
│ │ iemocap_dev.txt
│ │ iemocap_test.txt
| | iemocap_train.txt
|
└───daily_dialog/
│ │ dailydialog_dev.txt
│ │ dailydialog_test.txt
| | dailydialog_train.txt
|
python train.py -tr -wp 0 -bsz 1 -acc_step [2, 8] -lr 1e-4 -ptmlr 1e-5 -dpt 0.3 -bert_path roberta-[base, large] -epochs [20, 5]
python train.py -tr -wp 0 -bsz 1 -acc_step [2, 8] -lr 1e-4 -ptmlr 1e-5 -dpt 0.3 -bert_path -tsk emorynlp roberta-[base, large] -epochs [20, 5]
python train_iemocap.py -tr -wp 0 -bsz 1 -acc_step [2, 8] -lr 1e-4 -ptmlr 1e-5 -dpt 0.3 -bert_path roberta-[base, large] -epochs [20, 5]
python train_dailydialog.py -tr -wp 0 -bsz 1 -acc_step [2, 8] -lr 1e-4 -ptmlr 1e-5 -dpt 0.3 -bert_path roberta-[base, large] -epochs [20, 5]
Similarly for training without the speaker information the training scripts train_meld_emorynlp_without_speaker.py and train_iemocap_without_speaker.py.
Please download the model checkpoint from the link given below and keep it in this directory. Following script will run only on a GPU.
python inference.py -tr -wp 0 -bsz 1 -acc_step 8 -lr 1e-4 -ptmlr 1e-5 -dpt 0.3 -bert_path roberta-base -epochs 2 -postfix iemocap_inf -tsk iemocap -ft
| model | Checkpoint |
|---|---|
| Iemocap-roberta-base | iemocap-roberta-base.pkl |
Checkpoints are produced on a single A100 GPU.