distilbert-base-uncased-reviews_multilabel_clf_v2

This model is a fine-tuned version of distilbert-base-uncased. It achieves the following results on the evaluation set:

Loss: 0.1519
F1: 0.8697
Roc Auc: 0.9107
Accuracy: 0.5787

Model description

This is a multilabel classification model of whether different aspects of a product are mentioned in reviews.

For more information on how it was created, check out the following link: https://github.com/DunnBC22/NLP_Projects/blob/main/Multilabel%20Classification/Review%20Sentiments/Sentiments%20-%20Multilabel%20clf.ipynb

Intended uses & limitations

This model is intended to demonstrate my ability to solve a complex problem using technology.

Training and evaluation data

Dataset Source: https://www.kaggle.com/datasets/mohamedziauddin/mh-uhack-sentiments?select=train.csv

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2e-05
train_batch_size: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.6847	1.0	305	0.2425	0.7619	0.8209	0.3492
0.296	2.0	610	0.1786	0.8447	0.8847	0.5197
0.296	3.0	915	0.1634	0.8511	0.8937	0.5361
0.1476	4.0	1220	0.1544	0.8626	0.8999	0.5623
0.0986	5.0	1525	0.1490	0.8624	0.8994	0.5639
0.0986	6.0	1830	0.1521	0.8653	0.9041	0.5787
0.0686	7.0	2135	0.1511	0.8676	0.9110	0.5656
0.0686	8.0	2440	0.1501	0.8687	0.9104	0.5869
0.0525	9.0	2745	0.1519	0.8685	0.9089	0.5754
0.0432	10.0	3050	0.1519	0.8697	0.9107	0.5787

Framework versions

Transformers 4.26.1
Pytorch 1.12.1
Datasets 2.8.0
Tokenizers 0.12.1