distilbert-base-uncased-finetuned-emotions-text-classification

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

• Loss: 0.2130
• Accuracy: 0.928
• F1: 0.9278

Model description

More information needed

Intended uses & limitations

import pandas as pd
from transformers import pipeline
import pandas as pd
import matplotlib.pyplot as plt


model_id = 'ithattieu/distilbert-base-uncased-finetuned-emotions-text-classification'
classifier = pipeline('text-classification', model=model_id)

labels = ['sadness', 'joy', 'love', 'anger', 'fear', 'surprise']

text = 'I am not a really smart lady, which makes it challenging to grasp certain concepts, but you made it super easy to understand NLP and relevant applications. So, thank you!'

preds_df = pd.DataFrame(preds[0])
preds = classifier(text, return_all_scores=True)

preds_df = pd.DataFrame(preds[0])
plt.bar(labels, 100 * preds_df["score"], color='C0')
plt.title(f'"{text}"')
plt.ylabel("Class probability (%)")
plt.show()

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.8161	1.0	250	0.3051	0.912	0.9114
0.2464	2.0	500	0.2130	0.928	0.9278

Framework versions

• Transformers 4.41.2
• Pytorch 2.5.0.dev20240815
• Datasets 2.19.1
• Tokenizers 0.19.1