You can deploy this model and start earning money today!

kluvin/
bertweet-tweet-sentiment

Name: bertweet-tweet-sentiment
Availability: InStock

safetensors

roberta

Views

No views yet

BERTweet Tweet Sentiment Classifier

Fine-tuned vinai/bertweet-base for 3-class sentiment analysis on tweets (negative, neutral, positive).

Model Details

Base Model: vinai/bertweet-base
Task: Text Classification (Sentiment Analysis)
Dataset: SetFit/tweet_sentiment_extraction
Languages: English (Twitter/Social Media)

Training Details

Data Split

Train: 70% (19,236 samples)
Validation: 15% (4,122 samples)
Test: 15% (4,123 samples)
Stratified sampling to preserve class distribution

Hyperparameters

Learning rate: 1e-5
LR scheduler: Cosine with warmup
Batch size: 16 (train), 32 (eval)
Epochs: 2
Weight decay: 0.01 (L2 regularization)
Max sequence length: 64 tokens
Optimizer: AdamW

Training Infrastructure

Mixed precision training (fp16/bf16)
Dynamic padding with pad_to_multiple_of=8 for Tensor Core optimization
Early stopping based on validation accuracy

Usage

python

from transformers import pipeline

# Load the model
classifier = pipeline("sentiment-analysis", model="kluvin/bertweet-tweet-sentiment")

# Make predictions
result = classifier("I love this product!")
print(result)
# [{'label': 'positive', 'score': 0.95}]

Advanced Usage

python

from transformers import AutoModelForSequenceClassification, AutoTokenizer
import torch

model = AutoModelForSequenceClassification.from_pretrained("kluvin/bertweet-tweet-sentiment")
tokenizer = AutoTokenizer.from_pretrained("kluvin/bertweet-tweet-sentiment")

text = "This is the worst day ever"
inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=64)

with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.nn.functional.softmax(outputs.logits, dim=-1)

print(predictions)

Limitations

Trained specifically on Twitter/social media text with informal language
May not generalize well to formal text or other domains
Limited to English tweets

License

Inherits license from base model (vinai/bertweet-base): MIT

Citation

If you use this model, please cite the original BERTweet paper:

bibtex

@inproceedings{bertweet,
  title={BERTweet: A pre-trained language model for English Tweets},
  author={Nguyen, Dat Quoc and Vu, Thanh and Nguyen, Anh Tuan},
  booktitle={Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations},
  pages={9--14},
  year={2020}
}