Sarcasm Detection Model
This model is a fine-tuned version of sadia72/roberta-base-finetuned-sarcasm-news-headline-detection for sarcasm detection.
Model Description
This model was trained on a dataset for detecting sarcasm in text. It achieves state-of-the-art performance on this specific task.
Intended Uses & Limitations
- Intended use: This model is intended to detect sarcasm in text.
- Limitations: The model may not perform well on text that is very different from the training data.
Training and Evaluation Data
- Training Data: The model was trained on a dataset of sarcastic and non-sarcastic tweets.
- Evaluation Data: The model was evaluated on a held-out test set with similar characteristics as the training set.
Training Procedure
Training Hyperparameters
- Optimizer: AdamWeightDecay with a learning rate of 5e-4
- Batch Size: 18
- Epochs: 10
- Loss Function: Sparse Categorical Crossentropy
Training Results
Here are the results of training:
- Training Accuracy: 0.99615478515625
- Validation Accuracy: 0.9029778838157654
- Training Loss: 0.015514880418777466
- Validation Loss: 0.4263628423213959
Framework Versions
- Transformers: 4.42.3
- TensorFlow: 2.15.0
- Datasets: 2.20.0
- Tokenizers: 0.19.1
- Downloads last month
- 4