Fine-Tuned DistilBERT for Sentiment Analysis

Model Information

• Base Model: distilbert-base-uncased
• Task: Sequence Classification (Sentiment Analysis)
• Number of Labels: 4
• Labels: Irrelevant, Negative, Neutral, Positive

Training Metrics

• Test Accuracy: 0.8956 (89.56%)
• Test Precision: 0.8957
• Test Recall: 0.8956
• Test F1-Score: 0.8956

Training Configuration

• Epochs: 3
• Batch Size: 16
• Learning Rate: 2e-05
• Max Sequence Length: 128
• Optimizer: AdamW
• Weight Decay: 0.01

Dataset

• Source: Kaggle - Twitter Entity Sentiment Analysis
• Training Samples: 59058
• Validation Samples: 7382
• Test Samples: 7383

Usage

Files

• pytorch_model.bin: Model weights
• config.json: Model configuration
• tokenizer.json: Tokenizer configuration
• special_tokens_map.json: Special tokens mapping
• vocab.txt: Vocabulary file
• label_mappings.json: Label ID mappings