metadata
license: apache-2.0
base_model: distilbert-base-cased
tags:
- text-classification
metrics:
- accuracy
model-index:
- name: Sarcastic_01
results: []
language:
- en
Sarcastic_01
This model is a fine-tuned version of distilbert-base-cased on the "News Headlines Dataset For Sarcasm Detection dataset. It achieves the following results on the evaluation set:
- Loss: 0.5868
- Accuracy: 0.9234
Model description
This model was created for my university project on Sarcasm Detection.
Intended uses & limitations
To have better results preprocessing is needed:
- expand_contractions
- remove_punctuation
- tokenize
- remove_stopwords
- lemmatization
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: 32
- eval_batch_size: 32
- 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 | Accuracy |
|---|---|---|---|---|
| 0.3244 | 1.0 | 609 | 0.2266 | 0.9068 |
| 0.1832 | 2.0 | 1218 | 0.2167 | 0.9196 |
| 0.1092 | 3.0 | 1827 | 0.3083 | 0.9211 |
| 0.0645 | 4.0 | 2436 | 0.4030 | 0.9222 |
| 0.0232 | 5.0 | 3045 | 0.4807 | 0.9240 |
| 0.0141 | 6.0 | 3654 | 0.5271 | 0.9208 |
| 0.0092 | 7.0 | 4263 | 0.5352 | 0.9240 |
| 0.0068 | 8.0 | 4872 | 0.5630 | 0.9252 |
| 0.0054 | 9.0 | 5481 | 0.5799 | 0.9240 |
| 0.0023 | 10.0 | 6090 | 0.5868 | 0.9234 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1