<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
twitter-prosusai-finbert-sentiment-finetuned-memes-final
This model is a fine-tuned version of jayantapaul888/twitter-data-prosusai-finbert-sentiment-finetuned-memes on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.9363
- Accuracy: 0.8163
- Precision: 0.8166
- Recall: 0.8163
- F1: 0.8164
Model description
More information needed
Intended uses & limitations
More information needed
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: 6
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|---|---|
| No log | 1.0 | 294 | 0.3954 | 0.8198 | 0.8215 | 0.8198 | 0.8200 |
| 0.4754 | 2.0 | 588 | 0.4318 | 0.8203 | 0.8270 | 0.8203 | 0.8204 |
| 0.4754 | 3.0 | 882 | 0.5372 | 0.8230 | 0.8230 | 0.8230 | 0.8230 |
| 0.192 | 4.0 | 1176 | 0.7378 | 0.8196 | 0.8198 | 0.8196 | 0.8195 |
| 0.192 | 5.0 | 1470 | 0.8747 | 0.8168 | 0.8176 | 0.8168 | 0.8170 |
| 0.0726 | 6.0 | 1764 | 0.9363 | 0.8163 | 0.8166 | 0.8163 | 0.8164 |
Framework versions
- Transformers 4.24.0.dev0
- Pytorch 1.11.0+cu102
- Datasets 2.6.1
- Tokenizers 0.13.1
