metadata
license: mit
tags:
- generated_from_trainer
model-index:
- name: verdict-classifier-trinary
results: []
verdict-classifier-trinary
This model is a fine-tuned version of roberta-base on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1258
- F1 Macro: 0.8408
- F1 Misinformation: 0.9751
- F1 Factual: 0.9508
- F1 Other: 0.5965
- Prec Macro: 0.8323
- Prec Misinformation: 0.9818
- Prec Factual: 1.0
- Prec Other: 0.5152
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: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 8
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 462
- num_epochs: 1000
Training results
Training Loss | Epoch | Step | Validation Loss | F1 Macro | F1 Misinformation | F1 Factual | F1 Other | Prec Macro | Prec Misinformation | Prec Factual | Prec Other |
---|---|---|---|---|---|---|---|---|---|---|---|
1.034 | 0.98 | 57 | 0.9960 | 0.3136 | 0.9408 | 0.0 | 0.0 | 0.2961 | 0.8882 | 0.0 | 0.0 |
0.968 | 1.98 | 114 | 0.8945 | 0.3136 | 0.9408 | 0.0 | 0.0 | 0.2961 | 0.8882 | 0.0 | 0.0 |
0.9253 | 2.98 | 171 | 0.7182 | 0.3136 | 0.9408 | 0.0 | 0.0 | 0.2961 | 0.8882 | 0.0 | 0.0 |
0.8215 | 3.98 | 228 | 0.3112 | 0.4795 | 0.9454 | 0.0 | 0.4932 | 0.4351 | 0.9381 | 0.0 | 0.3673 |
0.5073 | 4.98 | 285 | 0.1564 | 0.8272 | 0.9703 | 0.9355 | 0.5758 | 0.8025 | 0.9883 | 0.9667 | 0.4524 |
0.3046 | 5.98 | 342 | 0.1258 | 0.8408 | 0.9751 | 0.9508 | 0.5965 | 0.8323 | 0.9818 | 1.0 | 0.5152 |
0.1971 | 6.98 | 399 | 0.1540 | 0.8458 | 0.9796 | 0.9538 | 0.6038 | 0.8258 | 0.9863 | 0.9394 | 0.5517 |
0.1494 | 7.98 | 456 | 0.1779 | 0.8504 | 0.9737 | 0.9524 | 0.625 | 0.8195 | 0.9907 | 0.9677 | 0.5 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.9.0+cu102
- Datasets 1.9.0
- Tokenizers 0.10.2