Using the ClimateBERT-f model as starting point,the TCFD-BERT language model is additionally pre-trained to include precise paragraphs related to climate change.
<!-- 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. -->
TCFD-BERT
It achieves the following results on the evaluation set:
- Loss: 1.1325
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- distributed_type: tpu
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.865 | 0.37 | 500 | 1.4460 |
| 1.6601 | 0.73 | 1000 | 1.3491 |
| 1.593 | 1.1 | 1500 | 1.3190 |
| 1.5336 | 1.46 | 2000 | 1.2801 |
| 1.5081 | 1.83 | 2500 | 1.2446 |
| 1.4547 | 2.19 | 3000 | 1.2281 |
| 1.4358 | 2.56 | 3500 | 1.2065 |
| 1.4121 | 2.92 | 4000 | 1.1874 |
| 1.396 | 3.29 | 4500 | 1.1817 |
| 1.383 | 3.65 | 5000 | 1.1747 |
| 1.3662 | 4.02 | 5500 | 1.1717 |
| 1.3545 | 4.38 | 6000 | 1.1567 |
| 1.3441 | 4.75 | 6500 | 1.1325 |
Framework versions
- Transformers 4.18.0
- Pytorch 1.9.0+cu102
- Datasets 2.1.0
- Tokenizers 0.12.1
