|
--- |
|
library_name: transformers |
|
license: mit |
|
base_model: xlm-roberta-large |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- accuracy |
|
model-index: |
|
- name: emotion_model |
|
results: [] |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# emotion_model |
|
|
|
This model is a fine-tuned version of [xlm-roberta-large](https://huggingface.co/xlm-roberta-large) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.1691 |
|
- Macro F1: 0.5721 |
|
- Micro F1: 0.7014 |
|
- Accuracy: 0.8780 |
|
|
|
## 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: 1.5e-05 |
|
- train_batch_size: 64 |
|
- eval_batch_size: 64 |
|
- seed: 42 |
|
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_ratio: 0.1 |
|
- num_epochs: 10 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Macro F1 | Micro F1 | Accuracy | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------:|:--------:|:--------:| |
|
| 0.2428 | 1.0 | 143 | 0.2269 | 0.0016 | 0.0028 | 0.7811 | |
|
| 0.1979 | 2.0 | 286 | 0.1774 | 0.4377 | 0.6399 | 0.8642 | |
|
| 0.1712 | 3.0 | 429 | 0.1669 | 0.4939 | 0.6727 | 0.8729 | |
|
| 0.1571 | 4.0 | 572 | 0.1635 | 0.5474 | 0.6889 | 0.8768 | |
|
| 0.1426 | 5.0 | 715 | 0.1666 | 0.5658 | 0.6881 | 0.8737 | |
|
| 0.1335 | 6.0 | 858 | 0.1665 | 0.5824 | 0.6999 | 0.8750 | |
|
| 0.1236 | 7.0 | 1001 | 0.1682 | 0.5765 | 0.6940 | 0.8735 | |
|
| 0.1152 | 8.0 | 1144 | 0.1697 | 0.5747 | 0.6964 | 0.8752 | |
|
| 0.1104 | 9.0 | 1287 | 0.1732 | 0.5708 | 0.6930 | 0.8732 | |
|
| 0.1069 | 10.0 | 1430 | 0.1742 | 0.5814 | 0.6959 | 0.8738 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.48.2 |
|
- Pytorch 2.3.1.post300 |
|
- Datasets 2.2.1 |
|
- Tokenizers 0.21.0 |
|
|
