|
--- |
|
license: mit |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- precision |
|
- recall |
|
- f1 |
|
- accuracy |
|
model-index: |
|
- name: camembert-base-finetuned-mixte-symbole-dd |
|
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. --> |
|
|
|
# camembert-base-finetuned-mixte-symbole-dd |
|
|
|
This model is a fine-tuned version of [camembert-base](https://huggingface.co/camembert-base) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.3576 |
|
- Precision: 0.8788 |
|
- Recall: 0.8986 |
|
- F1: 0.8886 |
|
- Accuracy: 0.9230 |
|
|
|
## 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: 16 |
|
- eval_batch_size: 16 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 5 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy | |
|
|:-------------:|:-----:|:-----:|:---------------:|:---------:|:------:|:------:|:--------:| |
|
| 0.1977 | 1.0 | 4322 | 0.2256 | 0.8388 | 0.8980 | 0.8674 | 0.9240 | |
|
| 0.1498 | 2.0 | 8644 | 0.2765 | 0.8666 | 0.8916 | 0.8789 | 0.9235 | |
|
| 0.1085 | 3.0 | 12966 | 0.2945 | 0.8666 | 0.9028 | 0.8844 | 0.9211 | |
|
| 0.0775 | 4.0 | 17288 | 0.3181 | 0.8735 | 0.9015 | 0.8873 | 0.9226 | |
|
| 0.0605 | 5.0 | 21610 | 0.3576 | 0.8788 | 0.8986 | 0.8886 | 0.9230 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.21.1 |
|
- Pytorch 1.12.1+cu113 |
|
- Datasets 2.4.0 |
|
- Tokenizers 0.12.1 |
|
|