metadata
license: mit
base_model: xlm-roberta-base
tags:
- generated_from_trainer
metrics:
- accuracy
- f1
model-index:
- name: xlm-roberta-base-IDMGSP-danish
results: []
datasets:
- ernlavr/IDMGSP-danish
language:
- da
xlm-roberta-base-IDMGSP-danish
This model is a fine-tuned version of xlm-roberta-base on the ernlavr/IDMGSP-danish dataset. It achieves the following results on the evaluation set:
- Loss: 1.0276
- Accuracy: {'accuracy': 0.8530452362901187}
- F1: {'f1': 0.8630636995172495}
Intended uses & limitations
Binary classification, label 0 - text is not AI generated; label 1 - text is AI generated
Training and evaluation data
ernlavr/IDMGSP-danish dataset.
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0001
- 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
- lr_scheduler_warmup_steps: 500
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
|---|---|---|---|---|---|
| 0.3631 | 1.0 | 496 | 0.9902 | {'accuracy': 0.5959059893858984} | {'f1': 0.7104310032596887} |
| 0.3208 | 2.0 | 992 | 1.0736 | {'accuracy': 0.7261814505938843} | {'f1': 0.780245411215901} |
| 0.2191 | 3.0 | 1488 | 0.3496 | {'accuracy': 0.8664392216325499} | {'f1': 0.8717077315208156} |
| 0.1548 | 4.0 | 1984 | 0.5604 | {'accuracy': 0.8155168056608542} | {'f1': 0.8378858538751943} |
| 0.1127 | 5.0 | 2480 | 0.4164 | {'accuracy': 0.8641647712913824} | {'f1': 0.871056735036584} |
| 0.1372 | 6.0 | 2976 | 0.5515 | {'accuracy': 0.8822340156684357} | {'f1': 0.8833833833833834} |
| 0.0279 | 7.0 | 3472 | 0.7203 | {'accuracy': 0.8458428102097548} | {'f1': 0.8573766658873042} |
| 0.0456 | 8.0 | 3968 | 0.8584 | {'accuracy': 0.8498862774829417} | {'f1': 0.8604651162790697} |
| 0.0095 | 9.0 | 4464 | 0.9214 | {'accuracy': 0.8512762193580996} | {'f1': 0.861415283174379} |
| 0.0076 | 10.0 | 4960 | 1.0276 | {'accuracy': 0.8530452362901187} | {'f1': 0.8630636995172495} |
Framework versions
- Transformers 4.35.0
- Pytorch 2.0.1
- Datasets 2.14.6
- Tokenizers 0.14.1