|
--- |
|
license: cc-by-sa-4.0 |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- accuracy |
|
model-index: |
|
- name: SloBertAA_Top5_WithOOC |
|
results: [] |
|
datasets: |
|
- gregorgabrovsek/RTVCommentsTop5UsersWithOOC |
|
language: |
|
- sl |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# SloBertAA_Top5_WithOOC |
|
|
|
This model is a fine-tuned version of [EMBEDDIA/sloberta](https://huggingface.co/EMBEDDIA/sloberta) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.5616 |
|
- Accuracy: 0.8946 |
|
|
|
## Other related models |
|
|
|
Models fine-tuned on the RTV datasets: |
|
|
|
| Base model | Includes the OOC class? | 5 classes | 10 classes | 20 classes | 50 classes | 100 classes | |
|
| ------------------- | ----------------------- |:---------:|:----------:|:----------:|:----------:|:-----------:| |
|
| SloBERTa | Yes | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top5_WithOOC) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top10_WithOOC) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top20_WithOOC) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top50_WithOOC) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top100_WithOOC) | |
|
| SloBERTa | No | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top5_WithoutOOC) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top10_WithoutOOC) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top20_WithoutOOC) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top50_WithoutOOC) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top100_WithoutOOC) | |
|
| BERT Multilingual | Yes | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top5_WithOOC_MultilingualBertBase) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top10_WithOOC_MultilingualBertBase) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top20_WithOOC_MultilingualBertBase) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top50_WithOOC_MultilingualBertBase) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top100_WithOOC_MultilingualBertBase) | |
|
| BERT Multilingual | No | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top5_WithoutOOC_MultilingualBertBase) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top10_WithoutOOC_MultilingualBertBase) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top20_WithoutOOC_MultilingualBertBase) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top50_WithoutOOC_MultilingualBertBase) | [link](https://huggingface.co/gregorgabrovsek/SloBertAA_Top100_WithoutOOC_MultilingualBertBase) | |
|
|
|
Models fine-tuned on the IMDb datasets: |
|
|
|
| Base model | Includes the OOC class? | 5 classes | 10 classes | 25 classes | 50 classes | 100 classes | |
|
| ------------------- | ----------------------- |:---------:|:----------:|:----------:|:----------:|:-----------:| |
|
| BERT Multilingual | No | [link](https://huggingface.co/gregorgabrovsek/BERT_AA_IMDB_Top5_WithoutOOC_MultilingualBertBase) |[link](https://huggingface.co/gregorgabrovsek/BERT_AA_IMDB_Top10_WithoutOOC_MultilingualBertBase) |[link](https://huggingface.co/gregorgabrovsek/BERT_AA_IMDB_Top25_WithoutOOC_MultilingualBertBase) |[link](https://huggingface.co/gregorgabrovsek/BERT_AA_IMDB_Top50_WithoutOOC_MultilingualBertBase) |[link](https://huggingface.co/gregorgabrovsek/BERT_AA_IMDB_Top100_WithoutOOC_MultilingualBertBase) | |
|
|
|
|
|
## 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: 12 |
|
- eval_batch_size: 12 |
|
- 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 | Accuracy | |
|
|:-------------:|:-----:|:-----:|:---------------:|:--------:| |
|
| 0.4211 | 1.0 | 10508 | 0.3823 | 0.8700 | |
|
| 0.3163 | 2.0 | 21016 | 0.3917 | 0.8772 | |
|
| 0.257 | 3.0 | 31524 | 0.3771 | 0.8925 | |
|
| 0.1874 | 4.0 | 42032 | 0.5059 | 0.8931 | |
|
| 0.129 | 5.0 | 52540 | 0.5616 | 0.8946 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.26.1 |
|
- Pytorch 1.8.0 |
|
- Datasets 2.10.1 |
|
- Tokenizers 0.13.2 |