pos-polish-gpt2-large
This model was trained from polish-gpt2-large on clarin-pl/nkjp-pos dataset. It achieves the following results on the evaluation set:
- Loss: 0.2290
- Precision: 0.8910
- Recall: 0.9328
- F1: 0.9114
- Accuracy: 0.9450
Model description
Trained from polish-gpt2-large
Intended uses & limitations
Part-of-speech tagging for Polish language. Tags description at the bottom of http://nkjp.pl/poliqarp/help/plse2.html
Training and evaluation data
Dataset: clarin-pl/nkjp-pos
Datacollator:
from transformers import DataCollatorForTokenClassification
data_collator = DataCollatorForTokenClassification(tokenizer=tokenizer)
Training procedure
GPU: RTX 3090
Training time: 01:15:31
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.0 | 0 | 3.8487 | 3.8487 | 3.8487 | 3.8487 | 3.8487 | |
| 0.1952 | 1.0 | 2444 | 0.1942 | 0.8865 | 0.9304 | 0.9079 | 0.9426 |
| 0.1287 | 2.0 | 4889 | 0.1984 | 0.8903 | 0.9322 | 0.9108 | 0.9449 |
| 0.0832 | 3.0 | 7332 | 0.2290 | 0.8910 | 0.9328 | 0.9114 | 0.9450 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0