albert-large-v2_ner_wikiann
This model is a fine-tuned version of albert-large-v2 on the wikiann dataset. It achieves the following results on the evaluation set:
- Loss: 0.3416
- Precision: 0.8240
- Recall: 0.8375
- F1: 0.8307
- Accuracy: 0.9270
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: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 5
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.3451 | 1.0 | 2500 | 0.3555 | 0.7745 | 0.7850 | 0.7797 | 0.9067 |
| 0.2995 | 2.0 | 5000 | 0.2927 | 0.7932 | 0.8240 | 0.8083 | 0.9205 |
| 0.252 | 3.0 | 7500 | 0.2936 | 0.8094 | 0.8236 | 0.8164 | 0.9239 |
| 0.1676 | 4.0 | 10000 | 0.3302 | 0.8256 | 0.8359 | 0.8307 | 0.9268 |
| 0.1489 | 5.0 | 12500 | 0.3416 | 0.8240 | 0.8375 | 0.8307 | 0.9270 |
Framework versions
- Transformers 4.20.1
- Pytorch 1.11.0
- Datasets 2.1.0
- Tokenizers 0.12.1