Instructions to use creat89/NER_FEDA_Latin2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use creat89/NER_FEDA_Latin2 with Transformers:
# Load model directly from transformers import AutoTokenizer, BERT_model_multidata tokenizer = AutoTokenizer.from_pretrained("creat89/NER_FEDA_Latin2") model = BERT_model_multidata.from_pretrained("creat89/NER_FEDA_Latin2", device_map="auto") - Notebooks
- Google Colab
- Kaggle
| seed: 12 | |
| file_extension: txt | |
| experiment_name: 2021latin_da-IOBES-S____ECn_ | |
| tags2use: NER_IOBES | |
| special_labesl: True | |
| force_size: False | |
| train_file: train_* | |
| masking_percentage: 0.0 | |
| seed: 12 | |
| epochs: 20 | |
| early_stop: 2 | |
| separator: | |
| test_file: test | |
| dev_file: dev_* | |
| crf: True | |
| uppercase: True | |
| lr: 2e-05 | |
| epsilon: 1e-08 | |
| bert_model: /home/lcabrera/BERT_Models/LaBSE | |
| mask_entities: False | |
| train_batch_size: 5 | |
| comment_line: -DOCSTART | |
| sequence_size: 128 | |
| multi_gpu: False | |
| fullwords_mask: None | |
| redundant_uppercase: None | |
| bert_hidden_size: 768 | |
| no_dev: False | |
| multidata_model: Daume | |
| training_datasets: ['train_0_wikiann_CsPlSL', 'train_1_final_csplsl', 'train_2_bsnlp17_CsPlSL', 'train_3_ssj', 'train_4_kpwr2', 'train_5_cnec', 'train_6_turku'] | |
| predict_boundaries: False | |
| biaffine: False | |
| uppercase_percentahe: 0.05 | |