bert-finetuned-chunks-tags

This model is a fine-tuned version of bert-base-cased on the conll2003 dataset.

Model description

This model is used for Chunks token classification and utilizes the CoNLL-2003 dataset. In processing this model, we used dynamic padding

Intended uses & limitations

This model can be used for token classification in the field of Chunk Chunking, also known as shallow parsing, is a token classification task where sequences of tokens are grouped into syntactically correlated units called chunks—such as noun phrases (NP), verb phrases (VP), or prepositional phrases (PP). Each token is assigned a chunk tag (e.g., B-NP, I-NP, O) indicating its position within a phrase. Chunk tags help capture the structure of a sentence beyond individual part-of-speech tags, which is useful in various NLP applications like parsing and information extraction.

Training and evaluation data

conll2003

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 3

Training results

TrainOutput(global_step=5268, training_loss=0.06017827317854085, metrics={'train_runtime': 584.6969, 'train_samples_per_second': 72.042, 'train_steps_per_second': 9.01, 'total_flos': 920888121858078.0, 'train_loss': 0.06017827317854085, 'epoch': 3.0})

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1