Swedish_BERT-based_L1_classifier

General Information:

Issue: This classifier aims to classify the l1 of the witer of a sentence in Swedish. This can be useful for forensic linguistics or specialised language learning.

Classifier: Linear classifier

Embedding: bert-base-multilingual-cased

Links:

Model: https://huggingface.co/uhygutfr/Swedish_BERT-based_L1_classifier (you are here)

Dataset: https://huggingface.co/datasets/uhygutfr/Swedish_sentences_with_l1

Demo: https://huggingface.co/spaces/uhygutfr/Swedish_BERT-based_L1_classifier_demo

File Breakdown:

model.safetensors This is the trained model weights. It contains the BERT parameters and classification head.

config.json The model configuration. It includes BERT architecture, number of labels, label names, label IDs, etc.

tokenizer.json This is the model's tokenizer from strings to BERT tokens.

tokenizer_config.json tokenizer settings

inference_config.json small file containing prediction_threshold, max_length, and task_type