# SQuAD Optimization

A fine-tuned version of the MobileBERT model on the SQuADv1 question answering dataset, optimized for QA tasks

This model is a question-answering model based on DeBERTa-base, fine-tuned on the SQuAD1 dataset and further fine-tuned on the adversarial_qa dataset.

This model is a question-answering model fine-tuned on the SQuAD dataset based on bert-base-uncased

This model is a cross-encoder trained on distilroberta-base for determining whether a given passage can answer a specific question, trained on the GLUE QNLI dataset.

MobileBERT is a lightweight version of BERT_LARGE, featuring a bottleneck structure design that balances self-attention mechanisms and feed-forward networks. This model is fine-tuned on the SQuAD1.1 dataset and is suitable for question-answering tasks.

This is a cross-encoder model based on the ELECTRA architecture, specifically designed for natural language inference (NLI) in question-answering tasks, determining whether a given question can be answered by a specific paragraph.

This model is pruned using the nn_pruning library, retaining 32% of the original weights, with a processing speed 2.15 times faster than the original version and an F1 score of 83.22

A two-stage training QA model based on ELECTRA-Large architecture, first trained on synthetic adversarial data and then fine-tuned on SQuAD and AdversarialQA datasets