File size: 834 Bytes
3abeb35 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 |
**[`microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext`](https://huggingface.co/microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext)** fine-tuned on **[`SQuAD V2`](https://rajpurkar.github.io/SQuAD-explorer/)** using **[`run_qa.py`](https://github.com/huggingface/transformers/blob/master/examples/pytorch/question-answering/run_qa.py)**
Tunning script:
```bash
BASE_MODEL=microsoft/BiomedNLP-PubMedBERT-base-uncased-abstract-fulltext
OUTPUT_DIR=~/Documents/projects/tunned_models/ms_pubmed_bert_squadv2/
python run_qa.py \
--model_name_or_path $BASE_MODEL\
--dataset_name squad_v2 \
--do_train \
--do_eval \
--version_2_with_negative \
--per_device_train_batch_size 12 \
--learning_rate 3e-5 \
--num_train_epochs 2 \
--max_seq_length 384 \
--doc_stride 128 \
--output_dir $OUTPUT_DIR
``` |