Frizio/minilm-uncased-squad2-covidqa

minilm-uncased-squad2 for QA on COVID-19

Overview

Language model: deepset/minilm-uncased-squad2
Language: English
Downstream-task: Extractive QA
Training data: SQuAD-style COV-19 QA
Infrastructure: A4000

Initially fine-tuned for https://github.com/CDCapobianco/COVID-Question-Answering-REST-API

Hyperparameters

batch_size = 24
n_epochs = 3
base_LM_model = "deepset/minilm-uncased-squad2"
max_seq_len = 384
learning_rate = 3e-5
lr_schedule = LinearWarmup
warmup_proportion = 0.1
doc_stride = 128
dev_split = 0
x_val_splits = 5
no_ans_boost = -100

---

license: cc-by-4.0

Performance

Single EM-Scores: [0.7441, 0.7938, 0.6666, 0.6576, 0.6445]
Single F1-Scores: [0.8261, 0.8748, 0.8188, 0.7633, 0.7935]
XVAL EM: 0.7013 XVAL f1: 0.8153

Usage

In Haystack

For doing QA at scale (i.e. many docs instead of single paragraph), you can load the model also in haystack:

reader = FARMReader(model_name_or_path="Frizio/minilm-uncased-squad2-covidqa")

In Transformers

from transformers import AutoModelForQuestionAnswering, AutoTokenizer, pipeline


model_name = "Frizio/minilm-uncased-squad2-covidqa"

# a) Get predictions
nlp = pipeline('question-answering', model=model_name, tokenizer=model_name)
QA_input = {
    'question': 'Why is model conversion important?',
    'context': 'The option to convert models between FARM and transformers gives freedom to the user and let people easily switch between frameworks.'
}
res = nlp(QA_input)

# b) Load model & tokenizer
model = AutoModelForQuestionAnswering.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)