๐ Electra-base for QA
This is an Electra-base model fine-tuned for extractive question answering on the SQuAD 2.0 dataset.
๐ Quick Start
Prerequisites
- The model is based on the English language and trained on SQuAD 2.0 for extractive question answering.
- The infrastructure used for training is 1x Tesla v100.
Code Example
You can refer to the example in FARM.
โจ Features
- Language Model: electra-base
- Language: English
- Downstream-task: Extractive QA
- Training data: SQuAD 2.0
- Eval data: SQuAD 2.0
- Infrastructure: 1x Tesla v100
๐ฆ Installation
No specific installation steps are provided in the original document.
๐ป Usage Examples
In Transformers
from transformers import AutoModelForQuestionAnswering, AutoTokenizer, pipeline
model_name = "deepset/electra-base-squad2"
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)
model = AutoModelForQuestionAnswering.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)
In FARM
from farm.modeling.adaptive_model import AdaptiveModel
from farm.modeling.tokenization import Tokenizer
from farm.infer import Inferencer
model_name = "deepset/electra-base-squad2"
nlp = Inferencer.load(model_name, task_type="question_answering")
QA_input = [{"questions": ["Why is model conversion important?"],
"text": "The option to convert models between FARM and transformers gives freedom to the user and let people easily switch between frameworks."}]
res = nlp.inference_from_dicts(dicts=QA_input)
model = AdaptiveModel.convert_from_transformers(model_name, device="cpu", task_type="question_answering")
tokenizer = Tokenizer.load(model_name)
In haystack
reader = FARMReader(model_name_or_path="deepset/electra-base-squad2")
reader = TransformersReader(model="deepset/electra-base-squad2",tokenizer="deepset/electra-base-squad2")
๐ Documentation
Hyperparameters
seed=42
batch_size = 32
n_epochs = 5
base_LM_model = "google/electra-base-discriminator"
max_seq_len = 384
learning_rate = 1e-4
lr_schedule = LinearWarmup
warmup_proportion = 0.1
doc_stride=128
max_query_length=64
Performance
Evaluated on the SQuAD 2.0 dev set with the official eval script.
"exact": 77.30144024256717,
"f1": 81.35438272008543,
"total": 11873,
"HasAns_exact": 74.34210526315789,
"HasAns_f1": 82.45961302894314,
"HasAns_total": 5928,
"NoAns_exact": 80.25231286795626,
"NoAns_f1": 80.25231286795626,
"NoAns_total": 5945
๐ง Technical Details
The model is fine-tuned on the SQuAD 2.0 dataset for extractive question answering. The hyperparameters are carefully selected to achieve good performance on the evaluation set.
๐ License
The license for this model is cc-by-4.0.
๐ Authors
- Vaishali Pal
vaishali.pal [at] deepset.ai
- Branden Chan:
branden.chan [at] deepset.ai
- Timo Mรถller:
timo.moeller [at] deepset.ai
- Malte Pietsch:
malte.pietsch [at] deepset.ai
- Tanay Soni:
tanay.soni [at] deepset.ai
Note
Borrowed this model from Haystack model repo for adding tensorflow model.