IProject-10/distilbert-base-uncased-finetuned-squad2

Model description

DistilBERT fine-tuned on SQuAD 2.0 : Encoder-based Transformer Language model. DistilBERT is a compact and efficient version of BERT (Bidirectional Encoder Representations from Transformers).
It employs a distillation process that transfers knowledge from a larger pretrained model (like BERT) to a smaller one.
Suitable for Question-Answering tasks, predicts answer spans within the context provided.

Language model: distilbert-base-uncased
Language: English
Downstream-task: Question-Answering
Training data: Train-set SQuAD 2.0
Evaluation data: Evaluation-set SQuAD 2.0
Hardware Accelerator used: GPU Tesla T4

Intended uses & limitations

For Question-Answering -

!pip install transformers
from transformers import pipeline
model_checkpoint = "IProject-10/distilbert-base-uncased-finetuned-squad2"
question_answerer = pipeline("question-answering", model=model_checkpoint)

context = """
🤗 Transformers is backed by the three most popular deep learning libraries — Jax, PyTorch and TensorFlow — with a seamless integration
between them. It's straightforward to train your models with one before loading them for inference with the other.
"""

question = "Which deep learning libraries back 🤗 Transformers?"
question_answerer(question=question, context=context)

Results

Evaluation on SQuAD 2.0 validation dataset:

 exact: 65.88056935904994,
 f1: 68.9782873196397,
 total': 11873,
 HasAns_exact: 68.15114709851552,
 HasAns_f1: 74.35546648888003,
 HasAns_total: 5928,
 NoAns_exact: 63.61648444070648,
 NoAns_f1: 63.61648444070648,
 NoAns_total: 5945,
 best_exact: 65.88056935904994,
 best_exact_thresh: 0.9993563294410706,
 best_f1: 68.97828731963992,
 best_f1_thresh: 0.9993563294410706,
 total_time_in_seconds: 122.51037029999998,
 samples_per_second: 96.91424465476456,
 latency_in_seconds: 0.01031840059799545

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 3e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
1.1952	1.0	8235	1.2246
0.8749	2.0	16470	1.3015
0.6708	3.0	24705	1.4648

This model is a fine-tuned version of distilbert-base-uncased on the squad_v2 dataset. It achieves the following results on the evaluation set:

• Loss: 1.4648

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.2
• Tokenizers 0.13.3