File size: 2,380 Bytes
c0c4505
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
---
license: apache-2.0
tags:
- Question Answering
metrics:
- squad
model-index:
- name: question-answering-roberta-base-s
  results: []
---

# Question Answering 
The model is intended to be used for Q&A task, given the question & context, the model would attempt to infer the answer text, answer span & confidence score.<br>
Model is encoder-only (roberta-base) with QuestionAnswering LM Head, fine-tuned on SQUADx dataset with **exact_match:** 86.14 & **f1:** 92.330 performance scores.

[Live Demo: Question Answering Encoders vs Generative](https://huggingface.co/spaces/anshoomehra/question_answering)

Please follow this link for [Generative Question Answering](https://huggingface.co/anshoomehra/question-answering-generative-t5-v1-base-s-q-c/)

Example code:
```
from transformers import pipeline

model_checkpoint = "anshoomehra/question-answering-roberta-base-s"

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 = pipeline("question-answering", model=model_checkpoint)
question_answerer(question=question, context=context)

```

## Training and evaluation data

SQUAD Split

## Training procedure

Preprocessing:
1. SQUAD Data longer chunks were sub-chunked with input context max-length 384 tokens and stride as 128 tokens.
2. Target answers readjusted for sub-chunks, sub-chunks with no-answers or partial answers were set to target answer span as (0,0)

Metrics:
1. Adjusted accordingly to handle sub-chunking.
2. n best = 20
3. skip answers with length zero or higher than max answer length (30)

### Training hyperparameters
Custom Training Loop:
The following hyperparameters were used during training:
- learning_rate: 2e-5
- train_batch_size: 32
- eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2

### Training results

| Epoch | F1       | Exact Match | 
|:-----:|:--------:|:-----------:|
| 1.0   | 91.3085  | 84.5412     |
| 2.0   | 92.3304  | 86.1400     |


### Framework versions

- Transformers 4.23.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.5.2
- Tokenizers 0.13.0