---
license: apache-2.0
base_model: google/electra-small-discriminator
tags:
- generated_from_keras_callback
model-index:
- name: nguyennghia0902/electra-small-discriminator_0.0001_16_15e
  results: []
language:
- vi
- en
metrics:
- accuracy
pipeline_tag: question-answering
datasets:
- nguyennghia0902/project02_textming_dataset
---

<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->

# nguyennghia0902/electra-small-discriminator_0.0001_16_15e

This model is a fine-tuned version of [google/electra-small-discriminator](https://huggingface.co/google/electra-small-discriminator) on [Vietnamese dataset](https://www.kaggle.com/datasets/duyminhnguyentran/csc15105).
It achieves the following results on the evaluation set:
- Train Loss: 0.4315
- Train End Logits Accuracy: 0.8714
- Train Start Logits Accuracy: 0.8580
- Validation Loss: 0.1470
- Validation End Logits Accuracy: 0.9577
- Validation Start Logits Accuracy: 0.9542
- Test Matching Accuracy: 0.90209
- Epoch: 15
- Train time: 21920.9752 seconds ~ 6.09 hours

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- Learning rate: 1e-4
- Batch size: 16
- optimizer: {
              'name': 'Adam',
              'learning_rate': {
                                'module': 'keras.optimizers.schedules',
                                'class_name': 'PolynomialDecay',
                                'config': {
                                          'initial_learning_rate': 0.0001,
                                          'decay_steps': 46905,
                                          'end_learning_rate': 0.0,
                                          'power': 1.0, 'cycle': False
                                          }
                                },
              'epsilon': 1e-08
              }
- training_precision: float32

### Training results

| Train Loss | Train End Logits Accuracy | Train Start Logits Accuracy | Validation Loss | Validation End Logits Accuracy | Validation Start Logits Accuracy | Epoch |
|:----------:|:-------------------------:|:---------------------------:|:---------------:|:------------------------------:|:--------------------------------:|:-----:|
| 2.9418     | 0.3441                    | 0.3115                      | 2.1831          | 0.4777                         | 0.4649                           | 0     |
| 2.2767     | 0.4696                    | 0.4357                      | 1.7802          | 0.5643                         | 0.5481                           | 1     |
| 1.9907     | 0.5234                    | 0.4941                      | 1.5055          | 0.6229                         | 0.6068                           | 2     |
| 1.7630     | 0.5690                    | 0.5440                      | 1.2348          | 0.6824                         | 0.6708                           | 3     |
| 1.5637     | 0.6086                    | 0.5842                      | 1.0345          | 0.7291                         | 0.7190                           | 4     |
| 1.3785     | 0.6500                    | 0.6241                      | 0.8309          | 0.7823                         | 0.7724                           | 5     |
| 1.2118     | 0.6880                    | 0.6604                      | 0.6918          | 0.8105                         | 0.8116                           | 6     |
| 1.0610     | 0.7222                    | 0.6963                      | 0.5471          | 0.8490                         | 0.8476                           | 7     |
| 0.9249     | 0.7495                    | 0.7272                      | 0.4426          | 0.8770                         | 0.8763                           | 8     |
| 0.8085     | 0.7777                    | 0.7585                      | 0.3695          | 0.8919                         | 0.8908                           | 9     |
| 0.7062     | 0.8018                    | 0.7843                      | 0.2773          | 0.9194                         | 0.9198                           | 10    |
| 0.6182     | 0.8232                    | 0.8043                      | 0.2323          | 0.9343                         | 0.9302                           | 11    |
| 0.5422     | 0.8414                    | 0.8267                      | 0.1807          | 0.9470                         | 0.9470                           | 12    |
| 0.4797     | 0.8588                    | 0.8443                      | 0.1570          | 0.9530                         | 0.9515                           | 13    |
| 0.4315     | 0.8714                    | 0.8580                      | 0.1470          | 0.9577                         | 0.9542                           | 14    |


### Framework versions

- Transformers 4.39.3
- TensorFlow 2.15.0
- Datasets 2.18.0
- Tokenizers 0.15.2

## How to use?
```python
from transformers import ElectraTokenizerFast, TFElectraForQuestionAnswering

model_hf = "nguyennghia0902/electra-small-discriminator_0.0001_16_15e"
tokenizer = ElectraTokenizerFast.from_pretrained(model_hf)
reload_model = TFElectraForQuestionAnswering.from_pretrained(model_hf)

question = "Ký túc xá Đại học Quốc gia Thành phố Hồ Chí Minh bao gồm có bao nhiêu khu?"
context = "Ký túc xá Đại học Quốc gia Thành phố Hồ Chí Minh (Ký túc xá ĐHQG-TPHCM) là hệ thống ký túc xá xây tại Khu đô thị Đại học Quốc gia Thành phố Hồ Chí Minh (còn gọi với tên phổ biến: Khu đô thị ĐHQG-HCM hay Làng Đại học Thủ Đức). Ký túc xá ĐHQG-TPHCM gồm có 02 khu: A và B. Địa chỉ: Đường Tạ Quang Bửu, Khu phố 6, phường Linh Trung, thành phố Thủ Đức, Thành phố Hồ Chí Minh, điện thoại: 1900 05 55 59 (111). "

inputs = tokenizer(question, context, return_offsets_mapping=True, return_tensors="tf", max_length=512, truncation=True)
offset_mapping = inputs.pop("offset_mapping")
outputs = reload_model(**inputs)
answer_start_index = int(tf.math.argmax(outputs.start_logits, axis=-1)[0])
answer_end_index = int(tf.math.argmax(outputs.end_logits, axis=-1)[0])
start_char = offset_mapping[0][answer_start_index][0]
end_char = offset_mapping[0][answer_end_index][1]
predicted_answer_text = context[start_char:end_char]

print(predicted_answer_text)
```