Update README.md
Browse files
README.md
CHANGED
|
@@ -10,7 +10,7 @@ tags:
|
|
| 10 |
|
| 11 |
[BERT-Tiny](https://huggingface.co/google/bert_uncased_L-12_H-128_A-2) created by [Google Research](https://github.com/google-research) and fine-tuned on [SQuAD 2.0](https://rajpurkar.github.io/SQuAD-explorer/) for **Q&A** downstream task.
|
| 12 |
|
| 13 |
-
**Mode size** (after training): **24.
|
| 14 |
|
| 15 |
## Details of BERT-Tiny and its 'family' (from their documentation)
|
| 16 |
|
|
@@ -38,8 +38,8 @@ The script for fine tuning can be found [here](https://github.com/huggingface/tr
|
|
| 38 |
|
| 39 |
| Metric | # Value |
|
| 40 |
| ------ | --------- |
|
| 41 |
-
| **EM** | **
|
| 42 |
-
| **F1** | **
|
| 43 |
|
| 44 |
|
| 45 |
| Model | EM | F1 score | SIZE (MB) |
|
|
@@ -56,8 +56,8 @@ from transformers import pipeline
|
|
| 56 |
|
| 57 |
qa_pipeline = pipeline(
|
| 58 |
"question-answering",
|
| 59 |
-
model="mrm8488/bert-tiny-finetuned-squadv2",
|
| 60 |
-
tokenizer="mrm8488/bert-tiny-finetuned-squadv2"
|
| 61 |
)
|
| 62 |
|
| 63 |
qa_pipeline({
|
|
@@ -65,41 +65,8 @@ qa_pipeline({
|
|
| 65 |
'question': "Who has been working hard for hugginface/transformers lately?"
|
| 66 |
|
| 67 |
})
|
| 68 |
-
|
| 69 |
-
# Output:
|
| 70 |
-
```
|
| 71 |
-
|
| 72 |
-
```json
|
| 73 |
-
{
|
| 74 |
-
"answer": "Manuel Romero",
|
| 75 |
-
"end": 13,
|
| 76 |
-
"score": 0.05684709993458714,
|
| 77 |
-
"start": 0
|
| 78 |
-
}
|
| 79 |
-
```
|
| 80 |
-
|
| 81 |
-
### Yes! That was easy 🎉 Let's try with another example
|
| 82 |
-
|
| 83 |
-
```python
|
| 84 |
-
qa_pipeline({
|
| 85 |
-
'context': "Manuel Romero has been working hardly in the repository hugginface/transformers lately",
|
| 86 |
-
'question': "For which company has worked Manuel Romero?"
|
| 87 |
-
})
|
| 88 |
-
|
| 89 |
-
# Output:
|
| 90 |
```
|
| 91 |
|
| 92 |
-
```json
|
| 93 |
-
{
|
| 94 |
-
"answer": "hugginface/transformers",
|
| 95 |
-
"end": 79,
|
| 96 |
-
"score": 0.11613431826808274,
|
| 97 |
-
"start": 56
|
| 98 |
-
}
|
| 99 |
-
```
|
| 100 |
-
|
| 101 |
-
### It works!! 🎉 🎉 🎉
|
| 102 |
-
|
| 103 |
> Created by [Manuel Romero/@mrm8488](https://twitter.com/mrm8488) | [LinkedIn](https://www.linkedin.com/in/manuel-romero-cs/)
|
| 104 |
|
| 105 |
> Made with <span style="color: #e25555;">♥</span> in Spain
|
|
|
|
| 10 |
|
| 11 |
[BERT-Tiny](https://huggingface.co/google/bert_uncased_L-12_H-128_A-2) created by [Google Research](https://github.com/google-research) and fine-tuned on [SQuAD 2.0](https://rajpurkar.github.io/SQuAD-explorer/) for **Q&A** downstream task.
|
| 12 |
|
| 13 |
+
**Mode size** (after training): **24.33 MB**
|
| 14 |
|
| 15 |
## Details of BERT-Tiny and its 'family' (from their documentation)
|
| 16 |
|
|
|
|
| 38 |
|
| 39 |
| Metric | # Value |
|
| 40 |
| ------ | --------- |
|
| 41 |
+
| **EM** | **57.12** |
|
| 42 |
+
| **F1** | **60.86** |
|
| 43 |
|
| 44 |
|
| 45 |
| Model | EM | F1 score | SIZE (MB) |
|
|
|
|
| 56 |
|
| 57 |
qa_pipeline = pipeline(
|
| 58 |
"question-answering",
|
| 59 |
+
model="mrm8488/bert-tiny-5-finetuned-squadv2",
|
| 60 |
+
tokenizer="mrm8488/bert-tiny-5-finetuned-squadv2"
|
| 61 |
)
|
| 62 |
|
| 63 |
qa_pipeline({
|
|
|
|
| 65 |
'question': "Who has been working hard for hugginface/transformers lately?"
|
| 66 |
|
| 67 |
})
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
```
|
| 69 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 70 |
> Created by [Manuel Romero/@mrm8488](https://twitter.com/mrm8488) | [LinkedIn](https://www.linkedin.com/in/manuel-romero-cs/)
|
| 71 |
|
| 72 |
> Made with <span style="color: #e25555;">♥</span> in Spain
|