Migrate model card from transformers-repo
Browse filesRead announcement at https://discuss.huggingface.co/t/announcement-all-model-cards-will-be-migrated-to-hf-co-model-repos/2755
Original file history: https://github.com/huggingface/transformers/commits/master/model_cards/mrm8488/spanbert-large-finetuned-squadv1/README.md
README.md
ADDED
@@ -0,0 +1,80 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language: en
|
3 |
+
thumbnail:
|
4 |
+
---
|
5 |
+
|
6 |
+
# SpanBERT large fine-tuned on SQuAD v1
|
7 |
+
|
8 |
+
[SpanBERT](https://github.com/facebookresearch/SpanBERT) created by [Facebook Research](https://github.com/facebookresearch) and fine-tuned on [SQuAD 1.1](https://rajpurkar.github.io/SQuAD-explorer/explore/1.1/dev/) for **Q&A** downstream task ([by them](https://github.com/facebookresearch/SpanBERT#finetuned-models-squad-1120-relation-extraction-coreference-resolution)).
|
9 |
+
|
10 |
+
## Details of SpanBERT
|
11 |
+
|
12 |
+
[SpanBERT: Improving Pre-training by Representing and Predicting Spans](https://arxiv.org/abs/1907.10529)
|
13 |
+
|
14 |
+
## Details of the downstream task (Q&A) - Dataset π π§ β
|
15 |
+
|
16 |
+
[SQuAD1.1](https://rajpurkar.github.io/SQuAD-explorer/)
|
17 |
+
|
18 |
+
## Model fine-tuning ποΈβ
|
19 |
+
|
20 |
+
You can get the fine-tuning script [here](https://github.com/facebookresearch/SpanBERT)
|
21 |
+
|
22 |
+
```bash
|
23 |
+
python code/run_squad.py \
|
24 |
+
--do_train \
|
25 |
+
--do_eval \
|
26 |
+
--model spanbert-large-cased \
|
27 |
+
--train_file train-v1.1.json \
|
28 |
+
--dev_file dev-v1.1.json \
|
29 |
+
--train_batch_size 32 \
|
30 |
+
--eval_batch_size 32 \
|
31 |
+
--learning_rate 2e-5 \
|
32 |
+
--num_train_epochs 4 \
|
33 |
+
--max_seq_length 512 \
|
34 |
+
--doc_stride 128 \
|
35 |
+
--eval_metric f1 \
|
36 |
+
--output_dir squad_output \
|
37 |
+
--fp16
|
38 |
+
```
|
39 |
+
|
40 |
+
## Results Comparison π
|
41 |
+
|
42 |
+
| | SQuAD 1.1 | SQuAD 2.0 | Coref | TACRED |
|
43 |
+
| ---------------------- | ------------- | --------- | ------- | ------ |
|
44 |
+
| | F1 | F1 | avg. F1 | F1 |
|
45 |
+
| BERT (base) | 88.5* | 76.5* | 73.1 | 67.7 |
|
46 |
+
| SpanBERT (base) | [92.4*](https://huggingface.co/mrm8488/spanbert-base-finetuned-squadv1) | [83.6*](https://huggingface.co/mrm8488/spanbert-base-finetuned-squadv2) | 77.4 | [68.2](https://huggingface.co/mrm8488/spanbert-base-finetuned-tacred) |
|
47 |
+
| BERT (large) | 91.3 | 83.3 | 77.1 | 66.4 |
|
48 |
+
| SpanBERT (large) | **94.6** (this) | [88.7](https://huggingface.co/mrm8488/spanbert-large-finetuned-squadv2) | 79.6 | [70.8](https://huggingface.co/mrm8488/spanbert-large-finetuned-tacred) |
|
49 |
+
|
50 |
+
|
51 |
+
Note: The numbers marked as * are evaluated on the development sets because those models were not submitted to the official SQuAD leaderboard. All the other numbers are test numbers.
|
52 |
+
|
53 |
+
## Model in action
|
54 |
+
|
55 |
+
Fast usage with **pipelines**:
|
56 |
+
|
57 |
+
```python
|
58 |
+
from transformers import pipeline
|
59 |
+
|
60 |
+
qa_pipeline = pipeline(
|
61 |
+
"question-answering",
|
62 |
+
model="mrm8488/spanbert-large-finetuned-squadv1",
|
63 |
+
tokenizer="SpanBERT/spanbert-large-cased"
|
64 |
+
)
|
65 |
+
|
66 |
+
qa_pipeline({
|
67 |
+
'context': "Manuel Romero has been working very hard in the repository hugginface/transformers lately",
|
68 |
+
'question': "How has been working Manuel Romero lately?"
|
69 |
+
|
70 |
+
})
|
71 |
+
|
72 |
+
# Output: {'answer': 'very hard in the repository hugginface/transformers',
|
73 |
+
'end': 82,
|
74 |
+
'score': 0.327230326857725,
|
75 |
+
'start': 31}
|
76 |
+
```
|
77 |
+
|
78 |
+
> Created by [Manuel Romero/@mrm8488](https://twitter.com/mrm8488)
|
79 |
+
|
80 |
+
> Made with <span style="color: #e25555;">♥</span> in Spain
|