File size: 3,034 Bytes
7706069
 
98e3346
 
 
 
 
7706069
98e3346
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
datasets:
- bookcorpus
- wikipedia
language:
- en
---

# BERT L8-H256 (uncased)

Mini BERT models from https://arxiv.org/abs/1908.08962 that the HF team didn't convert. The original [conversion script](https://github.com/huggingface/transformers/blob/main/src/transformers/models/bert/convert_bert_original_tf_checkpoint_to_pytorch.py) is used.

See the original Google repo: [google-research/bert](https://github.com/google-research/bert)

Note: it's not clear if these checkpoints have undergone knowledge distillation.

## Model variants

|   |H=128|H=256|H=512|H=768|
|---|:---:|:---:|:---:|:---:|
| **L=2**  |[2/128 (BERT-Tiny)][2_128]|[2/256][2_256]|[2/512][2_512]|[2/768][2_768]|
| **L=4**  |[4/128][4_128]|[4/256 (BERT-Mini)][4_256]|[4/512 (BERT-Small)][4_512]|[4/768][4_768]|
| **L=6**  |[6/128][6_128]|[6/256][6_256]|[6/512][6_512]|[6/768][6_768]|
| **L=8**  |[8/128][8_128]|[**8/256**][8_256]|[8/512 (BERT-Medium)][8_512]|[8/768][8_768]|
| **L=10** |[10/128][10_128]|[10/256][10_256]|[10/512][10_512]|[10/768][10_768]|
| **L=12** |[12/128][12_128]|[12/256][12_256]|[12/512][12_512]|[12/768 (BERT-Base, original)][12_768]|

[2_128]: https://huggingface.co/gaunernst/bert-tiny-uncased
[2_256]: https://huggingface.co/gaunernst/bert-L2-H256-uncased
[2_512]: https://huggingface.co/gaunernst/bert-L2-H512-uncased
[2_768]: https://huggingface.co/gaunernst/bert-L2-H768-uncased
[4_128]: https://huggingface.co/gaunernst/bert-L4-H128-uncased
[4_256]: https://huggingface.co/gaunernst/bert-mini-uncased
[4_512]: https://huggingface.co/gaunernst/bert-small-uncased
[4_768]: https://huggingface.co/gaunernst/bert-L4-H768-uncased
[6_128]: https://huggingface.co/gaunernst/bert-L6-H128-uncased
[6_256]: https://huggingface.co/gaunernst/bert-L6-H256-uncased
[6_512]: https://huggingface.co/gaunernst/bert-L6-H512-uncased
[6_768]: https://huggingface.co/gaunernst/bert-L6-H768-uncased
[8_128]: https://huggingface.co/gaunernst/bert-L8-H128-uncased
[8_256]: https://huggingface.co/gaunernst/bert-L8-H256-uncased
[8_512]: https://huggingface.co/gaunernst/bert-medium-uncased
[8_768]: https://huggingface.co/gaunernst/bert-L8-H768-uncased
[10_128]: https://huggingface.co/gaunernst/bert-L10-H128-uncased
[10_256]: https://huggingface.co/gaunernst/bert-L10-H256-uncased
[10_512]: https://huggingface.co/gaunernst/bert-L10-H512-uncased
[10_768]: https://huggingface.co/gaunernst/bert-L10-H768-uncased
[12_128]: https://huggingface.co/gaunernst/bert-L12-H128-uncased
[12_256]: https://huggingface.co/gaunernst/bert-L12-H256-uncased
[12_512]: https://huggingface.co/gaunernst/bert-L12-H512-uncased
[12_768]: https://huggingface.co/bert-base-uncased

## Usage

See other BERT model cards e.g. https://huggingface.co/bert-base-uncased

## Citation

```bibtex
@article{turc2019,
  title={Well-Read Students Learn Better: On the Importance of Pre-training Compact Models},
  author={Turc, Iulia and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina},
  journal={arXiv preprint arXiv:1908.08962v2 },
  year={2019}
}
```