David
commited on
Commit
•
c9bf790
1
Parent(s):
34323c5
Update README.md
Browse files
README.md
CHANGED
@@ -15,7 +15,7 @@ We release a `small` and `medium` version with the following configuration:
|
|
15 |
| Model | Layers | Embedding/Hidden Size | Params | Vocab Size | Max Sequence Length | Cased |
|
16 |
| --- | --- | --- | --- | --- | --- | --- |
|
17 |
| **SELECTRA small** | **12** | **256** | **22M** | **50k** | **512** | **True** |
|
18 |
-
| SELECTRA medium | 12 | 384 | 41M | 50k | 512 | True |
|
19 |
|
20 |
Selectra small is about 5 times smaller than BETO but achieves comparable results (see Metrics section below).
|
21 |
|
@@ -43,9 +43,11 @@ Estamos desayun ##ando pan rosa con tomate y aceite de
|
|
43 |
"""
|
44 |
```
|
45 |
|
46 |
-
However, you probably want to use this model to fine-tune it on a
|
|
|
47 |
|
48 |
-
-
|
|
|
49 |
|
50 |
## Metrics
|
51 |
|
@@ -59,7 +61,7 @@ We fine-tune our models on 4 different down-stream tasks:
|
|
59 |
For each task, we conduct 5 trials and state the mean and standard deviation of the metrics in the table below.
|
60 |
To compare our results to other Spanish language models, we provide the same metrics taken from [Table 4](https://huggingface.co/bertin-project/bertin-roberta-base-spanish#results) of the Bertin-project model card.
|
61 |
|
62 |
-
| Model | CoNLL2002 - POS (acc) | CoNLL2002 - NER (f1) | PAWS-X (acc) | XNLI (acc) | Params |
|
63 |
| --- | --- | --- | --- | --- | --- |
|
64 |
| SELECTRA small | 0.9653 +- 0.0007 | 0.863 +- 0.004 | 0.896 +- 0.002 | 0.784 +- 0.002 | **22M** |
|
65 |
| SELECTRA medium | 0.9677 +- 0.0004 | 0.870 +- 0.003 | 0.896 +- 0.002 | **0.804 +- 0.002** | 41M |
|
|
|
15 |
| Model | Layers | Embedding/Hidden Size | Params | Vocab Size | Max Sequence Length | Cased |
|
16 |
| --- | --- | --- | --- | --- | --- | --- |
|
17 |
| **SELECTRA small** | **12** | **256** | **22M** | **50k** | **512** | **True** |
|
18 |
+
| [SELECTRA medium](https://huggingface.co/Recognai/selectra_medium) | 12 | 384 | 41M | 50k | 512 | True |
|
19 |
|
20 |
Selectra small is about 5 times smaller than BETO but achieves comparable results (see Metrics section below).
|
21 |
|
|
|
43 |
"""
|
44 |
```
|
45 |
|
46 |
+
However, you probably want to use this model to fine-tune it on a downstream task.
|
47 |
+
We provide models fine-tuned on the [XNLI dataset](https://huggingface.co/datasets/xnli), which can be used together with the zero-shot classification pipeline:
|
48 |
|
49 |
+
- [Zero-shot SELECTRA small](https://huggingface.co/Recognai/zeroshot_selectra_small)
|
50 |
+
- [Zero-shot SELECTRA medium](https://huggingface.co/Recognai/zeroshot_selectra_medium)
|
51 |
|
52 |
## Metrics
|
53 |
|
|
|
61 |
For each task, we conduct 5 trials and state the mean and standard deviation of the metrics in the table below.
|
62 |
To compare our results to other Spanish language models, we provide the same metrics taken from [Table 4](https://huggingface.co/bertin-project/bertin-roberta-base-spanish#results) of the Bertin-project model card.
|
63 |
|
64 |
+
| Model | [CoNLL2002](https://huggingface.co/datasets/conll2002) - POS (acc) | [CoNLL2002](https://huggingface.co/datasets/conll2002) - NER (f1) | [PAWS-X](https://huggingface.co/datasets/paws-x) (acc) | [XNLI](https://huggingface.co/datasets/xnli) (acc) | Params |
|
65 |
| --- | --- | --- | --- | --- | --- |
|
66 |
| SELECTRA small | 0.9653 +- 0.0007 | 0.863 +- 0.004 | 0.896 +- 0.002 | 0.784 +- 0.002 | **22M** |
|
67 |
| SELECTRA medium | 0.9677 +- 0.0004 | 0.870 +- 0.003 | 0.896 +- 0.002 | **0.804 +- 0.002** | 41M |
|