diversifix
commited on
Commit
·
97f510a
1
Parent(s):
17e7bc5
Update README.md
Browse files
README.md
CHANGED
@@ -10,6 +10,8 @@ _Work in progress._
|
|
10 |
|
11 |
Language model for inclusive language in German, fine-tuned on [mT5](https://arxiv.org/abs/2010.11934).
|
12 |
|
|
|
|
|
13 |
## Tasks
|
14 |
|
15 |
- **DETECT**: Recognizes instances of the generic masculine, and of other exclusive language. To do.
|
@@ -24,9 +26,29 @@ Language model for inclusive language in German, fine-tuned on [mT5](https://arx
|
|
24 |
|
25 |
◀️ `Das wartende Kollegium wunderte sich.`
|
26 |
|
27 |
-
##
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
-
|
|
|
|
|
|
|
30 |
|
31 |
## License
|
32 |
|
|
|
10 |
|
11 |
Language model for inclusive language in German, fine-tuned on [mT5](https://arxiv.org/abs/2010.11934).
|
12 |
|
13 |
+
An experimental model version is released [on Huggingface](https://huggingface.co/diversifix/diversiformer).
|
14 |
+
|
15 |
## Tasks
|
16 |
|
17 |
- **DETECT**: Recognizes instances of the generic masculine, and of other exclusive language. To do.
|
|
|
26 |
|
27 |
◀️ `Das wartende Kollegium wunderte sich.`
|
28 |
|
29 |
+
## Usage
|
30 |
+
|
31 |
+
```python
|
32 |
+
from transformers import T5Tokenizer, TFT5ForConditionalGeneration
|
33 |
+
|
34 |
+
tokenizer = T5Tokenizer.from_pretrained("google/mt5-small")
|
35 |
+
model = TFT5ForConditionalGeneration.from_pretrained("diversifix/diversiformer")
|
36 |
+
|
37 |
+
def generate(prompt, tokenizer, model):
|
38 |
+
tokenized_text = tokenizer.encode(prompt, return_tensors="tf")
|
39 |
+
ids = model.generate(tokenized_text, max_length=500)
|
40 |
+
output = tokenizer.decode(ids[0], skip_special_tokens=True)
|
41 |
+
return output
|
42 |
+
|
43 |
+
prompts = [
|
44 |
+
'Ersetze "Schüler" durch "Schülerin oder Schüler": Die Schüler kamen zu spät.',
|
45 |
+
'Ersetze "Lehrer" durch "Kollegium": Die wartenden Lehrer wunderten sich.',
|
46 |
+
]
|
47 |
|
48 |
+
for prompt in prompts:
|
49 |
+
output = generate(prompt, tokenizer, model)
|
50 |
+
print(f"{prompt}\n{output}\n\n")
|
51 |
+
```
|
52 |
|
53 |
## License
|
54 |
|