File size: 941 Bytes
8d3bbeb
 
d557a0b
8d3bbeb
715094b
d557a0b
 
5e15d8e
613c95d
f8d9673
d557a0b
 
 
48d7f20
d557a0b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: afl-3.0
language: uk
---
## GPT2 trained to generate ЗНО (Ukrainian exam SAT type of thing) essays

Generated texts are not very cohesive yet but I'm working on it. 
The Hosted inference API outputs (on the right) are too short for some reason. Trying to fix it. 
Use the code from the example below. The model takes "ZNOTITLE: your essay title" inputs. 


### Example of usage:
```python
from transformers import AlbertTokenizer, GPT2LMHeadModel
tokenizer = AlbertTokenizer.from_pretrained("kyryl0s/gpt2-uk-zno-edition")
model = GPT2LMHeadModel.from_pretrained("kyryl0s/gpt2-uk-zno-edition")
input_ids = tokenizer.encode("ZNOTITLE: Чи можна трахатися?", add_special_tokens=False, return_tensors='pt')

outputs = model.generate(
    input_ids,
    do_sample=True,
    num_return_sequences=1,
    max_length=250
)

for i, out in enumerate(outputs):
    print("{}: {}".format(i, tokenizer.decode(out)))
    
```