File size: 1,051 Bytes
8d3bbeb
 
d557a0b
8d3bbeb
715094b
d557a0b
e868a93
 
613c95d
f8d9673
d557a0b
 
 
48d7f20
d557a0b
 
41b793e
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. <br />
The Hosted inference API outputs (on the right) are too short for some reason. Trying to fix it. <br />
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: За яку працю треба більше поважати людину - за фізичну чи інтелектуальну?", 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)))
    
```