---
language: 
- tr
thumbnail: 
tags:
- gpt2
- turkish

license: Apache 2.0
datasets:
- wikipedia-turkish
metrics:
- perplexity
- accuracy

widget:
- text: "Bu yazıyı bir bilgisayar yazdı. Yazarken"
  context: ""
- text: "İnternete kolay erişim sayesinde dünya daha da küçüldü. Bunun sonucunda"
  context: ""
  
---

# MyModel

## Model description

This is a GPT2-Small English based model finetuned and additionaly trainied with Wikipedia Articles in Turkish as of 28-10-2020

Work has been done on Pierre Guillou tutorial as on this page.
(https://github.com/piegu/fastai-projects/blob/master/finetuning-English-GPT2-any-language-Portuguese-HuggingFace-fastaiv2.ipynb) 

Code is converted to work with Fastai 2.X .

Using Google Colab for training. 

Additional tutorial and source will be in https://github.com/gorkemgoknar in later stage.

Current accuracy 28.9 %  , Perplexity : 86.71

Models are available:

* [gpt2-small-tuned-tr] (https://huggingface.co/gorkemgoknar/gpt2-small-turkish)

## Intended uses & limitations

#### How to use

#### Install

```python
from transformers import AutoTokenizer, AutoModelWithLMHead
import torch

tokenizer = AutoTokenizer.from_pretrained("gorkemgoknar/gpt2-small-turkish")
model = AutoModelWithLMHead.from_pretrained("gorkemgoknar/gpt2-small-turkish")

# Get sequence length max of 1024
tokenizer.model_max_length=1024 

model.eval()  # disable dropout (or leave in train mode to finetune)

```

#### Generate 1 word
```python
# input sequence
text = "Bu yazıyı bilgisayar yazdı."
inputs = tokenizer(text, return_tensors="pt")

# model output
outputs = model(**inputs, labels=inputs["input_ids"])
loss, logits = outputs[:2]
predicted_index = torch.argmax(logits[0, -1, :]).item()
predicted_text = tokenizer.decode([predicted_index])

# results
print('input text:', text)
print('predicted text:', predicted_text)

# input text: Quem era Jim Henson? Jim Henson era um
# predicted text:  homem

```

#### Generate Full Sequence
```python
# input sequence
text = "Bu yazıyı bilgisayar yazdı."
inputs = tokenizer(text, return_tensors="pt")

# model output using Top-k sampling text generation method
sample_outputs = model.generate(inputs.input_ids,
                                pad_token_id=50256,
                                do_sample=True, 
                                max_length=50, # put the token number you want
                                top_k=40,
                                num_return_sequences=1)

# generated sequence
for i, sample_output in enumerate(sample_outputs):
    print(">> Generated text {}\n\n{}".format(i+1, tokenizer.decode(sample_output.tolist())))

# >> Generated text
# Quem era Jim Henson? Jim Henson era um executivo de televisão e diretor de um grande estúdio de cinema mudo chamado Selig,
# depois que o diretor de cinema mudo Georges Seuray dirigiu vários filmes para a Columbia e o estúdio.    

```

#### Limitations and bias

The training data used for this model come from Turkish Wikipedia. We know it contains a lot of unfiltered content from the internet, which is far from neutral. 


## Training data

Wikipedia Turkish article dump as of 28-10-2020

## Training procedure


## Eval results

epoch	train_loss	valid_loss	accuracy	perplexity	time
0	6.922922	6.653488	0.148002	775.484253	2:26:41    
1	4.799396	4.633522	0.277028	102.875755	3:03:38    
2	4.610025	4.462641	0.289884	86.716248	2:34:50      



```