File size: 1,719 Bytes
4c883e0
 
 
 
 
 
 
6adc97a
 
 
 
ee0a5da
6adc97a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
76dd1d3
6adc97a
76dd1d3
6adc97a
76dd1d3
6adc97a
 
 
 
 
 
 
 
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
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
---
datasets:
- oscar
language:
- da
widget:
  - text: Der var engang
---

# What is this?

A GPT-2 model (medium version, ~354.8 M parameters) for Danish text generation. The model was not pre-trained from scratch but adapted from the English version using [CLP-Transfer](https://arxiv.org/abs/2301.09626).

# How to use

Test the model using the pipeline from the [🤗 Transformers](https://github.com/huggingface/transformers) library:

```python
from transformers import pipeline

generator = pipeline("text-generation", model = "KennethTM/gpt2-medium-danish")
text = generator("Manden arbejdede som")

print(text[0]["generated_text"])
```

Or load it using the Auto* classes:

```python
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("KennethTM/gpt2-medium-danish")
model = AutoModelForCausalLM.from_pretrained("KennethTM/gpt2-medium-danish")
```

# Model training

The training data are the Danish part of the [oscar dataset](https://huggingface.co/datasets/oscar) ('unshuffled_deduplicated_da') and a context length of 1024 tokens.

The model weights are initialized from the English [GPT-2 medium model](https://huggingface.co/gpt2-medium) ('source model') with new word token embeddings created from the Danish [GPT-2 small model](https://huggingface.co/KennethTM/gpt2-small-danish) ('helper model') using the [CLP-Transfer method](https://github.com/malteos/clp-transfer).

The model is trained using ~1.000.000 samples.

For reference, the model achieves a perplexity of 24.7 on 5.000 random validation samples.

The model is trained on an 8 GB GPU.

# Notes

This is a pre-trained model, for optimal performance it should be finetuned for new tasks.