Keras
File size: 1,219 Bytes
b5810aa
 
bee8b7b
b5810aa
bee8b7b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
library_name: keras
---
# Surrogate Model (Teeny-Tiny Castle)

This model is part of a tutorial tied to the [Teeny-Tiny Castle](https://github.com/Nkluge-correa/TeenyTinyCastle), an open-source repository containing educational tools for AI Ethics and Safety research. 

## How to Use

```python
from datasets import load_dataset
from huggingface_hub import hf_hub_download
from keras_preprocessing.sequence import pad_sequences

# Download the surrogate model
hf_hub_download(repo_id="AiresPucrs/surrogate-model",
                filename="surrogate_model.h5",
                local_dir="./",
                repo_type="model"
 )

# Download the surrogate tokenizer file
hf_hub_download(repo_id="AiresPucrs/surrogate-model",
                filename="tokenizer_surrogate_model.json",
                local_dir="./",
                repo_type="model"
 )
surrogate_model = tf.keras.models.load_model('./surrogate_model.h5')

with open('./tokenizer_surrogate_model.json') as fp:
 data = json.load(fp)
 tokenizer_surrogate = tf.keras.preprocessing.text.tokenizer_from_json(data)
 word_index_surrogate = tokenizer_surrogate.word_index
 fp.close()
```