Simon Strandgaard commited on
Commit
b5f2434
1 Parent(s): 6fee525

Here is my project.

Browse files
Files changed (4) hide show
  1. README.md +59 -3
  2. generate_dataset.py +104 -0
  3. random_data.py +11 -0
  4. test_random_data.py +21 -0
README.md CHANGED
@@ -1,3 +1,59 @@
1
- ---
2
- license: mit
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ task_categories:
4
+ - translation
5
+ language:
6
+ - en
7
+ tags:
8
+ - code
9
+ pretty_name: Base64 encode version1
10
+ size_categories:
11
+ - 10K<n<100K
12
+ configs:
13
+ - config_name: default
14
+ data_files:
15
+ - split: train
16
+ path: data/data.jsonl
17
+ ---
18
+
19
+ # Dataset: Base64 encode version1
20
+
21
+ This dataset is for improving base64 encoding capabilities.
22
+
23
+ `GPT 4o` is great at base64 encoding.
24
+
25
+ ```
26
+ user:
27
+ convert this hex data to base64:
28
+ 880567a1
29
+
30
+ assistant:
31
+ The base64 encoding of the hex data `880567a1` is `iAVnoQ==`.
32
+
33
+ user:
34
+ convert this json data representing a byte sequence to base64:
35
+ [30,41,183]
36
+
37
+ assistant:
38
+ The base64 encoding of the JSON data `[30,41,183]` is `Him3`.
39
+ ```
40
+
41
+ However `llama3` is terrible at base64 encoding.
42
+
43
+ Short examples of what `data.jsonl` looks like:
44
+
45
+ ```text
46
+ {"instruction": "Encode hex to Base64", "input": "ecfc2db9ba6049165b", "output": "7PwtubpgSRZb"}
47
+ {"instruction": "change HEX to base64", "input": "60926e782008", "output": "YJJueCAI"}
48
+ {"instruction": "Json to base64", "input": "[77,62,160,64,248,233,105,133,5,248,89,239]", "output": "TT6gQPjpaYUF+Fnv"}
49
+ {"instruction": "Change Json to BASE64", "input": "[10,59,42,251,112,1]", "output": "Cjsq+3AB"}
50
+ {"instruction": "Convert JSON to Base64", "input": "[236,201,129,100,238]", "output": "7MmBZO4="}
51
+ ```
52
+
53
+ # Generate dataset
54
+
55
+ ```
56
+ PROMPT> python generate_dataset.py
57
+ ```
58
+
59
+ This creates the `data.jsonl` file.
generate_dataset.py ADDED
@@ -0,0 +1,104 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from random_data import generate_random_byte_array
2
+ import json
3
+ import os
4
+ import random
5
+ import base64
6
+
7
+ def base64_encode_byte_array(byte_array):
8
+ return base64.b64encode(byte_array).decode('utf-8')
9
+
10
+ def generate_dataset_item(seed):
11
+ length = random.Random(seed + 1000).randint(0, 127)
12
+ byte_array = generate_random_byte_array(length=length, seed=seed + 1001)
13
+
14
+ input_formats = [
15
+ 'hex',
16
+ 'json'
17
+ ]
18
+ input_format = random.Random(seed + 1002).choice(input_formats)
19
+
20
+ names_hex = [
21
+ 'Hexadecimal',
22
+ 'hexadecimal',
23
+ 'hex',
24
+ 'Hex',
25
+ 'HEX',
26
+ ]
27
+ names_json = [
28
+ 'Json',
29
+ 'json',
30
+ 'JSON',
31
+ ]
32
+
33
+ name_input = None
34
+ if input_format == 'hex':
35
+ name_input = random.Random(seed + 1003).choice(names_hex)
36
+ else:
37
+ if input_format == 'json':
38
+ name_input = random.Random(seed + 1004).choice(names_json)
39
+
40
+ name_outputs = [
41
+ 'base64',
42
+ 'Base64',
43
+ 'BASE64',
44
+ ]
45
+ name_output = random.Random(seed + 1005).choice(name_outputs)
46
+
47
+ instructions = [
48
+ f'Encode {name_input} to {name_output}',
49
+ f'encode {name_input} to {name_output}',
50
+ f'convert {name_input} to {name_output}',
51
+ f'Convert {name_input} to {name_output}',
52
+ f'Transform {name_input} to {name_output}',
53
+ f'transform {name_input} to {name_output}',
54
+ f'Change {name_input} to {name_output}',
55
+ f'change {name_input} to {name_output}',
56
+ f'{name_input} to {name_output}',
57
+ f'{name_output} from {name_input}',
58
+ ]
59
+
60
+ instruction = random.Random(seed + 1006).choice(instructions)
61
+
62
+ output = base64_encode_byte_array(byte_array)
63
+
64
+ input = None
65
+ if input_format == 'hex':
66
+ input = byte_array.hex()
67
+ else:
68
+ if input_format == 'json':
69
+ input = json.dumps(list(byte_array), separators=(',', ':'))
70
+
71
+ dict = {
72
+ 'instruction': instruction,
73
+ 'input': input,
74
+ 'output': output
75
+ }
76
+ return dict
77
+
78
+ def generate_dataset(max_num_samples=1000, max_byte_size=1024*1024, seed_start=500000000):
79
+ dataset = []
80
+ dataset_byte_size = 0
81
+ for i in range(max_num_samples):
82
+ item = generate_dataset_item(seed_start + i)
83
+ bytes = len(json.dumps(item))
84
+ if dataset_byte_size + bytes > max_byte_size:
85
+ break
86
+ dataset_byte_size += bytes
87
+ dataset.append(item)
88
+ return dataset
89
+
90
+ dataset = generate_dataset(
91
+ max_num_samples=50000,
92
+ max_byte_size=1024*1024*20,
93
+ )
94
+
95
+ # Save dataset to file
96
+ filename = 'data.jsonl'
97
+ with open(filename, 'w') as f:
98
+ for item in dataset:
99
+ f.write(json.dumps(item) + '\n')
100
+
101
+ # Summary
102
+ file_size = os.path.getsize(filename)
103
+ print(f"Generated {len(dataset)} samples, saved to {filename}, file size: {file_size} bytes.")
104
+
random_data.py ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import random
2
+
3
+ def generate_random_byte_array(length, seed):
4
+ random.seed(seed)
5
+ return bytearray(random.randint(0, 255) for _ in range(length))
6
+
7
+ if __name__ == "__main__":
8
+ length = 10
9
+ seed = 42
10
+ byte_array = generate_random_byte_array(length, seed)
11
+ print(byte_array)
test_random_data.py ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import unittest
2
+ from random_data import generate_random_byte_array
3
+
4
+ class TestRandomData(unittest.TestCase):
5
+ def test_generate_random_byte_array_empty(self):
6
+ actual = generate_random_byte_array(length=0, seed=0)
7
+ expected = bytearray()
8
+ self.assertEqual(actual, expected)
9
+
10
+ def test_generate_random_byte_array_length1(self):
11
+ actual = generate_random_byte_array(length=1, seed=0)
12
+ expected = bytearray([0xc5])
13
+ self.assertEqual(actual, expected)
14
+
15
+ def test_generate_random_byte_array_length2(self):
16
+ actual = generate_random_byte_array(length=2, seed=0)
17
+ expected = bytearray([0xc5, 0xd7])
18
+ self.assertEqual(actual, expected)
19
+
20
+ if __name__ == '__main__':
21
+ unittest.main()