Hugging Face's logo

Datasets:
alvin319
/
semantic-memorization-partial-2023-09-03

Modalities:
Tabular
Formats:
parquet
Size:
10M - 100M
Libraries:
Datasets
Dask
License:
Dataset card Data Studio Files Community
alvin319 commited on
Commit
d96d640
·
1 Parent(s): c2b12af

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +180 -0
README.md CHANGED
@@ -1,5 +1,185 @@
1
  ---
2
  license: mit
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
4
 
5
  This dataset is a partial computation of metrics (memorized token frequencies, non-memorized token frequencies, sequence frequencies) needed for [research](https://github.com/EleutherAI/semantic-memorization).
 
1
  ---
2
  license: mit
3
+ configs:
4
+ - config_name: default
5
+ data_files:
6
+ - split: pile_deduped_70m
7
+ path: data/pile_deduped_70m-*
8
+ - split: memories_deduped_70m
9
+ path: data/memories_deduped_70m-*
10
+ - split: pile_deduped_160m
11
+ path: data/pile_deduped_160m-*
12
+ - split: memories_deduped_160m
13
+ path: data/memories_deduped_160m-*
14
+ - split: pile_deduped_410m
15
+ path: data/pile_deduped_410m-*
16
+ - split: memories_deduped_410m
17
+ path: data/memories_deduped_410m-*
18
+ - split: pile_deduped_1b
19
+ path: data/pile_deduped_1b-*
20
+ - split: memories_deduped_1b
21
+ path: data/memories_deduped_1b-*
22
+ - split: pile_deduped_1.4b
23
+ path: data/pile_deduped_1.4b-*
24
+ - split: memories_deduped_1.4b
25
+ path: data/memories_deduped_1.4b-*
26
+ - split: pile_deduped_2.8b
27
+ path: data/pile_deduped_2.8b-*
28
+ - split: memories_deduped_2.8b
29
+ path: data/memories_deduped_2.8b-*
30
+ - split: pile_deduped_6.9b
31
+ path: data/pile_deduped_6.9b-*
32
+ - split: memories_deduped_6.9b
33
+ path: data/memories_deduped_6.9b-*
34
+ - split: pile_deduped_12b
35
+ path: data/pile_deduped_12b-*
36
+ - split: memories_deduped_12b
37
+ path: data/memories_deduped_12b-*
38
+ - split: pile_duped_70m
39
+ path: data/pile_duped_70m-*
40
+ - split: memories_duped_70m
41
+ path: data/memories_duped_70m-*
42
+ - split: pile_duped_160m
43
+ path: data/pile_duped_160m-*
44
+ - split: memories_duped_160m
45
+ path: data/memories_duped_160m-*
46
+ - split: pile_duped_410m
47
+ path: data/pile_duped_410m-*
48
+ - split: memories_duped_410m
49
+ path: data/memories_duped_410m-*
50
+ - split: pile_duped_1b
51
+ path: data/pile_duped_1b-*
52
+ - split: memories_duped_1b
53
+ path: data/memories_duped_1b-*
54
+ - split: pile_duped_1.4b
55
+ path: data/pile_duped_1.4b-*
56
+ - split: memories_duped_1.4b
57
+ path: data/memories_duped_1.4b-*
58
+ - split: pile_duped_2.8b
59
+ path: data/pile_duped_2.8b-*
60
+ - split: memories_duped_2.8b
61
+ path: data/memories_duped_2.8b-*
62
+ - split: pile_duped_6.9b
63
+ path: data/pile_duped_6.9b-*
64
+ - split: memories_duped_6.9b
65
+ path: data/memories_duped_6.9b-*
66
+ - split: pile_duped_12b
67
+ path: data/pile_duped_12b-*
68
+ - split: memories_duped_12b
69
+ path: data/memories_duped_12b-*
70
+ dataset_info:
71
+ features:
72
+ - name: sequence_id
73
+ dtype: int64
74
+ - name: tokens
75
+ sequence: int64
76
+ - name: memorized_frequencies
77
+ sequence: int64
78
+ - name: non_memorized_frequencies
79
+ sequence: int64
80
+ - name: memorization_score
81
+ dtype: float64
82
+ - name: sequence_frequency
83
+ dtype: int64
84
+ splits:
85
+ - name: pile_deduped_70m
86
+ num_bytes: 7860000000
87
+ num_examples: 5000000
88
+ - name: memories_deduped_70m
89
+ num_bytes: 646796256
90
+ num_examples: 411448
91
+ - name: pile_deduped_160m
92
+ num_bytes: 7860000000
93
+ num_examples: 5000000
94
+ - name: memories_deduped_160m
95
+ num_bytes: 913638540
96
+ num_examples: 581195
97
+ - name: pile_deduped_410m
98
+ num_bytes: 7860000000
99
+ num_examples: 5000000
100
+ - name: memories_deduped_410m
101
+ num_bytes: 1274953308
102
+ num_examples: 811039
103
+ - name: pile_deduped_1b
104
+ num_bytes: 7860000000
105
+ num_examples: 5000000
106
+ - name: memories_deduped_1b
107
+ num_bytes: 1623663780
108
+ num_examples: 1032865
109
+ - name: pile_deduped_1.4b
110
+ num_bytes: 7860000000
111
+ num_examples: 5000000
112
+ - name: memories_deduped_1.4b
113
+ num_bytes: 1647608484
114
+ num_examples: 1048097
115
+ - name: pile_deduped_2.8b
116
+ num_bytes: 7860000000
117
+ num_examples: 5000000
118
+ - name: memories_deduped_2.8b
119
+ num_bytes: 2130391692
120
+ num_examples: 1355211
121
+ - name: pile_deduped_6.9b
122
+ num_bytes: 7860000000
123
+ num_examples: 5000000
124
+ - name: memories_deduped_6.9b
125
+ num_bytes: 2641422168
126
+ num_examples: 1680294
127
+ - name: pile_deduped_12b
128
+ num_bytes: 7860000000
129
+ num_examples: 5000000
130
+ - name: memories_deduped_12b
131
+ num_bytes: 2941549980
132
+ num_examples: 1871215
133
+ - name: pile_duped_70m
134
+ num_bytes: 7860000000
135
+ num_examples: 5000000
136
+ - name: memories_duped_70m
137
+ num_bytes: 729334116
138
+ num_examples: 463953
139
+ - name: pile_duped_160m
140
+ num_bytes: 7860000000
141
+ num_examples: 5000000
142
+ - name: memories_duped_160m
143
+ num_bytes: 1084165956
144
+ num_examples: 689673
145
+ - name: pile_duped_410m
146
+ num_bytes: 7860000000
147
+ num_examples: 5000000
148
+ - name: memories_duped_410m
149
+ num_bytes: 1525376052
150
+ num_examples: 970341
151
+ - name: pile_duped_1b
152
+ num_bytes: 7860000000
153
+ num_examples: 5000000
154
+ - name: memories_duped_1b
155
+ num_bytes: 1974653652
156
+ num_examples: 1256141
157
+ - name: pile_duped_1.4b
158
+ num_bytes: 7860000000
159
+ num_examples: 5000000
160
+ - name: memories_duped_1.4b
161
+ num_bytes: 2159490984
162
+ num_examples: 1373722
163
+ - name: pile_duped_2.8b
164
+ num_bytes: 7860000000
165
+ num_examples: 5000000
166
+ - name: memories_duped_2.8b
167
+ num_bytes: 2633221044
168
+ num_examples: 1675077
169
+ - name: pile_duped_6.9b
170
+ num_bytes: 7860000000
171
+ num_examples: 5000000
172
+ - name: memories_duped_6.9b
173
+ num_bytes: 3334163268
174
+ num_examples: 2120969
175
+ - name: pile_duped_12b
176
+ num_bytes: 7860000000
177
+ num_examples: 5000000
178
+ - name: memories_duped_12b
179
+ num_bytes: 3745016472
180
+ num_examples: 2382326
181
+ download_size: 11256676441
182
+ dataset_size: 156765445752
183
  ---
184
 
185
  This dataset is a partial computation of metrics (memorized token frequencies, non-memorized token frequencies, sequence frequencies) needed for [research](https://github.com/EleutherAI/semantic-memorization).