much1na commited on
Commit
ced418a
·
verified ·
1 Parent(s): e5ca0e9

Add dataset card

Browse files
Files changed (1) hide show
  1. README.md +169 -0
README.md ADDED
@@ -0,0 +1,169 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ language: en
4
+ configs:
5
+ - config_name: aerospace
6
+ data_files: aerospace/*.parquet
7
+ - config_name: agronomy
8
+ data_files: agronomy/*.parquet
9
+ - config_name: artistic
10
+ data_files: artistic/*.parquet
11
+ - config_name: astronomy
12
+ data_files: astronomy/*.parquet
13
+ - config_name: atmospheric_science
14
+ data_files: atmospheric_science/*.parquet
15
+ - config_name: automotive
16
+ data_files: automotive/*.parquet
17
+ - config_name: beauty
18
+ data_files: beauty/*.parquet
19
+ - config_name: biology
20
+ data_files: biology/*.parquet
21
+ - config_name: celebrity
22
+ data_files: celebrity/*.parquet
23
+ - config_name: chemistry
24
+ data_files: chemistry/*.parquet
25
+ - config_name: christianity
26
+ data_files: christianity/*.parquet
27
+ - config_name: civil_engineering
28
+ data_files:
29
+ - civil_engineering/*.parquet
30
+ - hydraulic_engineering/*.parquet
31
+ - config_name: communication_engineering
32
+ data_files: communication_engineering/*.parquet
33
+ - config_name: computer_science_and_technology
34
+ data_files: computer_science_and_technology/*.parquet
35
+ - config_name: design
36
+ data_files: design/*.parquet
37
+ - config_name: drama_and_film
38
+ data_files:
39
+ - drama_and_film/*.parquet
40
+ - movie/*.parquet
41
+ - config_name: economics
42
+ data_files: economics/*.parquet
43
+ - config_name: electronic_science
44
+ data_files: electronic_science/*.parquet
45
+ - config_name: entertainment
46
+ data_files: entertainment/*.parquet
47
+ - config_name: environmental_science
48
+ data_files: environmental_science/*.parquet
49
+ - config_name: fashion
50
+ data_files: fashion/*.parquet
51
+ - config_name: finance
52
+ data_files: finance/*.parquet
53
+ - config_name: food
54
+ data_files: food/*.parquet
55
+ - config_name: gamble
56
+ data_files: gamble/*.parquet
57
+ - config_name: game
58
+ data_files: game/*.parquet
59
+ - config_name: geography
60
+ data_files: geography/*.parquet
61
+ - config_name: health
62
+ data_files: health/*.parquet
63
+ - config_name: history
64
+ data_files: history/*.parquet
65
+ - config_name: hobby
66
+ data_files: hobby/*.parquet
67
+ - config_name: instrument_science
68
+ data_files: instrument_science/*.parquet
69
+ - config_name: journalism_and_media_communication
70
+ data_files: journalism_and_media_communication/*.parquet
71
+ - config_name: landscape_architecture
72
+ data_files: landscape_architecture/*.parquet
73
+ - config_name: law
74
+ data_files: law/*.parquet
75
+ - config_name: library
76
+ data_files: library/*.parquet
77
+ - config_name: literature
78
+ data_files: literature/*.parquet
79
+ - config_name: materials_science
80
+ data_files: materials_science/*.parquet
81
+ - config_name: mathematics
82
+ data_files: mathematics/*.parquet
83
+ - config_name: mechanical_engineering
84
+ data_files: mechanical_engineering/*.parquet
85
+ - config_name: medical
86
+ data_files: medical/*.parquet
87
+ - config_name: mining_engineering
88
+ data_files: mining_engineering/*.parquet
89
+ - config_name: music_and_dance
90
+ data_files: music_and_dance/*.parquet
91
+ - config_name: news
92
+ data_files:
93
+ - news/*.parquet
94
+ - topicality/*.parquet
95
+ - config_name: nuclear_science
96
+ data_files: nuclear_science/*.parquet
97
+ - config_name: ocean_science
98
+ data_files: ocean_science/*.parquet
99
+ - config_name: optical_engineering
100
+ data_files: optical_engineering/*.parquet
101
+ - config_name: painting
102
+ data_files: painting/*.parquet
103
+ - config_name: pet
104
+ data_files: pet/*.parquet
105
+ - config_name: petroleum_and_natural_gas_engineering
106
+ data_files: petroleum_and_natural_gas_engineering/*.parquet
107
+ - config_name: philosophy
108
+ data_files: philosophy/*.parquet
109
+ - config_name: photo
110
+ data_files: photo/*.parquet
111
+ - config_name: physics
112
+ data_files: physics/*.parquet
113
+ - config_name: politics
114
+ data_files: politics/*.parquet
115
+ - config_name: psychology
116
+ data_files: psychology/*.parquet
117
+ - config_name: public_administration
118
+ data_files: public_administration/*.parquet
119
+ - config_name: relationship
120
+ data_files: relationship/*.parquet
121
+ - config_name: sociology
122
+ data_files: sociology/*.parquet
123
+ - config_name: sports
124
+ data_files: sports/*.parquet
125
+ - config_name: statistics
126
+ data_files: statistics/*.parquet
127
+ - config_name: systems_science
128
+ data_files: systems_science/*.parquet
129
+ - config_name: textile_science
130
+ data_files: textile_science/*.parquet
131
+ - config_name: transportation_engineering
132
+ data_files: transportation_engineering/*.parquet
133
+ - config_name: travel
134
+ data_files: travel/*.parquet
135
+ - config_name: urban_planning
136
+ data_files: urban_planning/*.parquet
137
+ - config_name: weapons_science
138
+ data_files: weapons_science/*.parquet
139
+ - config_name: all
140
+ data_files: "*/*.parquet"
141
+ ---
142
+
143
+ # FineFineWeb-1M
144
+
145
+ 64 million (1M per class) rows sampled from
146
+ [FineFineWeb](https://huggingface.co/datasets/m-a-p/FineFineWeb).
147
+
148
+ ## Classes
149
+
150
+ FineFineWeb has 67 total classes. We merge 3 into other classes, which gives 64
151
+ classes:
152
+
153
+ - `topicality` into `news`
154
+ - `hydraulic_engineering` into `civil_engineering`
155
+ - `movie` into `drama_and_film`
156
+
157
+ ## Usage
158
+
159
+ ```python
160
+ from datasets import load_dataset
161
+
162
+ # One class, 1M rows, about 770 MiB each
163
+ ds_sports = load_dataset("much1na/FineFineWeb-1M", "sports", split="train")
164
+ ds_history = load_dataset("much1na/FineFineWeb-1M", "history", split="train")
165
+ ds_law = load_dataset("much1na/FineFineWeb-1M", "law", split="train")
166
+
167
+ # All classes, 64M rows, about 50,700 MiB
168
+ ds_all = load_dataset("much1na/FineFineWeb-1M", "all", split="train")
169
+ ```