Yeb Havinga commited on
Commit
72ad1eb
1 Parent(s): 05f41a7

Remove unused tiny configs, add description of micro config

Browse files
Files changed (2) hide show
  1. README.md +4 -1
  2. mc4_nl_cleaned.py +0 -24
README.md CHANGED
@@ -11,6 +11,8 @@ licenses:
11
  multilinguality:
12
  - monolingual
13
  size_categories:
 
 
14
  tiny:
15
  - 1M<n<10M
16
  small:
@@ -129,13 +131,14 @@ For ease of use under different storage capacities, the following incremental co
129
 
130
  | subset | train size (docs, words, download + preproc disk space) | validation size |
131
  |:-------|--------------------------------------------------------:|----------------:|
 
132
  | tiny | 6M docs, 2B words (6 GB + 15 GB) | 16k docs |
133
  | small | 15M docs, 6B words (14 GB + 36 GB) | 16k docs |
134
  | medium | 31M docs, 12B words (28 GB + 72 GB) | 32k docs |
135
  | large | 47M docs, 19B words (42 GB + 108 GB) | 48k docs |
136
  | full | 64M docs, 25B words (58 GB + 148 GB) | 64k docs |
137
 
138
- For each subset as `tiny` there also exists a config `tiny_en_nl` that interleaves examples from the cleaned
139
  `en` variant of C4.
140
 
141
  You can load any subset like this:
 
11
  multilinguality:
12
  - monolingual
13
  size_categories:
14
+ micro:
15
+ - 120k
16
  tiny:
17
  - 1M<n<10M
18
  small:
 
131
 
132
  | subset | train size (docs, words, download + preproc disk space) | validation size |
133
  |:-------|--------------------------------------------------------:|----------------:|
134
+ | micro | 125k docs, 23M words (<1GB) | 16k docs |
135
  | tiny | 6M docs, 2B words (6 GB + 15 GB) | 16k docs |
136
  | small | 15M docs, 6B words (14 GB + 36 GB) | 16k docs |
137
  | medium | 31M docs, 12B words (28 GB + 72 GB) | 32k docs |
138
  | large | 47M docs, 19B words (42 GB + 108 GB) | 48k docs |
139
  | full | 64M docs, 25B words (58 GB + 148 GB) | 64k docs |
140
 
141
+ For each subset there also exists a config `<name>_en_nl` that interleaves `nl` and `en` examples from the cleaned
142
  `en` variant of C4.
143
 
144
  You can load any subset like this:
mc4_nl_cleaned.py CHANGED
@@ -64,16 +64,6 @@ _CONFIGS = dict(
64
  medium={"train": 500, "validation": 2, "estimate": "50GB"},
65
  large={"train": 750, "validation": 3, "estimate": "75GB"},
66
  full={"train": 1024, "validation": 4, "estimate": "103GB"},
67
- tiny_0={"start": 0, "train": 100, "validation": 1, "estimate": "10GB"},
68
- tiny_1={"start": 100, "train": 100, "validation": 1, "estimate": "10GB"},
69
- tiny_2={"start": 200, "train": 100, "validation": 1, "estimate": "10GB"},
70
- tiny_3={"start": 300, "train": 100, "validation": 1, "estimate": "10GB"},
71
- tiny_4={"start": 400, "train": 100, "validation": 1, "estimate": "10GB"},
72
- tiny_5={"start": 500, "train": 100, "validation": 1, "estimate": "10GB"},
73
- tiny_6={"start": 600, "train": 100, "validation": 1, "estimate": "10GB"},
74
- tiny_7={"start": 700, "train": 100, "validation": 1, "estimate": "10GB"},
75
- tiny_8={"start": 800, "train": 100, "validation": 1, "estimate": "10GB"},
76
- tiny_9={"start": 900, "train": 100, "validation": 1, "estimate": "10GB"},
77
  )
78
 
79
 
@@ -105,20 +95,6 @@ class Mc4(datasets.GeneratorBasedBuilder):
105
  for name in _CONFIG_NAMES
106
  ]
107
 
108
- for i in range(10):
109
- BUILDER_CONFIGS.append(
110
- Mc4NlCleanedConfig(
111
- name=f"tiny_{i}",
112
- version=datasets.Version("1.0.0"),
113
- description=textwrap.dedent(
114
- f"""\
115
- The tiny_{i} slice of the full cleaned version of the Dutch portion of the multilingual C4 corpus.
116
- Estimated size of compressed files: 10GB
117
- """
118
- ),
119
- ),
120
- )
121
-
122
  BUILDER_CONFIGS += [
123
  Mc4NlCleanedConfig(
124
  name=f"{name}{_CONFIG_EN_NL_SUFFIX}",
 
64
  medium={"train": 500, "validation": 2, "estimate": "50GB"},
65
  large={"train": 750, "validation": 3, "estimate": "75GB"},
66
  full={"train": 1024, "validation": 4, "estimate": "103GB"},
 
 
 
 
 
 
 
 
 
 
67
  )
68
 
69
 
 
95
  for name in _CONFIG_NAMES
96
  ]
97
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
98
  BUILDER_CONFIGS += [
99
  Mc4NlCleanedConfig(
100
  name=f"{name}{_CONFIG_EN_NL_SUFFIX}",