Remove unused tiny configs, add description of micro config

README.md

@@ -11,6 +11,8 @@ licenses:
 multilinguality:
 - monolingual
 size_categories:
+  micro:
+    - 120k
   tiny:
     - 1M<n<10M
   small:
@@ -129,13 +131,14 @@ For ease of use under different storage capacities, the following incremental co
 
 | subset | train size (docs, words, download + preproc disk space) | validation size |
 |:-------|--------------------------------------------------------:|----------------:|
+| micro  |                             125k docs, 23M words (<1GB) |        16k docs |
 | tiny   |                        6M docs, 2B words (6 GB + 15 GB) |        16k docs |
 | small  |                      15M docs, 6B words (14 GB + 36 GB) |        16k docs |
 | medium |                     31M docs, 12B words (28 GB + 72 GB) |        32k docs |
 | large  |                    47M docs, 19B words (42 GB + 108 GB) |        48k docs |
 | full   |                    64M docs, 25B words (58 GB + 148 GB) |        64k docs |
 
-For each subset as `tiny` there also exists a config `tiny_en_nl` that interleaves examples from the cleaned
+For each subset there also exists a config `<name>_en_nl` that interleaves `nl` and `en` examples from the cleaned
 `en` variant of C4.
 
 You can load any subset like this:

mc4_nl_cleaned.py

@@ -64,16 +64,6 @@ _CONFIGS = dict(
     medium={"train": 500, "validation": 2, "estimate": "50GB"},
     large={"train": 750, "validation": 3, "estimate": "75GB"},
     full={"train": 1024, "validation": 4, "estimate": "103GB"},
-    tiny_0={"start": 0, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_1={"start": 100, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_2={"start": 200, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_3={"start": 300, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_4={"start": 400, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_5={"start": 500, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_6={"start": 600, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_7={"start": 700, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_8={"start": 800, "train": 100, "validation": 1, "estimate": "10GB"},
-    tiny_9={"start": 900, "train": 100, "validation": 1, "estimate": "10GB"},
 )
 
 
@@ -105,20 +95,6 @@ class Mc4(datasets.GeneratorBasedBuilder):
         for name in _CONFIG_NAMES
     ]
 
-    for i in range(10):
-        BUILDER_CONFIGS.append(
-            Mc4NlCleanedConfig(
-                name=f"tiny_{i}",
-                version=datasets.Version("1.0.0"),
-                description=textwrap.dedent(
-                    f"""\
-                    The tiny_{i} slice of the full cleaned version of the Dutch portion of the multilingual C4 corpus.
-                    Estimated size of compressed files: 10GB
-                    """
-                ),
-            ),
-        )
-
     BUILDER_CONFIGS += [
         Mc4NlCleanedConfig(
             name=f"{name}{_CONFIG_EN_NL_SUFFIX}",