hylcool commited on
Commit
1c99644
1 Parent(s): 8230129

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +85 -0
README.md CHANGED
@@ -1,3 +1,88 @@
1
  ---
2
  license: apache-2.0
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: apache-2.0
3
+ task_categories:
4
+ - text-generation
5
+ language:
6
+ - en
7
+ tags:
8
+ - data-juicer
9
+ - pretraining
10
+ size_categories:
11
+ - 10M<n<100M
12
  ---
13
+
14
+ # RedPajama -- CommonCrawl-2021-04 (refined by Data-Juicer)
15
+
16
+ A refined version of CommonCrawl-2021-04 dataset in RedPajama by [Data-Juicer](https://github.com/alibaba/data-juicer). Removing some "bad" samples from the original dataset to make it higher-quality.
17
+
18
+ This dataset is usually used to pretrain a Large Language Model.
19
+
20
+ **Notice**: Here is a small subset for previewing. The whole dataset is available [here](https://dail-wlcb.oss-cn-wulanchabu.aliyuncs.com/LLM_data/our_refined_datasets/pretraining/redpajama-cc-refine-results/redpajama-cc-2021-04-refine-result.jsonl) (About 284GB).
21
+
22
+ ## Dataset Information
23
+
24
+ - Number of samples: 44,724,752 (Keep ~45.23% from the original dataset)
25
+
26
+ ## Refining Recipe
27
+ ```yaml
28
+ # global parameters
29
+ project_name: 'Data-Juicer-recipes-cc-2021-04'
30
+ dataset_path: '/path/to/your/dataset' # path to your dataset directory or file
31
+ export_path: '/path/to/your/dataset.jsonl'
32
+
33
+ np: 50 # number of subprocess to process your dataset
34
+ open_tracer: true
35
+
36
+ # process schedule
37
+ # a list of several process operators with their arguments
38
+ process:
39
+ - document_simhash_deduplicator:
40
+ tokenization: space
41
+ window_size: 6
42
+ lowercase: true
43
+ ignore_pattern: '\p{P}'
44
+ num_blocks: 6
45
+ hamming_distance: 4
46
+
47
+ - clean_email_mapper:
48
+ - clean_links_mapper:
49
+ - fix_unicode_mapper:
50
+ - punctuation_normalization_mapper:
51
+ - whitespace_normalization_mapper:
52
+
53
+ - alphanumeric_filter:
54
+ tokenization: false
55
+ min_ratio: 0.7494 # 3sigma
56
+ max_ratio: 0.8595 # 3sigma -- 1001790
57
+ - average_line_length_filter: # for code
58
+ max_len: 1500 # < 3sigma (2817) -- 541131
59
+ - character_repetition_filter:
60
+ rep_len: 10
61
+ max_ratio: 0.3 # > 3sigma (0.1463) -- 159152
62
+ - flagged_words_filter:
63
+ lang: en
64
+ tokenization: true
65
+ max_ratio: 0.0019 # 3sigma -- 184714
66
+ - language_id_score_filter: # remove language filter
67
+ min_score: 0.786 # 3sigma -- 1995115
68
+ - maximum_line_length_filter: # for code
69
+ max_len: 5000 # < 3sigma -- 1076085
70
+ - perplexity_filter:
71
+ lang: en
72
+ max_ppl: 5000 # < 3sigma -- 906649
73
+ - special_characters_filter:
74
+ min_ratio: 0.15 # > 3sigma
75
+ max_ratio: 0.35 # > 3sigma -- 1046590
76
+ - text_length_filter:
77
+ max_len: 61592 # 3sigma -- 1114727
78
+ - words_num_filter:
79
+ lang: en
80
+ tokenization: true
81
+ min_num: 20 # > 3sigma
82
+ max_num: 12241 # 3sigma -- 1120334
83
+ - word_repetition_filter:
84
+ lang: en
85
+ tokenization: true
86
+ rep_len: 10
87
+ max_ratio: 0.3105 # 3sigma -- 2234933
88
+ ```