hylcool commited on
Commit
76674ef
1 Parent(s): 94e19f0

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +82 -0
README.md CHANGED
@@ -1,3 +1,85 @@
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
+ - 100K<n<1M
12
  ---
13
+
14
+ # RedPajama -- Book (refined by Data-Juicer)
15
+
16
+ A refined version of Book 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-book-refine-result.jsonl) (About 91GB).
21
+
22
+ ## Dataset Information
23
+
24
+ - Number of samples: 195,983 (Keep ~95.51% from the original dataset)
25
+
26
+ ## Refining Recipe
27
+ ```yaml
28
+ # global parameters
29
+ project_name: 'Data-Juicer-recipes-book'
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
+ - clean_email_mapper:
40
+ - clean_links_mapper:
41
+ - fix_unicode_mapper:
42
+ - punctuation_normalization_mapper:
43
+ - whitespace_normalization_mapper:
44
+
45
+ - alphanumeric_filter:
46
+ tokenization: false
47
+ min_ratio: 0.55 # <3sigma (0.697)
48
+ max_ratio: 0.854 # 3sigma
49
+ - average_line_length_filter: # for code
50
+ max_len: 500 # >3sigma (364)
51
+ - character_repetition_filter:
52
+ rep_len: 10
53
+ max_ratio: 0.2 # >3sigma (0.12)
54
+ - flagged_words_filter:
55
+ lang: en
56
+ tokenization: true
57
+ max_ratio: 0.00047 # 3sigma
58
+ - language_id_score_filter: # remove language filter
59
+ min_score: 0.2
60
+ - maximum_line_length_filter: # for code
61
+ max_len: 13381 # 3sigma
62
+ - perplexity_filter:
63
+ lang: en
64
+ max_ppl: 6000 # <3sigma (16516)
65
+ - special_characters_filter:
66
+ max_ratio: 0.5 # >3sigma (0.32)
67
+ - words_num_filter:
68
+ lang: en
69
+ tokenization: true
70
+ min_num: 1000
71
+ max_num: 539754 # 3sigma
72
+ - word_repetition_filter:
73
+ lang: en
74
+ tokenization: true
75
+ rep_len: 10
76
+ max_ratio: 0.194 # 3sigma
77
+
78
+ - document_simhash_deduplicator:
79
+ tokenization: space
80
+ window_size: 6
81
+ lowercase: true
82
+ ignore_pattern: '\p{P}'
83
+ num_blocks: 6
84
+ hamming_distance: 4
85
+ ```