liujch1998 commited on
Commit
0067690
1 Parent(s): 8c4a00c

Remove Dolma 1.6

Browse files
Files changed (1) hide show
  1. constants.py +1 -1
constants.py CHANGED
@@ -3,7 +3,6 @@ import os
3
  # options
4
  INDEX_BY_DESC = {
5
  'Dolma-v1.7 (2.6T tokens)': 'v4_dolma-v1_7_llama',
6
- 'Dolma-v1.6 (3.1T tokens)': 'v4_dolma-v1_6_llama',
7
  'RedPajama (1.4T tokens)': 'v4_rpj_llama_s4',
8
  'Pile-train (380B tokens)': 'v4_piletrain_llama',
9
  'C4-train (200B tokens)': 'v4_c4train_llama',
@@ -11,6 +10,7 @@ INDEX_BY_DESC = {
11
  # 'Pile-val (GPT-2 tokenizer), 380M tokens': 'v4_pileval_gpt2',
12
  # 'Dolma-v1.6-sample (OLMo tokenizer), 8.0B tokens': 'v4_dolmasample_olmo',
13
  # 'Dolma-v1.6-sample (9.2B tokens)': 'v4_dolma-v1_6-sample_llama',
 
14
  # 'Dolma-v1.6-wiki (4.3B tokens)': 'v4_dolma-v1_6-wiki_llama',
15
  # 'Dolma-v1.6-books (5.8B tokens)': 'v4_dolma-v1_6-books_llama',
16
  # 'Dolma-v1.6-pes2o (69B tokens)': 'v4_dolma-v1_6-pes2o_llama',
 
3
  # options
4
  INDEX_BY_DESC = {
5
  'Dolma-v1.7 (2.6T tokens)': 'v4_dolma-v1_7_llama',
 
6
  'RedPajama (1.4T tokens)': 'v4_rpj_llama_s4',
7
  'Pile-train (380B tokens)': 'v4_piletrain_llama',
8
  'C4-train (200B tokens)': 'v4_c4train_llama',
 
10
  # 'Pile-val (GPT-2 tokenizer), 380M tokens': 'v4_pileval_gpt2',
11
  # 'Dolma-v1.6-sample (OLMo tokenizer), 8.0B tokens': 'v4_dolmasample_olmo',
12
  # 'Dolma-v1.6-sample (9.2B tokens)': 'v4_dolma-v1_6-sample_llama',
13
+ # 'Dolma-v1.6 (3.1T tokens)': 'v4_dolma-v1_6_llama',
14
  # 'Dolma-v1.6-wiki (4.3B tokens)': 'v4_dolma-v1_6-wiki_llama',
15
  # 'Dolma-v1.6-books (5.8B tokens)': 'v4_dolma-v1_6-books_llama',
16
  # 'Dolma-v1.6-pes2o (69B tokens)': 'v4_dolma-v1_6-pes2o_llama',