birgermoell commited on
Commit
b452553
1 Parent(s): d1850c7

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +51 -4
README.md CHANGED
@@ -10,6 +10,45 @@ part of the wiki40b dataset.
10
 
11
  https://huggingface.co/datasets/wiki40b
12
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
13
 
14
  ## Data cleaning and preprocessing
15
  The data was cleaned and preprocessed using the following script. Make sure to install depencies for beam_runner to make the dataset work.
@@ -26,10 +65,18 @@ def load_and_clean_wiki():
26
  return filtered_dataset
27
 
28
  def filter_wikipedia(batch):
29
- batch["text"] = " ".join(batch["text"].split("\n_START_SECTION_\n"))
30
- batch["text"] = " ".join(batch["text"].split("\n_START_ARTICLE_\n"))
31
- batch["text"] = " ".join(batch["text"].split("\n_START_ARTICLE_\n"))
32
- batch["text"] = " ".join(batch["text"].split("\n_START_PARAGRAPH_\n"))
 
 
 
 
 
 
 
 
33
  batch["text"] = " ".join(batch["text"].split("_NEWLINE_"))
34
  batch["text"] = " ".join(batch["text"].split("\xa0"))
35
  return batch
 
10
 
11
  https://huggingface.co/datasets/wiki40b
12
 
13
+ ## Model series
14
+ This model is part of a series of models training on TPU with Flax Jax during Huggingface Flax/Jax challenge.
15
+
16
+ ## Gpt models
17
+
18
+ ## Swedish Gpt
19
+ https://huggingface.co/birgermoell/swedish-gpt/
20
+
21
+ ## Swedish gpt wiki
22
+ https://huggingface.co/flax-community/swe-gpt-wiki
23
+
24
+ # Nordic gpt wiki
25
+ https://huggingface.co/flax-community/nordic-gpt-wiki
26
+
27
+ ## Dansk gpt wiki
28
+ https://huggingface.co/flax-community/dansk-gpt-wiki
29
+
30
+ ## Norsk gpt wiki
31
+ https://huggingface.co/flax-community/norsk-gpt-wiki
32
+
33
+ ## Roberta models
34
+
35
+ ## Nordic Roberta Wiki
36
+ https://huggingface.co/flax-community/nordic-roberta-wiki
37
+
38
+ ## Swe Roberta Wiki Oscar
39
+ https://huggingface.co/flax-community/swe-roberta-wiki-oscar
40
+
41
+ ## Roberta Swedish Scandi
42
+ https://huggingface.co/birgermoell/roberta-swedish-scandi
43
+
44
+ ## Roberta Swedish
45
+ https://huggingface.co/birgermoell/roberta-swedish
46
+
47
+ ## Swedish T5 model
48
+ https://huggingface.co/birgermoell/t5-base-swedish
49
+
50
+
51
+
52
 
53
  ## Data cleaning and preprocessing
54
  The data was cleaned and preprocessed using the following script. Make sure to install depencies for beam_runner to make the dataset work.
 
65
  return filtered_dataset
66
 
67
  def filter_wikipedia(batch):
68
+ batch["text"] = " ".join(batch["text"].split("\
69
+ _START_SECTION_\
70
+ "))
71
+ batch["text"] = " ".join(batch["text"].split("\
72
+ _START_ARTICLE_\
73
+ "))
74
+ batch["text"] = " ".join(batch["text"].split("\
75
+ _START_ARTICLE_\
76
+ "))
77
+ batch["text"] = " ".join(batch["text"].split("\
78
+ _START_PARAGRAPH_\
79
+ "))
80
  batch["text"] = " ".join(batch["text"].split("_NEWLINE_"))
81
  batch["text"] = " ".join(batch["text"].split("\xa0"))
82
  return batch