system HF staff commited on
Commit
e4a8655
1 Parent(s): ea3bbd2

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +4 -0
README.md CHANGED
@@ -31,6 +31,7 @@ For text generation, like other XLNet models, a longer context will help generat
31
  ## Training data
32
 
33
  The following is the list of data sources. Total characters is about 507M.
 
34
  | Data | % |
35
  | ------------------------------------------------- | --: |
36
  | News Articles / Blogs | 58% |
@@ -40,6 +41,7 @@ The following is the list of data sources. Total characters is about 507M.
40
  | Online Fiction | 1% |
41
 
42
  The following is the distribution of different languages within the corpus.
 
43
  | Language | % |
44
  | ------------------------------------------------- | --: |
45
  | Standard Chinese | 62% |
@@ -49,6 +51,7 @@ The following is the distribution of different languages within the corpus.
49
  ## Training procedure
50
 
51
  Model was trained on a single TPUv3 from the official repo with the default parameters.
 
52
  | Parameter | Value |
53
  | ------------------------------------------------ | ----: |
54
  | Batch Size | 32 |
@@ -60,6 +63,7 @@ Model was trained on a single TPUv3 from the official repo with the default para
60
  ## Eval results
61
 
62
  Average evaluation task results over 10 runs. Comparison using the original repo model and code. Chinese models are available from [Joint Laboratory of HIT and iFLYTEK Research (HFL)](https://huggingface.co/hfl)
 
63
  | Model | DRCD (EM/F1) | openrice-senti | lihkg-cat | wordshk-sem |
64
  |:-----------:|:------------:|:--------------:|:---------:|:-----------:|
65
  | Chinese | 82.8 / 91.8 | 79.8 | 70.7 | 72.0 / 78.9*|
 
31
  ## Training data
32
 
33
  The following is the list of data sources. Total characters is about 507M.
34
+
35
  | Data | % |
36
  | ------------------------------------------------- | --: |
37
  | News Articles / Blogs | 58% |
 
41
  | Online Fiction | 1% |
42
 
43
  The following is the distribution of different languages within the corpus.
44
+
45
  | Language | % |
46
  | ------------------------------------------------- | --: |
47
  | Standard Chinese | 62% |
 
51
  ## Training procedure
52
 
53
  Model was trained on a single TPUv3 from the official repo with the default parameters.
54
+
55
  | Parameter | Value |
56
  | ------------------------------------------------ | ----: |
57
  | Batch Size | 32 |
 
63
  ## Eval results
64
 
65
  Average evaluation task results over 10 runs. Comparison using the original repo model and code. Chinese models are available from [Joint Laboratory of HIT and iFLYTEK Research (HFL)](https://huggingface.co/hfl)
66
+
67
  | Model | DRCD (EM/F1) | openrice-senti | lihkg-cat | wordshk-sem |
68
  |:-----------:|:------------:|:--------------:|:---------:|:-----------:|
69
  | Chinese | 82.8 / 91.8 | 79.8 | 70.7 | 72.0 / 78.9*|