Hugging Face's logo

emiltj
/
da_dacy_small_DANSK_ner

Token Classification spaCy Danish Eval Results
Model card Files Community
emiltj commited on
Commit
0c716bc
•
1 Parent(s): 045d3b0

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -35
README.md CHANGED
@@ -70,13 +70,14 @@ DaCy also contains guides on usage of the package as well as behavioural test fo
70
 
71
  ### Performance tables
72
 
73
- The three tables below show the F1-scores for the three DaCy fine-grained models.
74
- | | DaCy large | DaCy medium | DaCy small |
75
  |:---:|:---:|:---:|:---:|
76
- | F1-score | 0.823 | 0.806 | 0.776 |
77
  | Recall | 0.834 | 0.818 | 0.77 |
78
  | Precision | 0.813 | 0.794 | 0.781 |
79
 
 
80
  | Named-entity type | DaCy large | DaCy medium | DaCy small |
81
  |:---:|:---:|:---:|:---:|
82
  | CARDINAL | 0.874 | 0.781 | 0.887 |
@@ -98,6 +99,7 @@ The three tables below show the F1-scores for the three DaCy fine-grained models
98
  | TIME | 0.643 | 0.571 | 0.71 |
99
  | WORK OF ART | 0.494 | 0.639 | 0.488 |
100
 
 
101
  | Domain | DaCy large | DaCy medium | DaCy small |
102
  |:---:|:---:|:---:|:---:|
103
  | All domains combined | 0.823 | 0.806 | 0.776 |
@@ -109,35 +111,3 @@ The three tables below show the F1-scores for the three DaCy fine-grained models
109
  | Web | 0.826 | 0.802 | 0.756 |
110
  | Wiki and Books | 0.778 | 0.838 | 0.709 |
111
 
112
-
113
-
114
- | Domain | DaCy large | DaCy medium | DaCy small |
115
- |:--------------------: |:----------: |:-----------: |:----------: |
116
- | All domains combined | 0.823 | 0.806 | 0.776 |
117
- | Conversation | 0.796 | 0.718 | 0.82|
118
- | Dannet | 0.75 | 0.667 | 1 |
119
- | Legal | 0.852 | 0.854 | 0.866 |
120
- | News | 0.841 | 0.759 | 0.86 |
121
- | Social Media | 0.793 | 0.847 | 0.8 |
122
- | Web | 0.826 | 0.802 | 0.756 |
123
- | Wiki and Books | 0.778 | 0.838 | 0.709 |
124
-
125
- | Domain | DaCy large | DaCy medium | DaCy small |
126
- |:--------------------: |:----------: |:-----------: |:----------: |
127
- | All domains combined | 0.823 | 0.806 | 0.776 |
128
- | Conversation | 0.796 | 0.718 | 0.82 |
129
- | Dannet | 0.75 | 0.667 | 1 |
130
- | Legal | 0.852 | 0.854 | 0.866 |
131
- | News | 0.841 | 0.759 | 0.86 |
132
- | Social Media | 0.793 | 0.847 | 0.8 |
133
- | Web | 0.826 | 0.802 | 0.756 |
134
- | Wiki and Books | 0.778 | 0.838 | 0.709 |
135
-
136
-
137
- | | DaCy fine-grained model | | |
138
- |:---:|:---:|:---:|:---:|
139
- | | | Medium | Small |
140
- | F1-score | 0.823 | 0.806 | 0.776 |
141
- | Recall | 0.834 | 0.818 | 0.77 |
142
- | Precision | 0.813 | 0.794 | 0.781 |
143
-
70
 
71
  ### Performance tables
72
 
73
+ The table below shows the F1, recall and precision of the three DaCy fine-grained models.
74
+ | Score | DaCy large | DaCy medium | DaCy small |
75
  |:---:|:---:|:---:|:---:|
76
+ | F1 | 0.823 | 0.806 | 0.776 |
77
  | Recall | 0.834 | 0.818 | 0.77 |
78
  | Precision | 0.813 | 0.794 | 0.781 |
79
 
80
+ The table below shows the F1 of the three DaCy fine-grained models within each named entity type.
81
  | Named-entity type | DaCy large | DaCy medium | DaCy small |
82
  |:---:|:---:|:---:|:---:|
83
  | CARDINAL | 0.874 | 0.781 | 0.887 |
99
  | TIME | 0.643 | 0.571 | 0.71 |
100
  | WORK OF ART | 0.494 | 0.639 | 0.488 |
101
 
102
+ The table below shows the F1 of the three DaCy fine-grained models within each domain of texts in DANSK.
103
  | Domain | DaCy large | DaCy medium | DaCy small |
104
  |:---:|:---:|:---:|:---:|
105
  | All domains combined | 0.823 | 0.806 | 0.776 |
111
  | Web | 0.826 | 0.802 | 0.756 |
112
  | Wiki and Books | 0.778 | 0.838 | 0.709 |
113