pierreguillou commited on
Commit
3ac173e
·
1 Parent(s): 6ce253b

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +19 -2
README.md CHANGED
@@ -48,7 +48,7 @@ model-index:
48
  value: 0.7739
49
  - name: accuracy
50
  type: accuracy
51
- value: 0.87
52
  ---
53
 
54
  # Document Understanding model (finetuned LiLT base at paragraph level on DocLayNet base)
@@ -65,7 +65,24 @@ It achieves the following results on the evaluation set:
65
 
66
  ## Accuracy at paragraph level
67
 
68
- ![Accuracy of the labeled blocks of the dataset DocLayNet base of test (model: LayoutXLM base finetuned on DocLayNet base)](https://huggingface.co/pierreguillou/layout-xlm-base-finetuned-with-DocLayNet-base-at-paragraphlevel-ml512/resolve/main/docs/accuracy_labeled_blocks_DocLayNet_base_test_LayoutXLM_base_paragraph_level_512.png)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
69
 
70
  ## References
71
 
 
48
  value: 0.7739
49
  - name: accuracy
50
  type: accuracy
51
+ value: 0.8655
52
  ---
53
 
54
  # Document Understanding model (finetuned LiLT base at paragraph level on DocLayNet base)
 
65
 
66
  ## Accuracy at paragraph level
67
 
68
+ ![Confusion matrix of the labeled blocks of the dataset DocLayNet base of test (model: LayoutXLM base finetuned on DocLayNet base)](https://huggingface.co/pierreguillou/layout-xlm-base-finetuned-with-DocLayNet-base-at-paragraphlevel-ml512/resolve/main/docs/confusion_matrix_labeled_blocks_DocLayNet_base_test_LayoutXLM_base_paragraph_level_512.png)
69
+
70
+ Paragraph Accuracy: 86.55%
71
+
72
+ ![Paragraphs labels vs accuracy (%) of the dataset DocLayNet base of test (model: LayoutXLM base finetuned on DocLayNet base))](https://huggingface.co/pierreguillou/layout-xlm-base-finetuned-with-DocLayNet-base-at-paragraphlevel-ml512/resolve/main/docs/confusion_matrix_labeled_blocks_DocLayNet_base_test_LayoutXLM_base_paragraph_level_512.png)
73
+
74
+ Accuracy by label
75
+ - Caption: 63.76%
76
+ - Footnote: 31.91%
77
+ - Formula: 95.33%
78
+ - List-item: 79.31%
79
+ - Page-footer: 99.51%
80
+ - Page-header: 88.75%
81
+ - Picture: 90.91%
82
+ - Section-header: 83.16%
83
+ - Table: 68.25%
84
+ - Text: 91.37%
85
+ - Title: 50.0%
86
 
87
  ## References
88