Hugging Face's logo

impira
/
layoutlm-invoices

Document Question Answering Transformers PyTorch Safetensors English layoutlm pdf invoices Inference Endpoints
Model card Files Community
8
Ankur Goyal commited on
Commit
0771b61
•
1 Parent(s): 1a67f8e

Improve README

Browse files
Files changed (4) hide show
  1. README.md +6 -2
  2. after.png +0 -0
  3. before.png +0 -0
  4. demo.png +0 -0
README.md CHANGED
@@ -16,9 +16,13 @@ invoices as well as both [SQuAD2.0](https://huggingface.co/datasets/squad_v2) an
16
  ## Non-consecutive tokens
17
 
18
  Unlike other QA models, which can only extract consecutive tokens (because they predict the start and end of a sequence), this model can predict longer-range, non-consecutive sequences with an additional
19
- classifier head. For example, it can extract the two-line address as below:
20
 
21
- ![Two-line Address](./demo.png)
 
 
 
 
22
 
23
  ## Getting started with the model
24
 
16
  ## Non-consecutive tokens
17
 
18
  Unlike other QA models, which can only extract consecutive tokens (because they predict the start and end of a sequence), this model can predict longer-range, non-consecutive sequences with an additional
19
+ classifier head. For example, QA models often encounter this failure mode:
20
 
21
+ ![Broken Address](./before.png)
22
+
23
+ However this model is able to predict non-consecutive tokens and therefore the address correctly:
24
+
25
+ ![Two-line Address](./after.png)
26
 
27
  ## Getting started with the model
28
 
after.png ADDED
before.png ADDED
demo.png DELETED
Binary file (630 kB)