pierreguillou
commited on
Commit
•
b05819d
1
Parent(s):
9c8291e
Update app.py
Browse files
app.py
CHANGED
@@ -52,13 +52,14 @@ os.system('python -m pip install --upgrade pip')
|
|
52 |
|
53 |
## model / feature extractor / tokenizer
|
54 |
|
55 |
-
from transformers import LayoutLMv2ForTokenClassification # LayoutXLMTokenizerFast,
|
56 |
-
|
57 |
import torch
|
58 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
59 |
|
60 |
# model
|
61 |
-
|
|
|
|
|
|
|
62 |
model = LayoutLMv2ForTokenClassification.from_pretrained(model_id);
|
63 |
model.to(device);
|
64 |
|
@@ -68,6 +69,7 @@ feature_extractor = LayoutLMv2FeatureExtractor(apply_ocr=False)
|
|
68 |
|
69 |
# tokenizer
|
70 |
from transformers import AutoTokenizer
|
|
|
71 |
tokenizer = AutoTokenizer.from_pretrained(tokenizer_id)
|
72 |
|
73 |
# APP outputs
|
|
|
52 |
|
53 |
## model / feature extractor / tokenizer
|
54 |
|
|
|
|
|
55 |
import torch
|
56 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
57 |
|
58 |
# model
|
59 |
+
from transformers import LayoutLMv2ForTokenClassification
|
60 |
+
|
61 |
+
model_id = "pierreguillou/layout-xlm-base-finetuned-with-DocLayNet-base-at-linelevel-ml384"
|
62 |
+
|
63 |
model = LayoutLMv2ForTokenClassification.from_pretrained(model_id);
|
64 |
model.to(device);
|
65 |
|
|
|
69 |
|
70 |
# tokenizer
|
71 |
from transformers import AutoTokenizer
|
72 |
+
tokenizer_id = "xlm-roberta-base"
|
73 |
tokenizer = AutoTokenizer.from_pretrained(tokenizer_id)
|
74 |
|
75 |
# APP outputs
|