yeager07 commited on
Commit
ce75d5d
·
1 Parent(s): 6362bd8

Add application file

Browse files
Files changed (1) hide show
  1. app.py +41 -4
app.py CHANGED
@@ -1,7 +1,44 @@
1
  import gradio as gr
 
 
 
2
 
3
- def greet(name):
4
- return "Hello " + name + "!!"
5
 
6
- demo = gr.Interface(fn=greet, inputs="text", outputs="text")
7
- demo.launch()
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  import gradio as gr
2
+ import onnxruntime as rt
3
+ from transformers import AutoTokenizer
4
+ import torch, json
5
 
6
+ # Load tokenizer
7
+ tokenizer = AutoTokenizer.from_pretrained("allenai/scibert_scivocab_uncased")
8
 
9
+ # Load encoded subject dictionary
10
+ with open("subject_types_encoded.json", "r") as fp:
11
+ encode_subject_types = json.load(fp)
12
+
13
+ subjects = list(encode_subject_types.keys())
14
+
15
+ # Load ONNX model
16
+ inf_session = rt.InferenceSession("Scientific Articles-classifier-quantized.onnx")
17
+ input_name = inf_session.get_inputs()[0].name
18
+ output_name = inf_session.get_outputs()[0].name
19
+
20
+ # Prediction function
21
+ def classify_subjects(abstract: str):
22
+ inputs = tokenizer(
23
+ abstract,
24
+ padding="max_length",
25
+ truncation=True,
26
+ max_length=512,
27
+ return_tensors="np"
28
+ )
29
+ logits = inf_session.run([output_name], {input_name: inputs["input_ids"]})[0]
30
+ logits = torch.FloatTensor(logits)
31
+ probs = torch.sigmoid(logits)[0]
32
+ return dict(zip(subjects, map(float, probs)))
33
+
34
+ # Gradio Interface
35
+ label = gr.Label(num_top_classes=10)
36
+ iface = gr.Interface(
37
+ fn=classify_subjects,
38
+ inputs=gr.Textbox(lines=5, placeholder="Enter research paper abstract here..."),
39
+ outputs=label,
40
+ title="Multi-Label Classification of Computer Science Research Articles",
41
+ description="Predict multiple subject categories from computer science research paper abstracts."
42
+ )
43
+
44
+ iface.launch()