kassemsabeh
commited on
Commit
·
42466a5
1
Parent(s):
b0cd2d3
Add application
Browse files- app.py +66 -0
- requirements.txt +2 -0
app.py
ADDED
@@ -0,0 +1,66 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
import os
|
4 |
+
from transformers import AutoTokenizer, T5ForConditionalGeneration
|
5 |
+
|
6 |
+
model_id = 'ksabeh/t5-base-qpave'
|
7 |
+
max_input_length = 512
|
8 |
+
max_target_length = 20
|
9 |
+
auth_token = os.environ.get('TOKEN')
|
10 |
+
|
11 |
+
model = T5ForConditionalGeneration.from_pretrained(model_id, use_auth_token=auth_token)
|
12 |
+
tokenizer = AutoTokenizer.from_pretrained(model_id, use_auth_token=auth_token)
|
13 |
+
|
14 |
+
def predict(cg_attribute, text, fg_attribute, category):
|
15 |
+
input = f"{fg_attribute}: {text}"
|
16 |
+
model_input = tokenizer(input, max_length=max_input_length, truncation=True,
|
17 |
+
padding="max_length")
|
18 |
+
model_input = {k:torch.unsqueeze(torch.tensor(v),dim=0) for k,v in model_input.items()}
|
19 |
+
predictions = model.generate(**model_input, num_beams=4, do_sample=True, max_length=10)
|
20 |
+
return tokenizer.batch_decode(predictions, skip_special_tokens=True)[0]
|
21 |
+
|
22 |
+
# iface = gr.Interface(
|
23 |
+
# predict,
|
24 |
+
# inputs=["text", "text", "text", "text"],
|
25 |
+
# outputs=['text'],
|
26 |
+
# title="QPAVE",
|
27 |
+
# examples=[["Arriba Salsa Garlic and Cilantro, 16 oz", "Food"],
|
28 |
+
# ["MV Verholen Black GPS Ball Mount for BMW K1200S K1200R K1300S K1300R Black GPS Ball Mount VER-4901-10181", "Toys"],
|
29 |
+
# ["Mitsubishi 3000GT License Plate Frame (Zince Metal)", "Automotive"],
|
30 |
+
# ["Fun Fire Truck Pinata Personalized", "Toys"],
|
31 |
+
# ["White Chocolate Caramel Gourmet Popcorn Kelly", "Food"]
|
32 |
+
# ]
|
33 |
+
# )
|
34 |
+
|
35 |
+
# iface.launch()
|
36 |
+
|
37 |
+
demo = gr.Interface(
|
38 |
+
predict,
|
39 |
+
[
|
40 |
+
gr.Textbox(
|
41 |
+
label = "Coarse-grained Attribute",
|
42 |
+
info = "The coarse-grained attribute name",
|
43 |
+
lines = 1,
|
44 |
+
),
|
45 |
+
gr.Textbox(
|
46 |
+
label = "Context",
|
47 |
+
info = "The value of the coarse-grained attribute",
|
48 |
+
lines = 1,
|
49 |
+
),
|
50 |
+
gr.Textbox(
|
51 |
+
label = "Fine-grained Attribute",
|
52 |
+
info = "The target fine-grained attribute name",
|
53 |
+
lines = 1,
|
54 |
+
),
|
55 |
+
gr.Textbox(
|
56 |
+
label = "Category",
|
57 |
+
info = "The product category",
|
58 |
+
lines = 1,
|
59 |
+
)
|
60 |
+
],
|
61 |
+
"dataframe",
|
62 |
+
title="QPAVE",
|
63 |
+
examples=[["Processor", "3ghz intel core i5", "Brand Name", "Computers & Tablets"]
|
64 |
+
],
|
65 |
+
cache_examples = True
|
66 |
+
)
|
requirements.txt
ADDED
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
1 |
+
transformers
|
2 |
+
torch
|