Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -1,51 +1,11 @@
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
-
import asyncio
|
4 |
from helper_functions import *
|
5 |
from rank_bm25 import BM25L
|
6 |
-
import nest_asyncio
|
7 |
import time
|
8 |
-
nest_asyncio.apply()
|
9 |
-
from aiogoogletrans import Translator
|
10 |
import pprint
|
11 |
-
|
12 |
-
# Initialize the translator
|
13 |
-
translator = Translator()
|
14 |
-
|
15 |
-
async def translate_bulk(bulk: list) -> list:
|
16 |
-
"""
|
17 |
-
Translate the given text to English and return the translated text.
|
18 |
-
|
19 |
-
Args:
|
20 |
-
- text (str): The text to translate.
|
21 |
-
|
22 |
-
Returns:
|
23 |
-
- str: The translated text.
|
24 |
-
"""
|
25 |
-
try:
|
26 |
-
translated_bulk = await translator.translate(bulk, dest="en")
|
27 |
-
translated_bulk = [
|
28 |
-
translated_text.text.lower().strip() for translated_text in translated_bulk
|
29 |
-
]
|
30 |
-
except Exception as e:
|
31 |
-
print(f"Bulk Translation failed: {e}")
|
32 |
-
translated_bulk = [
|
33 |
-
text.lower().strip() for text in bulk
|
34 |
-
] # Use original text if translation fails
|
35 |
-
return translated_bulk
|
36 |
-
|
37 |
-
async def encode_document(document: str):
|
38 |
-
"""_summary_
|
39 |
-
|
40 |
-
Args:
|
41 |
-
document (str): _description_
|
42 |
-
|
43 |
-
Returns:
|
44 |
-
_type_: _description_
|
45 |
-
"""
|
46 |
-
return semantic_model.encode(document, convert_to_tensor=True)
|
47 |
|
48 |
-
|
49 |
start_time = time.time()
|
50 |
normalized_query_list = (
|
51 |
[normalizer.clean_text(query)]
|
@@ -102,14 +62,8 @@ async def predict(query):
|
|
102 |
categorize_time = categorize_end_time - categorize_start_time
|
103 |
except Exception as e:
|
104 |
return {"error": f"An error occurred while categorizing products: {e}"}
|
105 |
-
|
106 |
-
|
107 |
-
translation_start_time = time.time()
|
108 |
-
representation_list = await translate_bulk(tasks)
|
109 |
-
except Exception as e:
|
110 |
-
representation_list = tasks
|
111 |
-
print(f"An error occurred while translating: {e}")
|
112 |
-
translation_time = time.time() - translation_start_time
|
113 |
|
114 |
try:
|
115 |
# Tokenize representations for keyword search
|
@@ -123,16 +77,9 @@ async def predict(query):
|
|
123 |
|
124 |
# Encode representations for semantic search
|
125 |
encode_start_time = time.time()
|
126 |
-
|
127 |
-
embeddings = await asyncio.gather(
|
128 |
-
*[encode_document(document) for document in representation_list]
|
129 |
-
)
|
130 |
-
doc_embeddings = torch.stack(embeddings)
|
131 |
-
except Exception as e:
|
132 |
-
doc_embeddings = semantic_model.encode(
|
133 |
representation_list, convert_to_tensor=True
|
134 |
)
|
135 |
-
print(f"An error occurred while encoding documents: {e}")
|
136 |
encode_end_time = time.time()
|
137 |
encode_time = encode_end_time - encode_start_time
|
138 |
|
@@ -159,8 +106,7 @@ async def predict(query):
|
|
159 |
|
160 |
hits = {"results": results, "time_taken": time_taken, "normalize_query_time": normalize_query_time,
|
161 |
"request_time": request_time, "normalization_time": normalization_time,
|
162 |
-
"
|
163 |
-
"tokenization_time": tokenization_time, "encode_time": encode_time,
|
164 |
"calculate_interrelations_time": calculate_interrelations_time,
|
165 |
"process_time": process_time_taken}
|
166 |
|
|
|
1 |
import gradio as gr
|
2 |
import torch
|
|
|
3 |
from helper_functions import *
|
4 |
from rank_bm25 import BM25L
|
|
|
5 |
import time
|
|
|
|
|
6 |
import pprint
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
|
8 |
+
def predict(query):
|
9 |
start_time = time.time()
|
10 |
normalized_query_list = (
|
11 |
[normalizer.clean_text(query)]
|
|
|
62 |
categorize_time = categorize_end_time - categorize_start_time
|
63 |
except Exception as e:
|
64 |
return {"error": f"An error occurred while categorizing products: {e}"}
|
65 |
+
|
66 |
+
representation_list = tasks
|
|
|
|
|
|
|
|
|
|
|
|
|
67 |
|
68 |
try:
|
69 |
# Tokenize representations for keyword search
|
|
|
77 |
|
78 |
# Encode representations for semantic search
|
79 |
encode_start_time = time.time()
|
80 |
+
doc_embeddings = semantic_model.encode(
|
|
|
|
|
|
|
|
|
|
|
|
|
81 |
representation_list, convert_to_tensor=True
|
82 |
)
|
|
|
83 |
encode_end_time = time.time()
|
84 |
encode_time = encode_end_time - encode_start_time
|
85 |
|
|
|
106 |
|
107 |
hits = {"results": results, "time_taken": time_taken, "normalize_query_time": normalize_query_time,
|
108 |
"request_time": request_time, "normalization_time": normalization_time,
|
109 |
+
"categorize_time": categorize_time, "tokenization_time": tokenization_time, "encode_time": encode_time,
|
|
|
110 |
"calculate_interrelations_time": calculate_interrelations_time,
|
111 |
"process_time": process_time_taken}
|
112 |
|