Spaces:
Sleeping
Sleeping
Chananchida
commited on
Update app.py
Browse files
app.py
CHANGED
@@ -94,42 +94,30 @@ def predict_test(model, tokenizer, embedding_model, context, question, index):
|
|
94 |
question_vector = prepare_sentences_vector([question_vector])
|
95 |
distances, indices = faiss_search(index, question_vector, 3) # Retrieve top 3 indices
|
96 |
|
97 |
-
# most_similar_contexts = []
|
98 |
most_similar_contexts = ''
|
99 |
for i in range(3): # Loop through top 3 indices
|
100 |
most_sim_context = context[indices[0][i]].strip()
|
101 |
# most_similar_contexts.append(most_sim_context)
|
102 |
most_similar_contexts += str(i)+': '+most_sim_context + "\n\n"
|
103 |
|
104 |
-
_time = time.time() - t
|
105 |
-
output = {
|
106 |
-
"user_question": question,
|
107 |
-
"answer": most_similar_contexts,
|
108 |
-
# "answer": Answer,
|
109 |
-
"totaltime": round(_time, 3),
|
110 |
-
"distance": round(distances[0][0], 4)
|
111 |
-
}
|
112 |
-
# print('\nAnswer:',Answer)
|
113 |
-
|
114 |
return most_similar_contexts
|
115 |
|
116 |
-
def chat_interface(question, history):
|
117 |
-
response = predict_test(model, tokenizer, embedding_model, context, question, index)
|
118 |
-
return response
|
119 |
-
|
120 |
-
examples=['ภูมิทัศน์สื่อไทยในปี 2567 มีแนวโน้มว่า ',
|
121 |
-
'Fragmentation คือ',
|
122 |
-
'ติ๊กต๊อก คือ',
|
123 |
-
'รายงานจาก Reuters Institute'
|
124 |
-
]
|
125 |
-
interface = gr.ChatInterface(fn=chat_interface,
|
126 |
-
examples=examples)
|
127 |
|
128 |
|
129 |
if __name__ == "__main__":
|
130 |
-
# Load your model, tokenizer, data, and index here...
|
131 |
-
# model, tokenizer = load_model('wangchanberta-hyp')
|
132 |
embedding_model = load_embedding_model()
|
133 |
-
# df = load_data()
|
134 |
index = set_index(prepare_sentences_vector(get_embeddings(embedding_model, context)))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
135 |
interface.launch()
|
|
|
94 |
question_vector = prepare_sentences_vector([question_vector])
|
95 |
distances, indices = faiss_search(index, question_vector, 3) # Retrieve top 3 indices
|
96 |
|
|
|
97 |
most_similar_contexts = ''
|
98 |
for i in range(3): # Loop through top 3 indices
|
99 |
most_sim_context = context[indices[0][i]].strip()
|
100 |
# most_similar_contexts.append(most_sim_context)
|
101 |
most_similar_contexts += str(i)+': '+most_sim_context + "\n\n"
|
102 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
103 |
return most_similar_contexts
|
104 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
105 |
|
106 |
|
107 |
if __name__ == "__main__":
|
|
|
|
|
108 |
embedding_model = load_embedding_model()
|
|
|
109 |
index = set_index(prepare_sentences_vector(get_embeddings(embedding_model, context)))
|
110 |
+
|
111 |
+
def chat_interface(question, history):
|
112 |
+
response = predict_test(model, tokenizer, embedding_model, context, question, index)
|
113 |
+
return response
|
114 |
+
|
115 |
+
examples=['ภูมิทัศน์สื่อไทยในปี 2567 มีแนวโน้มว่า ',
|
116 |
+
'Fragmentation คือ',
|
117 |
+
'ติ๊กต๊อก คือ',
|
118 |
+
'รายงานจาก Reuters Institute'
|
119 |
+
]
|
120 |
+
interface = gr.ChatInterface(fn=chat_interface,
|
121 |
+
examples=examples)
|
122 |
+
|
123 |
interface.launch()
|