init application

app.py

@@ -0,0 +1,48 @@
+from sentence_transformers import SentenceTransformer, util
+import os
+from tqdm import tqdm
+import pandas as pd
+import json
+import pickle
+import torch
+import gradio as gr
+
+
+with open('new_transcript.json', 'r', encoding='utf-8') as openfile:
+
+    # Reading from json file
+    json_object1 = json.load(openfile)
+
+json_object1[0]
+
+model = SentenceTransformer('keepitreal/vietnamese-sbert', device='cpu')
+
+#Load sentences & embeddings from disc
+with open('embeddings.pkl', "rb") as fIn:
+    stored_data = pickle.load(fIn)
+    stored_sentences = stored_data['sentences']
+    stored_embeddings = stored_data['embeddings']
+
+emb = torch.from_numpy(stored_embeddings)
+
+
+def semantic_search(query, top_k=20):
+  query_embedding = model.encode(query, convert_to_tensor=True)
+
+  # We use cosine-similarity and torch.topk to find the highest 5 scores
+  cos_scores = util.cos_sim(query_embedding, emb)[0]
+  top_results = torch.topk(cos_scores, k=top_k)
+
+  str_results = ""
+  for score, idx in zip(top_results[0], top_results[1]):
+    str_results += str(json_object1[idx]) + " - (Score: {:.4f})".format(score) + "\n"
+
+  return str_results
+
+
+demo = gr.Interface(
+    fn=semantic_search,
+    inputs=gr.Textbox(lines=2, placeholder="Input text query..."),
+    outputs="text",
+)
+demo.launch(share=True)

embeddings.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8f424f4016bb2d019a4e2bc611cea027eb722212b18f7350614adfbaf89c687
+size 143095812

requirements.txt

@@ -0,0 +1,5 @@
+sentence-transformers==2.2.2
+torch==2.0.1+cu118
+tqdm==4.66.1
+gradio==3.46.1
+pandas==1.5.3