Update app.py

app.py

@@ -1,7 +1,56 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
+# SEMANTIC SEARCH STEP 1
+from sentence_transformers import SentenceTransformer
+import torch
 
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+# SEMANTIC SEARCH STEP 2 --> EDIT WITH YOUR OWN KNOWLEDGEBASE WHEN READY
+with open("water_cycle.txt", "r", encoding="utf-8") as file:
+  water_cycle_text = file.read()
+print(water_cycle_text)
+
+# SEMANTIC SEARCH STEP 3
+def preprocess_text(text):
+  cleaned_text = text.strip()
+  chunks = cleaned_text.split("\n")
+  cleaned_chunks = []
+  for chunk in chunks:
+    stripped_chunk = chunk.strip()
+    cleaned_chunks.append(stripped_chunk)
+  print(cleaned_chunks)
+  print(len(cleaned_chunks))
+  return cleaned_chunks
+
+cleaned_chunks = preprocess_text(water_cycle_text) # edit this with my knowledgebase when ready
+
+# SEMANTIC SEARCH STEP 4
+model = SentenceTransformer('all-MiniLM-L6-v2')
+
+def create_embeddings(text_chunks):
+  chunk_embeddings = model.encode(text_chunks, convert_to_tensor=True) # Replace ... with the text_chunks list
+  print(chunk_embeddings)
+  print(chunk_embeddings.shape)
+  return chunk_embeddings
+
+chunk_embeddings = create_embeddings(cleaned_chunks)
+
+# SEMANTIC SEARCH STEP 5
+def get_top_chunks(query, chunk_embeddings, text_chunks):
+  query_embedding = model.encode(query, convert_to_tensor=True) # Complete this line
+  query_embedding_normalized = query_embedding / query_embedding.norm()
+  chunk_embeddings_normalized = chunk_embeddings / chunk_embeddings.norm(dim=1, keepdim=True)
+  similarities = torch.matmul(chunk_embeddings_normalized, query_embedding_normalized) # Complete this line
+  print(similarities)
+  top_indices = torch.topk(similarities, k=3).indices
+  print(top_indices)
+  top_chunks = []
+  for i in top_indices:
+    relevant_info = text_chunks[i]
+    top_chunks.append(relevant_info)
+
+  return top_chunks
+
+client = InferenceClient("microsoft/phi-4")
 
 def respond(message, history):