Update app.py

app.py

@@ -1,11 +1,65 @@
 import gradio as gr
 import random
 from huggingface_hub import InferenceClient
+from sentence_transformers import SentenceTransformer
+import torch
 
+with open("knowledge.txt", "r", encoding="utf-8") as file:
+  recent = file.read()
+  # opens the text, saves as "file"
+  # reads the text and saves as water_cycle_text variable
+
+print(recent)
+
+cleaned_text = recent.strip()
+# cleaning up the text
+chunks = cleaned_text.split("\n")
+# seperating the text into one sentence pieces
+cleaned_chunks = []
+# creating an empty list to put the cleaned chunks in
+
+for chunk in chunks:
+  stripped_chunk = chunk.strip()
+  if stripped_chunk:
+    cleaned_chunks.append(stripped_chunk)
+    # loop through chunks and add not empty chunks to cleaned_chunks list
+print(cleaned_chunks)
+
+model = SentenceTransformer('all-MiniLM-L6-v2')
+
+chunk_embeddings = model.encode(cleaned_chunks, convert_to_tensor=True)
+# encode the model, pass through my cleaned chunks and convert to vector embeddings (not arrays)
+print(chunk_embeddings)
+
+def get_top_chunks(query):
+  # create my function taking query as parameter
+  query_embedding = model.encode(query, convert_to_tensor=True)
+  # encode query to vector embedding for comparison
+  query_embedding_normalized = query_embedding / query_embedding.norm()
+  # normalize my query to 1; allows for comparison of meaning
+  chunk_embeddings_normalized = chunk_embeddings / chunk_embeddings.norm(dim=1, keepdim=True)
+  # normailizing chunks for comparison of meaning
+
+  similarities = torch.matmul(chunk_embeddings_normalized, query_embedding_normalized)
+  print(similarities)
+  # using matmul (matrix multiplication method) to compare query to chunks
+  top_indices = torch.topk(similarities, k=3).indices
+  print(top_indices)
+  # get the indices of the chunks that are most similar to query
+
+  top_chunks = []
+
+  for i in top_indices:
+    chunk = chunks[i]
+    # for each index number in top_indices, get back the text
+    top_chunks.append(chunk)
+    # values of each index number is added to top_chunks
+  return top_chunks
+    
 client = InferenceClient('Qwen/Qwen2.5-72B-Instruct')
 #client is where you can change the LLM model!
 def respond(message,history):
-    messages = [{'role': 'system', 'content': 'You give really good gift ideas and are super helpful! You also tell me the price of each item. Give me 5 gift ideas if I ask'}]
+    messages = [{'role': 'system', 'content': 'You give really good gift ideas and are super helpful! You also tell me the price of each item. Give me 5 gift ideas if I ask}]
     if history:
         messages.extend(history)