Update app.py

app.py

@@ -25,23 +25,24 @@ info2_md_chunks = textwrap.wrap(info2_md_content, chunk_size)
 # Combine both sets of chunks
 all_chunks = info_md_chunks + info2_md_chunks
 
-# Initialize history with these chunks
-history = []
-for chunk in all_chunks:
-    history.append(("System Information", chunk))
-
-def format_prompt_mixtral(message, history):
+def format_prompt_mixtral(message, history, chunks):
     prompt = "<s>"
     prompt += f"{system_prompt_text}\n\n"  # Add the system prompt
 
+    # Include the initial context from the chunks
+    for chunk in chunks:
+        prompt += f"[INST] System Information [/INST] {chunk}</s> "
+
+    # Add conversation history
     if history:
         for user_prompt, bot_response in history:
-            prompt += f"[INST] {user_prompt} [/INST]"
-            prompt += f" {bot_response}</s> "
+            prompt += f"[INST] {user_prompt} [/INST] {bot_response}</s> "
+    
+    # Add the current user message
     prompt += f"[INST] {message} [/INST]"
     return prompt
 
-def chat_inf(prompt, history, seed, temp, tokens, top_p, rep_p):
+def chat_inf(message, history, seed, temp, tokens, top_p, rep_p):
     generate_kwargs = dict(
         temperature=temp,
         max_new_tokens=tokens,
@@ -51,13 +52,13 @@ def chat_inf(prompt, history, seed, temp, tokens, top_p, rep_p):
         seed=seed,
     )
 
-    formatted_prompt = format_prompt_mixtral(prompt, history)
+    formatted_prompt = format_prompt_mixtral(message, history, all_chunks)
     stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
     output = ""
     for response in stream:
         output += response.token.text
-        yield [(prompt, output)]
-    history.append((prompt, output))
+        yield [(message, output)]
+    history.append((message, output))
     yield history
 
 def clear_fn():