Update app.py

app.py

@@ -10,21 +10,26 @@ client = InferenceClient(model)
 # Embedded system prompt
 system_prompt_text = "You are a smart and helpful co-worker of Thailand based multi-national company PTT, and PTTEP. You help with any kind of request and provide a detailed answer to the question. But if you are asked about something unethical or dangerous, you must refuse and provide a safe and respectful way to handle that."
 
-# Read the content of the info.md file
+# Read the content of the info.md and info2.md files
 with open("info.md", "r") as file:
     info_md_content = file.read()
 
-# Chunk the info.md content into smaller sections
+with open("info2.md", "r") as file:
+    info2_md_content = file.read()
+
+# Chunk the info.md and info2.md content into smaller sections
 chunk_size = 2500  # Adjust this size as needed
 info_md_chunks = textwrap.wrap(info_md_content, chunk_size)
+info2_md_chunks = textwrap.wrap(info2_md_content, chunk_size)
 
 def get_all_chunks(chunks):
     return "\n\n".join(chunks)
 
-def format_prompt_mixtral(message, history, info_md_chunks):
+def format_prompt_mixtral(message, history, info_md_chunks, info2_md_chunks):
     prompt = "<s>"
-    all_chunks = get_all_chunks(info_md_chunks)
-    prompt += f"{all_chunks}\n\n"  # Add all chunks of info.md at the beginning
+    # Sequentially add chunks from both files
+    all_chunks = get_all_chunks(info_md_chunks + info2_md_chunks)
+    prompt += f"{all_chunks}\n\n"  # Add all chunks at the beginning
     prompt += f"{system_prompt_text}\n\n"  # Add the system prompt
 
     if history:
@@ -44,7 +49,7 @@ def chat_inf(prompt, history, seed, temp, tokens, top_p, rep_p):
         seed=seed,
     )
 
-    formatted_prompt = format_prompt_mixtral(prompt, history, info_md_chunks)
+    formatted_prompt = format_prompt_mixtral(prompt, history, info_md_chunks, info2_md_chunks)
     stream = client.text_generation(formatted_prompt, **generate_kwargs, stream=True, details=True, return_full_text=False)
     output = ""
     for response in stream:
@@ -96,9 +101,3 @@ with gr.Blocks() as app:  # Add auth here
     clear_btn.click(clear_fn, None, [inp, chat])
 
 app.queue(default_concurrency_limit=10).launch(share=True, auth=("admin", "0112358"))
-
-
-
-I have 2000 lines in info.md file, and the model throws error due to character limit.
-Even though I divide chunks, I added all together which is a bad choice.
-what can I do?