Update app.py

app.py

@@ -5,7 +5,7 @@ import os
 from huggingface_hub import hf_hub_download
 import base64
 from llama_cpp import Llama
-from llama_cpp.llama_chat_format import Llava16ChatHandler
+from llama_cpp.llama_chat_format import NanoLlavaChatHandler
 
 
 os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
@@ -27,18 +27,23 @@ CSS = """
 }
 """
 
-# chat_handler = Llava16ChatHandler.from_pretrained(
-#   repo_id="openbmb/MiniCPM-Llama3-V-2_5-gguf",
-#   filename="*mmproj*",
-# )
+chat_handler = NanoLlavaChatHandler.from_pretrained(
+    repo_id="openbmb/MiniCPM-Llama3-V-2_5-gguf",
+    filename="*mmproj*",
+)
 
 llm = Llama.from_pretrained(
   repo_id="openbmb/MiniCPM-Llama3-V-2_5-gguf",
   filename="ggml-model-Q5_K_M.gguf",
-# chat_handler=chat_handler,
+  chat_handler=chat_handler,
   n_ctx=2048, # n_ctx should be increased to accommodate the image embedding
 )
 
+def image_to_base64_data_uri(file_path):
+    with open(file_path, "rb") as img_file:
+        base64_data = base64.b64encode(img_file.read()).decode('utf-8')
+        return f"data:image/png;base64,{base64_data}"
+
 @spaces.GPU(queue=False)
 def stream_chat(message, history: list, temperature: float, max_new_tokens: int):
     print(f'message is - {message}')
@@ -46,7 +51,7 @@ def stream_chat(message, history: list, temperature: float, max_new_tokens: int)
     messages = []
 
     if message["files"]:
-        image = Image.open(message["files"][-1]).convert('RGB')
+        image = image_to_base64_data_uri(message["files"][-1])
         messages.append({
             "role": "user", 
             "content": [
@@ -59,7 +64,7 @@ def stream_chat(message, history: list, temperature: float, max_new_tokens: int)
             raise gr.Error("Please upload an image first.")
             image = None
         else:
-            image = Image.open(history[0][0][0])
+            image = image_to_base64_data_uri(history[0][0][0])
             for prompt, answer in history:
                 if answer is None:
                     messages.extend([{