Update app.py

app.py

@@ -5,7 +5,8 @@ import os
 from huggingface_hub import hf_hub_download
 import base64
 from llama_cpp import Llama
-from llama_cpp.llama_tokenizer import LlamaHFTokenizer
+# from llama_cpp.llama_chat_format import Llava15ChatHandler
+# from llama_cpp.llama_tokenizer import LlamaHFTokenizer
 
 os.environ["HF_HUB_ENABLE_HF_TRANSFER"] = "1"
 MODEL_LIST = ["openbmb/MiniCPM-Llama3-V-2_5","openbmb/MiniCPM-Llama3-V-2_5-int4"]
@@ -26,19 +27,34 @@ CSS = """
 }
 """
 
-tokenizer = LlamaHFTokenizer.from_pretrained(
-    repo_id="openbmb/MiniCPM-Llama3-V-2_5-gguf",
-    filename="*mmproj*",
-)
+# chat_handler = Llava15ChatHandler.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",
-  tokenizer=tokenizer,
+  filename=["ggml-model-Q5_K_M.gguf","*mmproj*"],
   n_ctx=4096,
   verbose=True
 )
 
+'''
+filenames = [
+    "*mmproj*",
+    "ggml-model-Q5_K_M.gguf"
+]
+
+for filename in filenames:
+    downloaded_model_path = hf_hub_download(
+        repo_id="openbmb/MiniCPM-Llama3-V-2_5-gguf",
+        filename=filename,
+        local_dir="model"
+    )
+'''
+
+
 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')