Update app.py

app.py

@@ -27,10 +27,10 @@ os.makedirs(OUTPUT_DIR, exist_ok=True)
 
 # --- LLAMA.CPP SETUP ---
 LLAMA_CPP_DIR = Path("llama.cpp")
-LLAMA_CPP_CONVERT_SCRIPT = LLAMA_CPP_DIR / "convert.py"
-LLAMA_CPP_QUANTIZE_SCRIPT = LLAMA_CPP_DIR / "quantize" # This is a compiled binary
+## FIX: The conversion script was renamed in the llama.cpp repository.
+LLAMA_CPP_CONVERT_SCRIPT = LLAMA_CPP_DIR / "convert-hf-to-gguf.py" 
+LLAMA_CPP_QUANTIZE_SCRIPT = LLAMA_CPP_DIR / "quantize"
 
-## FIXED FUNCTION: Replaced 'make' with 'cmake' for the build process.
 def setup_llama_cpp():
     """Clones and builds llama.cpp if not already present."""
     if not LLAMA_CPP_DIR.exists():
@@ -43,22 +43,11 @@ def setup_llama_cpp():
             logging.error(error_msg, exc_info=True)
             raise RuntimeError(error_msg)
 
-    # If the binary doesn't exist, try to build it with CMake.
     if not LLAMA_CPP_QUANTIZE_SCRIPT.exists():
         logging.info("llama.cpp 'quantize' binary not found. Building with CMake...")
         try:
-            # Step 1: Configure the build directory
-            subprocess.run(
-                ["cmake", "."],
-                cwd=str(LLAMA_CPP_DIR),  # Run command inside the llama.cpp directory
-                check=True, capture_output=True, text=True
-            )
-            # Step 2: Build the 'quantize' target
-            subprocess.run(
-                ["cmake", "--build", ".", "--target", "quantize"],
-                cwd=str(LLAMA_CPP_DIR),
-                check=True, capture_output=True, text=True
-            )
+            subprocess.run(["cmake", "."], cwd=str(LLAMA_CPP_DIR), check=True, capture_output=True, text=True)
+            subprocess.run(["cmake", "--build", ".", "--target", "quantize"], cwd=str(LLAMA_CPP_DIR), check=True, capture_output=True, text=True)
             logging.info("'quantize' binary built successfully with CMake.")
         except subprocess.CalledProcessError as e:
             error_msg = f"Failed to build llama.cpp with CMake. Error: {e.stderr}"
@@ -144,7 +133,7 @@ def stage_3_4_gguf_quantize(model_path_or_id: str, original_model_id: str, quant
     quantized_gguf_path = os.path.join(gguf_path, "model.gguf")
 
     try:
-        log_stream += f"Executing llama.cpp convert.py script on '{model_path_or_id}'...\n"
+        log_stream += f"Executing llama.cpp conversion script on '{model_path_or_id}'...\n"
         convert_command = ["python3", str(LLAMA_CPP_CONVERT_SCRIPT), model_path_or_id, "--outfile", f16_gguf_path, "--outtype", "f16"]
         process = subprocess.run(convert_command, check=True, capture_output=True, text=True)
         log_stream += process.stdout