Upload app.py

app.py

@@ -54,14 +54,24 @@ image_encoder = SigLIPImageEncoder(model_name=image_model_name, embed_dim=image_
 image_encoder.eval() # Set to evaluation mode
 
 # Load Phi-3 model using llama.cpp
-base_model = Llama(
-    model_path=phi3_model_path,
-    n_gpu_layers=0,  # Ensure no GPU usage
-    n_ctx=2048,       # Adjust context length as needed
-    verbose=True,
+#base_model = Llama(
+#    model_path=phi3_model_path,
+#    n_gpu_layers=0,  # Ensure no GPU usage
+#    n_ctx=2048,       # Adjust context length as needed
+#    verbose=True,
+#)
+
+
+base_model = Llama.from_pretrained(
+	repo_id="QuantFactory/Phi-3-mini-4k-instruct-GGUF",
+	filename="Phi-3-mini-4k-instruct.Q2_K.gguf",
+    n_gpu_layers=0,
+    n_ctx=2048,
+    verbose=True 
 )
 
 
+# Load and merge
 model = PeftModel.from_pretrained(base_model, peft_model_path, offload_dir='./offload')
 model = model.merge_and_unload()
 print("phi-3 model loaded sucessfully")