Version 1.2 files

Added support for Metal/MPS and AMD GPU acceleration (via ROCm).

Provided detailed installation instructions.

Added "check.gpu.py" to test whether GPU-acceleration is installed correctly.

Author: BBC-Esq

check_gpu.py

@@ -0,0 +1,24 @@
+import torch
+
+if torch.cuda.is_available():
+    print("CUDA is available!")
+    print("CUDA version:", torch.version.cuda)
+else:
+    print("CUDA is not available.")
+
+print()
+
+if torch.backends.mps.is_available():
+    print("Metal/MPS is available!")
+else:
+    print("Metal/MPS is not available.")
+
+print("If you want to check the version of Metal and MPS on your macOS device, you can go to \"About This Mac\" -> \"System Report\" -> \"Graphics/Displays\" and look for information related to Metal and MPS.")
+
+print()
+
+if torch.version.hip is not None:
+    print("ROCm is available!")
+    print("ROCm version:", torch.version.hip)
+else:
+    print("ROCm is not available.")

gui_logic.py

@@ -32,9 +32,9 @@ def buttons(self):
 class DocQA_Logic:
     def __init__(self, gui: DocQA_GUI):
         self.gui = gui
-        self.embed_model_name = ""  # Store the selected embedding model name
+        self.embed_model_name = ""
 
-        # Connect the buttons to their respective actions
+        # Connecting the GUI buttons to their logic
         self.gui.download_embedding_model_button.config(command=self.download_embedding_model)
         self.gui.select_embedding_model_button.config(command=self.select_embedding_model_directory)
         self.gui.choose_documents_button.config(command=self.choose_documents)
@@ -48,27 +48,27 @@ def download_embedding_model(self):
 
         # Opening the dialog window
         dialog = DownloadModelDialog(self.gui.root)
-        selected_model = dialog.model_var.get()  # this gets the selected model's name
+        selected_model = dialog.model_var.get()
 
         if selected_model:
             # Construct the URL for the Hugging Face model repository
             model_url = f"https://huggingface.co/{selected_model}"
 
-            # Define the target directory for the download
+            # Define the directory to download the model to
             target_directory = os.path.join("Embedding_Models", selected_model.replace("/", "--"))
 
-            # Clone the repository using the subprocess module
+            # Clone the repository to the directory
             subprocess.run(["git", "clone", model_url, target_directory])
 
     def select_embedding_model_directory(self):
         initial_dir = 'Embedding_Models' if os.path.exists('Embedding_Models') else os.path.expanduser("~")
         chosen_directory = filedialog.askdirectory(initialdir=initial_dir, title="Select Embedding Model Directory")
 
-        # Store the chosen directory locally
+        # Choose the model directory to use
         if chosen_directory:
             self.embedding_model_directory = chosen_directory
 
-            # Also update the global variable in server_connector.py
+            # Update the global variable in server_connector.py
             server_connector.EMBEDDING_MODEL_NAME = chosen_directory
 
             # Optionally, you can print or display a confirmation to the user
@@ -85,13 +85,12 @@ def choose_documents(self):
 
             for file_path in file_paths:
                 shutil.copy(file_path, docs_folder)
-                # Add any additional logic to handle the selected files
 
     def create_chromadb(self):
         current_dir = os.path.dirname(os.path.realpath(__file__))
         vector_db_folder = os.path.join(current_dir, "Vector_DB")
 
-        # Check if the "Vector_DB" folder exists, and create it if not
+        # Create the "Vector_DB" folder if it doesn't exist
         if not os.path.exists(vector_db_folder):
             os.mkdir(vector_db_folder)
 

ingest_improved.py

@@ -77,7 +77,15 @@ def split_documents(documents: list[Document]) -> tuple[list[Document], list[Doc
     return documents, []  # We're only processing PDFs now, no more split based on extensions
 
 def main():
-    device_type = "cuda"  # Change to 'cpu' if needed
+# Determine the appropriate compute device
+    if torch.cuda.is_available() and torch.version.cuda:
+        device_type = "cuda"
+    elif torch.backends.mps.is_available():
+        device_type = "mps"
+    elif torch.cuda.is_available() and torch.version.hip:
+        device_type = "cuda"
+    else:
+        device_type = "cpu"
 
     logging.info(f"Loading documents from {SOURCE_DIRECTORY}")
     documents = load_documents(SOURCE_DIRECTORY)
@@ -92,9 +100,13 @@ def main():
         embeddings = HuggingFaceInstructEmbeddings(
             model_name=EMBEDDING_MODEL_NAME,
             model_kwargs={"device": device_type},
+            query_instruction="Represent the legal treatise for retrieval."
         )
     else:
-        embeddings = HuggingFaceEmbeddings(model_name=EMBEDDING_MODEL_NAME)
+        embeddings = HuggingFaceEmbeddings(
+            model_name=EMBEDDING_MODEL_NAME,
+            model_kwargs={"device": device_type},
+        )
 
     # Delete contents of the PERSIST_DIRECTORY before creating the vector database
     if os.path.exists(PERSIST_DIRECTORY):