Update app.py

app.py

@@ -1,24 +1,27 @@
-import os
+from huggingface_hub import hf_hub_list, hf_hub_download
 from transformers import AutoModel, AutoTokenizer
 
-# Directory containing GGUF files
-gguf_directory = "/path/to/gguf/files"  # Replace with your actual directory path
+# Hugging Face model repository
+model_repo = "mradermacher/Qwen2-2B-RepleteCoder-DHM-GGUF"
 
-# Filter files containing 'Q2_K'
-def find_q2_k_models(directory):
-    return [file for file in os.listdir(directory) if "Q2_K" in file and file.endswith(".gguf")]
-
-# Load the appropriate model
-def load_q2_k_model(gguf_directory):
-    q2_k_files = find_q2_k_models(gguf_directory)
+# Find files containing 'Q2_K'
+def find_q2_k_files(repo_id):
+    files = hf_hub_list(repo_id)
+    q2_k_files = [file for file in files if "Q2_K" in file.filename and file.filename.endswith(".gguf")]
     if not q2_k_files:
-        raise ValueError("No GGUF files with 'Q2_K' found in the specified directory.")
-    # Load the first matching file (or implement logic for user selection)
-    model_path = os.path.join(gguf_directory, q2_k_files[0])
-    print(f"Loading model from: {model_path}")
+        raise ValueError(f"No files containing 'Q2_K' found in the repository {repo_id}.")
+    return q2_k_files
+
+# Download and load the desired model
+def load_q2_k_model(repo_id):
+    q2_k_files = find_q2_k_files(repo_id)
+    # For simplicity, load the first matching file
+    chosen_file = q2_k_files[0]
+    print(f"Loading model from: {chosen_file.filename}")
+    model_path = hf_hub_download(repo_id, chosen_file.filename)
     model = AutoModel.from_pretrained(model_path)
     tokenizer = AutoTokenizer.from_pretrained(model_path)
     return model, tokenizer
 
 # Initialize the model
-model, tokenizer = load_q2_k_model(gguf_directory)
+model, tokenizer = load_q2_k_model(model_repo)