fixes

app.py

@@ -18,7 +18,7 @@ from torchao.quantization import (
 
 MAP_QUANT_TYPE_TO_NAME = {
     "Int4WeightOnly": "int4wo",
-    "GemliteUIntXWeightOnly": "intxwo-gemlite"
+    "GemliteUIntXWeightOnly": "intxwo-gemlite",
     "Int8WeightOnly": "int8wo",
     "Int8DynamicActivationInt8Weight": "int8da8w8",
     "Float8WeightOnly": "float8wo",
@@ -61,14 +61,15 @@ def check_model_exists(
             if (
                 quantization_type in ["Int4WeightOnly", "GemliteUIntXWeightOnly"]
             ) and (group_size is not None):
-                repo_name = f"{username}/{model_name.split('/')[-1]}-ao-{MAP_QUANT_TYPE_TO_NAME[quantization_type.lower()]}-gs{group_size}"
+                repo_name = f"{username}/{model_name.split('/')[-1]}-ao-{MAP_QUANT_TYPE_TO_NAME[quantization_type]}-gs{group_size}"
             else:
-                repo_name = f"{username}/{model_name.split('/')[-1]}-ao-{MAP_QUANT_TYPE_TO_NAME[quantization_type.lower()]}"
+                repo_name = f"{username}/{model_name.split('/')[-1]}-ao-{MAP_QUANT_TYPE_TO_NAME[quantization_type]}"
         if repo_name in model_names:
             return f"Model '{repo_name}' already exists in your repository."
         else:
             return None  # Model does not exist
     except Exception as e:
+        # raise e
         return f"Error checking model existence: {str(e)}"
 
 
@@ -237,9 +238,9 @@ def save_model(
             if (
                 quantization_type in ["Int4WeightOnly", "GemliteUIntXWeightOnly"]
             ) and (group_size is not None):
-                repo_name = f"{username}/{model_name.split('/')[-1]}-ao-{MAP_QUANT_TYPE_TO_NAME[quantization_type.lower()]}-gs{group_size}"
+                repo_name = f"{username}/{model_name.split('/')[-1]}-ao-{MAP_QUANT_TYPE_TO_NAME[quantization_type]}-gs{group_size}"
             else:
-                repo_name = f"{username}/{model_name.split('/')[-1]}-ao-{MAP_QUANT_TYPE_TO_NAME[quantization_type.lower()]}"
+                repo_name = f"{username}/{model_name.split('/')[-1]}-ao-{MAP_QUANT_TYPE_TO_NAME[quantization_type]}"
         progress(0.70, desc="Creating model card")
         model_card = create_model_card(model_name, quantization_type, group_size)
         with open(os.path.join(tmpdirname, "README.md"), "w") as f:
@@ -494,14 +495,14 @@ with gr.Blocks(css=css) as demo:
                         info="Select the Quantization method",
                         choices=[
                             "Int4WeightOnly",
-                            "GemliteUIntXWeightOnly"
+                            "GemliteUIntXWeightOnly",
                             "Int8WeightOnly",
                             "Int8DynamicActivationInt8Weight",
                             "Float8WeightOnly",
                             "Float8DynamicActivationFloat8Weight",
                             "autoquant",
                         ],
-                        value="int8_weight_only",
+                        value="Int8WeightOnly",
                         filterable=False,
                         show_label=False,
                     )
@@ -509,7 +510,7 @@ with gr.Blocks(css=css) as demo:
                     group_size = gr.Textbox(
                         info="Group Size (only for int4_weight_only and int8_weight_only)",
                         value="128",
-                        interactive=(quantization_type.value == "int4_weight_only" or quantization_type.value == "int8_weight_only"),
+                        interactive=(quantization_type.value == "Int4WeightOnly" or quantization_type.value == "Int8WeightOnly"),
                         show_label=False,
                     )