Update app.py

app.py

@@ -7,7 +7,7 @@ def get_hf_model_args(hf_model_name_or_path):
     try:
         config = AutoConfig.from_pretrained(hf_model_name_or_path, trust_remote_code=True).to_dict()
     except Exception as e:
-        return None, f"Error fetching Hugging Face model: {str(e)}"
+        raise gr.Error(f"Error fetching Hugging Face model: {str(e)}")
     
     # Extract relevant values from the config
     num_layers = config.get("num_hidden_layers", None)
@@ -22,14 +22,11 @@ def get_hf_model_args(hf_model_name_or_path):
         "num_attention_heads": num_attention_heads,
         "vocab_size": vocab_size,
         "sequence_length": sequence_length,
-    }, None
+    }
 
 # ---- Update Gradio inputs with Hugging Face model config ---- #
 def update_from_hf_model(hf_model_name_or_path):
-    model_params, hf_error = get_hf_model_args(hf_model_name_or_path)
-    
-    if hf_error:
-        return gr.update(), gr.update(), gr.update(), gr.update(), gr.update(), hf_error
+    model_params = get_hf_model_args(hf_model_name_or_path)
     
     return (gr.update(value=model_params["num_layers"]), 
             gr.update(value=model_params["hidden_size"]),
@@ -40,10 +37,7 @@ def update_from_hf_model(hf_model_name_or_path):
 
 # ---- Memory Calculation ---- #
 def calc_mem(hf_model_name_or_path, num_gpus, tensor_parallel_size, pipeline_parallel_size, batch_size_per_gpu, sequence_length, vocab_size, hidden_size, num_attention_heads, num_layers, ffn_expansion_factor, is_mixed_precision, misc_mem_gib):
-    model_params, hf_error = get_hf_model_args(hf_model_name_or_path) if hf_model_name_or_path else (None, None)
-
-    if hf_error:
-        return hf_error
+    model_params = get_hf_model_args(hf_model_name_or_path) if hf_model_name_or_path else None
     
     if model_params:
         num_layers = model_params["num_layers"] or num_layers