支持fusemoe对int4的scale合zero合并读取操作

vllm/model_executor/model_loader/utils.py

@@ -103,7 +103,7 @@ def get_model_architecture(
         try:
             if os.getenv('AWQ_MOE_SZ') == None:
                 os.environ['AWQ_MOE_SZ'] = '1'
-            if os.getenv('AWQ_PAD') == None and (torch.cuda.get_device_properties(torch.cuda.current_device()).name == 'K100_AI') and (torch.cuda.get_device_properties(torch.cuda.current_device()).multi_processor_count == 120):
+            if os.getenv('AWQ_PAD') == None and (torch.cuda.get_device_properties(torch.cuda.current_device()).multi_processor_count == 120):
                 os.environ['AWQ_PAD'] = '1'
         except Exception as e:
             if os.getenv('AWQ_PAD') != '0':