Make universal tokenizer, xformers, sdp-attention apply to monkey patch

modules/models.py

@@ -149,8 +149,7 @@ def load_model(model_name):
             print("Warning: applying the monkey patch for using LoRAs in 4-bit mode.\nIt may cause undefined behavior outside its intended scope.")
             from modules.monkey_patch_gptq_lora import load_model_llama
 
-            model, tokenizer = load_model_llama(model_name)
-            return model, tokenizer
+            model, _ = load_model_llama(model_name)
 
         # No monkey patch
         else:

modules/monkey_patch_gptq_lora.py

@@ -18,17 +18,16 @@ from modules.GPTQ_loader import find_quantized_model_file
 replace_peft_model_with_gptq_lora_model()
 
 def load_model_llama(model_name):
-
     config_path = str(Path(f'{shared.args.model_dir}/{model_name}'))
     model_path = str(find_quantized_model_file(model_name))
     model, tokenizer = load_llama_model_4bit_low_ram(config_path, model_path, groupsize=shared.args.groupsize, is_v1_model=False)
-
     for n, m in model.named_modules():
         if isinstance(m, Autograd4bitQuantLinear) or isinstance(m, Linear4bitLt):
             if m.is_v1_model:
                 m.zeros = m.zeros.half()
             m.scales = m.scales.half()
             m.bias = m.bias.half()
+
     autograd_4bit.use_new = True
     autograd_4bit.auto_switch = True
 
@@ -36,11 +35,4 @@ def load_model_llama(model_name):
     wrapper = AMPWrapper(model)
     wrapper.apply_generate()
 
-    try:
-        tokenizer.eos_token_id = 2
-        tokenizer.bos_token_id = 1
-        tokenizer.pad_token_id = 0
-    except:
-        pass
-
     return model, tokenizer