diff --git a/.gitignore b/.gitignore
index 1b7f0fb..d98b81d 100644
--- a/.gitignore
+++ b/.gitignore
@@ -9,6 +9,8 @@ torch-dumps/*
 *pycache*
 */*pycache*
 */*/pycache*
+venv/
+.venv/
 
 settings.json
 img_bot*
diff --git a/modules/models.py b/modules/models.py
index 63060d4..f07e738 100644
--- a/modules/models.py
+++ b/modules/models.py
@@ -47,7 +47,12 @@ def load_model(model_name):
         if any(size in shared.model_name.lower() for size in ('13b', '20b', '30b')):
             model = AutoModelForCausalLM.from_pretrained(Path(f"models/{shared.model_name}"), device_map='auto', load_in_8bit=True)
         else:
-            model = AutoModelForCausalLM.from_pretrained(Path(f"models/{shared.model_name}"), low_cpu_mem_usage=True, torch_dtype=torch.bfloat16 if shared.args.bf16 else torch.float16).cuda()
+            model = AutoModelForCausalLM.from_pretrained(Path(f"models/{shared.model_name}"), low_cpu_mem_usage=True, torch_dtype=torch.bfloat16 if shared.args.bf16 else torch.float16)
+            if torch.has_mps:
+                device = torch.device('mps')
+                model = model.to(device)
+            else:
+                model = model.cuda()
 
     # FlexGen
     elif shared.args.flexgen:
@@ -97,7 +102,7 @@ def load_model(model_name):
     # Custom
     else:
         params = {"low_cpu_mem_usage": True}
-        if not shared.args.cpu and not torch.cuda.is_available():
+        if not any((shared.args.cpu, torch.cuda.is_available(), torch.has_mps)):
             print("Warning: torch.cuda.is_available() returned False.\nThis means that no GPU has been detected.\nFalling back to CPU mode.\n")
             shared.args.cpu = True
 
diff --git a/modules/text_generation.py b/modules/text_generation.py
index e5b4ad9..1d11de1 100644
--- a/modules/text_generation.py
+++ b/modules/text_generation.py
@@ -33,6 +33,9 @@ def encode(prompt, tokens_to_generate=0, add_special_tokens=True):
             return input_ids.numpy()
         elif shared.args.deepspeed:
             return input_ids.to(device=local_rank)
+        elif torch.has_mps:
+            device = torch.device('mps')
+            return input_ids.to(device)
         else:
             return input_ids.cuda()