diff --git a/bert/model.py b/bert/model.py
index 11a24659..f1e05313 100644
--- a/bert/model.py
+++ b/bert/model.py
@@ -1,7 +1,7 @@
 import argparse
 from dataclasses import dataclass
 from pathlib import Path
-from typing import List, Optional
+from typing import List, Optional, Tuple
 
 import mlx.core as mx
 import mlx.nn as nn
@@ -120,7 +120,7 @@ class Bert(nn.Module):
         input_ids: mx.array,
         token_type_ids: mx.array,
         attention_mask: mx.array = None,
-    ) -> tuple[mx.array, mx.array]:
+    ) -> Tuple[mx.array, mx.array]:
         x = self.embeddings(input_ids, token_type_ids)
 
         if attention_mask is not None:
@@ -132,7 +132,7 @@ class Bert(nn.Module):
         return y, mx.tanh(self.pooler(y[:, 0]))
 
 
-def load_model(bert_model: str, weights_path: str) -> tuple[Bert, BertTokenizer]:
+def load_model(bert_model: str, weights_path: str) -> Tuple[Bert, BertTokenizer]:
     if not Path(weights_path).exists():
         raise ValueError(f"No model weights found in {weights_path}")
 
diff --git a/llms/mlx_lm/models/phi2.py b/llms/mlx_lm/models/phi2.py
index 51b5e390..49c91706 100644
--- a/llms/mlx_lm/models/phi2.py
+++ b/llms/mlx_lm/models/phi2.py
@@ -1,5 +1,6 @@
 import math
 from dataclasses import dataclass
+from typing import Tuple
 
 import mlx.core as mx
 import mlx.nn as nn
@@ -128,7 +129,7 @@ class Model(nn.Module):
         x: mx.array,
         mask: mx.array = None,
         cache: mx.array = None,
-    ) -> tuple[mx.array, mx.array]:
+    ) -> Tuple[mx.array, mx.array]:
         mask = None
         if x.shape[1] > 1:
             mask = nn.MultiHeadAttention.create_additive_causal_mask(x.shape[1])
diff --git a/llms/setup.py b/llms/setup.py
index 3b3fdb7e..78839b2b 100644
--- a/llms/setup.py
+++ b/llms/setup.py
@@ -8,7 +8,7 @@ with open(Path(__file__).parent / "mlx_lm/requirements.txt") as fid:
     requirements = [str(r) for r in pkg_resources.parse_requirements(fid)]
 setup(
     name="mlx-lm",
-    version="0.0.1",
+    version="0.0.2",
     description="LLMs on Apple silicon with MLX and the Hugging Face Hub",
     long_description=open("README.md", encoding="utf-8").read(),
     long_description_content_type="text/markdown",