diff --git a/.github/workflows/pytest.yaml b/.github/workflows/pytest.yaml
new file mode 100644
index 0000000..15a98e7
--- /dev/null
+++ b/.github/workflows/pytest.yaml
@@ -0,0 +1,24 @@
+name: pytest
+
+on: [push]
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        python-version: ["3.10", "3.11", "3.12"]
+    steps:
+    - uses: actions/checkout@v3
+    - name: Set up Python ${{ matrix.python-version }}
+      uses: actions/setup-python@v3
+      with:
+        python-version: ${{ matrix.python-version }}
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade uv pytest pytest-cov
+        uv pip install -e . --system
+        uv pip install -e ".[dev]" --system
+    - name: Analysing the code with pytest
+      run: |
+        pytest -vvvv  --cov=shadow-peft tests/
diff --git a/tests/test_shadow_peft.py b/tests/test_shadow_peft.py
index ffae772..7ae5ece 100644
--- a/tests/test_shadow_peft.py
+++ b/tests/test_shadow_peft.py
@@ -87,6 +87,7 @@ def test_save_load_roundtrip_matches_outputs(tmp_path: Path):
     torch.manual_seed(0)
     base1 = _tiny_llama(num_layers=4)
     base1.eval()
+    base1_state = {k: v.detach().clone() for k, v in base1.state_dict().items()}
 
     cfg = ShadowConfig(
         num_shadow_layers=1,
@@ -107,7 +108,7 @@ def test_save_load_roundtrip_matches_outputs(tmp_path: Path):
 
     # Create a new base with identical weights.
     base2 = _tiny_llama(num_layers=4)
-    base2.load_state_dict(base1.state_dict())
+    base2.load_state_dict(base1_state)
     base2.eval()
 
     m2 = ShadowPeftModel.from_pretrained(base2, save_dir, is_trainable=False)
@@ -217,5 +218,3 @@ def test_implicit_shadow_model_uses_shadow_intermediate_size():
     )
     m = get_shadow_model(base, cfg)
     assert m.shadow_model.config.intermediate_size == 12
-
-