diff --git a/requirements-dev.txt b/requirements-dev.txt
index 84b5700d3..f02124df1 100644
--- a/requirements-dev.txt
+++ b/requirements-dev.txt
@@ -1,4 +1,4 @@
-accelerate>=0.11.0
+accelerate>=0.22.0
 fire
 flaky
 future>=0.17.1
diff --git a/skorch/tests/test_hf.py b/skorch/tests/test_hf.py
index e41c30496..63f0c2476 100644
--- a/skorch/tests/test_hf.py
+++ b/skorch/tests/test_hf.py
@@ -660,10 +660,32 @@ def test_mixed_precision_save_load_params(
         accelerator = accelerator_cls(mixed_precision=mixed_precision)
         net = net_cls(accelerator=accelerator)
         net.initialize()
+
         filename = tmp_path / 'accel-net-params.pth'
         net.save_params(f_params=filename)
         net.load_params(f_params=filename)
 
+    @pytest.mark.parametrize('mixed_precision', ['fp16', 'bf16', 'no'])
+    def test_mixed_precision_inference(
+            self, net_cls, accelerator_cls, data, mixed_precision, tmp_path
+    ):
+        from accelerate.utils import is_bf16_available
+
+        if (mixed_precision != 'no') and not torch.cuda.is_available():
+            pytest.skip('skipping AMP test because device does not support it')
+        if (mixed_precision == 'bf16') and not is_bf16_available():
+            pytest.skip('skipping bf16 test because device does not support it')
+
+        X, y = data
+        accelerator = accelerator_cls(mixed_precision=mixed_precision)
+        net = net_cls(accelerator=accelerator)
+        net.fit(X, y)
+        net.predict(X)
+        net.predict_proba(X)
+
+        Xt = torch.from_numpy(X).to(net.device)
+        net.forward(Xt)
+
     def test_force_cpu(self, net_cls, accelerator_cls, data):
         accelerator = accelerator_cls(device_placement=False, cpu=True)
         net = net_cls(accelerator=accelerator)