diff --git a/embeddings.py b/embeddings.py
index 111ba04..618f478 100644
--- a/embeddings.py
+++ b/embeddings.py
@@ -67,8 +67,9 @@ class CogVideoXPatchEmbed(nn.Module):
             post_time_compression_frames,
             self.spatial_interpolation_scale,
             self.temporal_interpolation_scale,
+            output_type="pt",
         )
-        pos_embedding = torch.from_numpy(pos_embedding).flatten(0, 1)
+        pos_embedding = pos_embedding.flatten(0, 1)
         joint_pos_embedding = torch.zeros(
             1, self.max_text_seq_length + num_patches, self.embed_dim, requires_grad=False
         )
diff --git a/requirements.txt b/requirements.txt
index 8ab8109..287e442 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,5 +1,5 @@
 huggingface_hub
-diffusers>=0.31.0
+diffusers>=0.33.1
 accelerate>=0.33.0
 einops
 peft