Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

Remove line made unnecessary (and wrong) after transformer_options was added to NextDiT's _forward definition (#11118)

Jedrzej Kosinski0ec05b1481d1
diff --git a/comfy/ldm/lumina/model.py b/comfy/ldm/lumina/model.py
index f1c1a0ec3..6c24fed9b 100644
--- a/comfy/ldm/lumina/model.py
+++ b/comfy/ldm/lumina/model.py
@@ -586,7 +586,6 @@ class NextDiT(nn.Module):
cap_feats = self.cap_embedder(cap_feats) # (N, L, D) # todo check if able to batchify w.o. redundant compute
patches = transformer_options.get("patches", {})
- transformer_options = kwargs.get("transformer_options", {})
x_is_tensor = isinstance(x, torch.Tensor)
img, mask, img_size, cap_size, freqs_cis = self.patchify_and_embed(x, cap_feats, cap_mask, t, num_tokens, transformer_options=transformer_options)
freqs_cis = freqs_cis.to(img.device)