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.

Don't leave very large hidden states in the clip vision output.

comfyanonymoused58730658d0
diff --git a/comfy/clip_vision.py b/comfy/clip_vision.py
index 9b95ae003..1206c680d 100644
--- a/comfy/clip_vision.py
+++ b/comfy/clip_vision.py
@@ -56,6 +56,7 @@ class ClipVisionModel():
if t is not None:
if k == 'hidden_states':
outputs["penultimate_hidden_states"] = t[-2].cpu()
+ outputs["hidden_states"] = None
else:
outputs[k] = t.cpu()