Open-Sora: Democratizing Efficient Video Production for All
Python
Force fp16 input to fp32 to avoid nan output in timestep_transform
HangXu04d2ee0182db
diff --git a/opensora/schedulers/rf/rectified_flow.py b/opensora/schedulers/rf/rectified_flow.pyindex 58d7b48..8acaff5 100644--- a/opensora/schedulers/rf/rectified_flow.py+++ b/opensora/schedulers/rf/rectified_flow.py@@ -15,6 +15,11 @@ def timestep_transform(scale=1.0,num_timesteps=1,):+ # Force fp16 input to fp32 to avoid nan output+ for key in ["height", "width", "num_frames"]:+ if model_kwargs[key].dtype == torch.float16:+ model_kwargs[key] = model_kwargs[key].float()+t = t / num_timestepsresolution = model_kwargs["height"] * model_kwargs["width"]ratio_space = (resolution / base_resolution).sqrt()