Hunter Senft-Grupp 68df63096b fix: use float32 for Qwen image rotary embeddings ... Fixes the error when using torch.compile with bfloat16 inference: 'No backend can handle apply_rope1: eager: freqs_cis: dtype torch.bfloat16 not in {torch.float32}' The apply_rope1 function uses addcmul_ in-place operations that fail under torch.compile when freqs_cis is bfloat16. This restores the behavior from commit 4cd88186 which was inadvertently reverted in commit c4a6b389.		2026-01-06 22:42:55 -05:00
..
controlnet.py	Fix qwen controlnet regression. (#10657)	2025-11-05 18:07:35 -05:00
model.py	fix: use float32 for Qwen image rotary embeddings	2026-01-06 22:42:55 -05:00