From dc1ff288e8046f5526ad48dbcdaaccbc2502c00d Mon Sep 17 00:00:00 2001 From: hailin Date: Fri, 6 Mar 2026 03:20:02 -0800 Subject: [PATCH] fix: save null vectors as bfloat16 to match training dtype Co-Authored-By: Claude Sonnet 4.6 --- scripts/diffusion/gen_smoke_data.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/scripts/diffusion/gen_smoke_data.py b/scripts/diffusion/gen_smoke_data.py index 839f8f2..6a9634e 100644 --- a/scripts/diffusion/gen_smoke_data.py +++ b/scripts/diffusion/gen_smoke_data.py @@ -75,8 +75,9 @@ def main(): print(f"Wrote {N_SAMPLES} samples to {csv_path}") # null vectors（空提示的 embedding，用于 cfg dropout） - null_t5 = torch.zeros(1, T5_SEQ, T5_DIM) - null_clip = torch.zeros(1, CLIP_DIM) + # 必须用 bf16，与训练 dtype 一致（train.py 加载后不做 dtype 转换） + null_t5 = torch.zeros(1, T5_SEQ, T5_DIM, dtype=torch.bfloat16) + null_clip = torch.zeros(1, CLIP_DIM, dtype=torch.bfloat16) torch.save(null_t5, f"{NULL_DIR}/null_t5.pt") torch.save(null_clip, f"{NULL_DIR}/null_clip.pt") print(f"Wrote null vectors to {NULL_DIR}/")