Precomputation folder name based on model name causes HunyuanVideo Lora training error

[LiuXiaolong19920720]

System Info / 系統信息

python3.10
pytorch2.5.1
cuda12.1

Information / 问题信息

• The official example scripts / 官方的示例脚本
• My own modified scripts / 我自己修改的脚本和任务

Reproduction / 复现过程

1. follow https://github.com/a-r-r-o-w/finetrainers/blob/main/docs/training/hunyuan_video.md

Expected behavior / 期待表现

Model name based precomputation folder name is used when precompute latents and conditions. But the path in PrecomputedDataset keep use the old format path.

class PrecomputedDataset(Dataset):
    def __init__(self, data_root: str) -> None:
        super().__init__()

        self.data_root = Path(data_root)

        self.latents_path = self.data_root / PRECOMPUTED_DIR_NAME / PRECOMPUTED_LATENTS_DIR_NAME
        self.conditions_path = self.data_root / PRECOMPUTED_DIR_NAME / PRECOMPUTED_CONDITIONS_DIR_NAME

[a-r-r-o-w]

Yes, sorry this slipped through. This is being fixed in #199