Open-Sora: Democratizing Efficient Video Production for All
Python
dataset trainscript
Shen-Chenhui2c391a636b89
diff --git a/scripts/train-vae-v2.py b/scripts/train-vae-v2.pyindex c65c3b3..621f7bd 100644--- a/scripts/train-vae-v2.py+++ b/scripts/train-vae-v2.py@@ -58,7 +58,7 @@ def main():if exp_dir is None:experiment_index = len(glob(f"{cfg.outputs}/*")) - 1model_name = cfg.model["type"].replace("/", "-")- exp_name = f"{experiment_index:03d}-F{cfg.num_frames}S{cfg.frame_interval}-{model_name}"+ exp_name = f"{experiment_index:03d}-F{cfg.dataset.num_frames}S{cfg.dataset.frame_interval}-{model_name}"exp_dir = f"{cfg.outputs}/{exp_name}"assert os.path.exists(exp_dir)@@ -267,8 +267,8 @@ def main():# calculate discriminator_time_paddingdisc_time_downsample_factor = 2 ** len(cfg.discriminator.channel_multipliers)- if cfg.num_frames % disc_time_downsample_factor != 0:- disc_time_padding = disc_time_downsample_factor - cfg.num_frames % disc_time_downsample_factor+ if cfg.dataset.num_frames % disc_time_downsample_factor != 0:+ disc_time_padding = disc_time_downsample_factor - cfg.dataset.num_frames % disc_time_downsample_factorelse:disc_time_padding = 0video_contains_first_frame = cfg.video_contains_first_frame@@ -314,8 +314,8 @@ def main():# supprt for image or video inputsassert x.ndim in {4, 5}, f"received input of {x.ndim} dimensions" # either image or videoassert (- x.shape[-2:] == cfg.image_size- ), f"received input size {x.shape[-2:]}, but config image size is {cfg.image_size}"+ x.shape[-2:] == cfg.dataset.image_size+ ), f"received input size {x.shape[-2:]}, but config image size is {cfg.dataset.image_size}"is_image = x.ndim == 4if is_image:video = rearrange(x, "b c ... -> b c 1 ...")