From d204e532919e509360be2a8bc10ea89e9b21f79e Mon Sep 17 00:00:00 2001 From: Aryan Date: Wed, 16 Oct 2024 16:25:41 +0530 Subject: [PATCH] [core] improve VAE encode/decode framewise batching (#9684) * update * apply suggestions from review --------- Co-authored-by: Sayak Paul --- .../models/autoencoders/autoencoder_kl_cogvideox.py | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/src/diffusers/models/autoencoders/autoencoder_kl_cogvideox.py b/src/diffusers/models/autoencoders/autoencoder_kl_cogvideox.py index 7834206ddb..68b49d72ac 100644 --- a/src/diffusers/models/autoencoders/autoencoder_kl_cogvideox.py +++ b/src/diffusers/models/autoencoders/autoencoder_kl_cogvideox.py @@ -1182,7 +1182,8 @@ class AutoencoderKLCogVideoX(ModelMixin, ConfigMixin, FromOriginalModelMixin): frame_batch_size = self.num_sample_frames_batch_size # Note: We expect the number of frames to be either `1` or `frame_batch_size * k` or `frame_batch_size * k + 1` for some k. - num_batches = num_frames // frame_batch_size if num_frames > 1 else 1 + # As the extra single frame is handled inside the loop, it is not required to round up here. + num_batches = max(num_frames // frame_batch_size, 1) conv_cache = None enc = [] @@ -1330,7 +1331,8 @@ class AutoencoderKLCogVideoX(ModelMixin, ConfigMixin, FromOriginalModelMixin): row = [] for j in range(0, width, overlap_width): # Note: We expect the number of frames to be either `1` or `frame_batch_size * k` or `frame_batch_size * k + 1` for some k. - num_batches = num_frames // frame_batch_size if num_frames > 1 else 1 + # As the extra single frame is handled inside the loop, it is not required to round up here. + num_batches = max(num_frames // frame_batch_size, 1) conv_cache = None time = [] @@ -1409,7 +1411,7 @@ class AutoencoderKLCogVideoX(ModelMixin, ConfigMixin, FromOriginalModelMixin): for i in range(0, height, overlap_height): row = [] for j in range(0, width, overlap_width): - num_batches = num_frames // frame_batch_size + num_batches = max(num_frames // frame_batch_size, 1) conv_cache = None time = []