2024-12-10 20:37:19,609 - __main__ - INFO - num_videos: 50
output_root_dir: outputs_video_video_prompts
output_dir: ${output_root_dir}/${now:%Y-%m-%d_%H-%M-%S}/benchmark_video_generation
generate_kwargs:
  negative_prompt: ''
  num_inference_steps: 50
  guidance_scale: 6
  num_frames: 45
prompt_sources:
- static/prompts/20241130_open_chatgpt4o_videos.txt
get_pipeline:
  _target_: accediff.utils.factory.pipeline_factory.get_cogvideox_pipeline_with_tome_3d
  repo_id: THUDM/CogVideoX-2b
  hook_kwargs:
    merge_ratio: 0.3
    st: 4
    sx: 6
    sy: 9
  tome_modele_names:
  - transformer_blocks.{0..27..2}.attn1

2024-12-10 20:37:21,517 - bitsandbytes.cextension - DEBUG - Loading bitsandbytes native library from: /scratch/yz10381/miniconda3/envs/diffusers/lib/python3.10/site-packages/bitsandbytes/libbitsandbytes_cuda124.so
2024-12-10 20:37:22,214 - matplotlib - DEBUG - matplotlib data path: /scratch/yz10381/miniconda3/envs/diffusers/lib/python3.10/site-packages/matplotlib/mpl-data
2024-12-10 20:37:22,218 - matplotlib - DEBUG - CONFIGDIR=/home/yz10381/.config/matplotlib
2024-12-10 20:37:22,218 - matplotlib - DEBUG - interactive is False
2024-12-10 20:37:22,218 - matplotlib - DEBUG - platform is linux
2024-12-10 20:37:22,248 - matplotlib - DEBUG - CACHEDIR=/scratch/yz10381/.cache/matplotlib
2024-12-10 20:37:22,248 - matplotlib.font_manager - DEBUG - Using fontManager instance from /scratch/yz10381/.cache/matplotlib/fontlist-v390.json
2024-12-10 20:37:22,366 - urllib3.connectionpool - DEBUG - Starting new HTTPS connection (1): huggingface.co:443
2024-12-10 20:37:22,419 - urllib3.connectionpool - DEBUG - https://huggingface.co:443 "GET /api/models/THUDM/CogVideoX-2b HTTP/11" 200 1943
2024-12-10 20:37:22,475 - urllib3.connectionpool - DEBUG - https://huggingface.co:443 "HEAD /THUDM/CogVideoX-2b/resolve/main/model_index.json HTTP/11" 200 0
2024-12-10 20:37:29,220 - accediff.utils.factory.pipeline_factory - INFO - Available config keys: odict_keys(['num_attention_heads', 'attention_head_dim', 'in_channels', 'out_channels', 'flip_sin_to_cos', 'freq_shift', 'time_embed_dim', 'text_embed_dim', 'num_layers', 'dropout', 'attention_bias', 'sample_width', 'sample_height', 'sample_frames', 'patch_size', 'temporal_compression_ratio', 'max_text_seq_length', 'activation_fn', 'timestep_activation_fn', 'norm_elementwise_affine', 'norm_eps', 'spatial_interpolation_scale', 'temporal_interpolation_scale', 'use_rotary_positional_embeddings', 'use_learned_positional_embeddings', '_use_default_values', '_class_name', '_diffusers_version', '_name_or_path'])
2024-12-10 20:37:29,220 - accediff.utils.factory.pipeline_factory - INFO - f: 12, h: 30, w: 45, st: 4, sx: 6, sy: 9
2024-12-10 20:37:29,222 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.0.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,222 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.0.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,222 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.2.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,222 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.2.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,222 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.4.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,222 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.4.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.6.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.6.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.8.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.8.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.10.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.10.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.12.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.12.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.14.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.14.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.16.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.16.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.18.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.18.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.20.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.20.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.22.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.22.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.24.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.24.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.26.attn1 with pre forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,224 - accediff.utils.factory.pipeline_factory - INFO - Wrapping transformer_blocks.26.attn1 with post forward hook <function _get_cogvideox_tome_token_merging_hooks_3d.<locals>.tome_post_forward_hook at 0x14cd00587c70>
2024-12-10 20:37:29,226 - __main__ - INFO - Saving videos to outputs_video_video_prompts/2024-12-10_20-37-19/benchmark_video_generation/videos
2024-12-10 20:37:29,227 - __main__ - INFO - Running iter_generate_prompt2video with 3 warmup iterations
2024-12-10 20:38:11,009 - __main__ - INFO - Warmup step 1: 41.23896765708923 seconds
2024-12-10 20:38:50,900 - __main__ - INFO - Warmup step 2: 39.46807265281677 seconds
2024-12-10 20:39:30,254 - __main__ - INFO - Warmup step 3: 38.93011784553528 seconds
2024-12-10 21:10:36,819 - __main__ - INFO - Total time: 1847.8315742015839 seconds for 47 iterations, throughput: 0.02543521858603801 iterations/second