diff --git "a/test/rainy/weather/distilled_step4_bg0.1_rm1/debug.log" "b/test/rainy/weather/distilled_step4_bg0.1_rm1/debug.log" new file mode 100644--- /dev/null +++ "b/test/rainy/weather/distilled_step4_bg0.1_rm1/debug.log" @@ -0,0 +1,1484 @@ +[06-21 23:14:18|DEBUG|packages/cosmos-oss/cosmos_oss/init.py:176:init_output_dir] Flags(internal=False, training=True, smoke=False, verbose=False, experimental_checkpoints=True) +[06-21 23:14:19|DEBUG|cosmos_transfer2/_src/imaginaire/utils/env_parsers/env_parser.py:86:get_val_dict] getting val dict of CredentialEnvParser +[06-21 23:14:19|DEBUG|cosmos_transfer2/_src/imaginaire/utils/env_parsers/env_parser.py:86:get_val_dict] getting val dict of CredentialEnvParser +[06-21 23:14:30|CRITICAL|cosmos_transfer2/_src/interactive/checkpointer/dcp.py:181:] for the back comptiable pytorch! New DefaultLoadPlanner class is created. +[06-21 23:14:31|CRITICAL|cosmos_transfer2/_src/predict2/checkpointer/dcp.py:181:] for the back comptiable pytorch! New DefaultLoadPlanner class is created. +[06-21 23:14:31|DEBUG|cosmos_transfer2/inference.py:46:__init__] SetupArguments(output_dir=PosixPath('/scratch/ssaha/cosmos-transfer2.5/outputs/edge_threshold_grid/rainy/weather/distilled_step4_bg0.1_rm1') model='edge/distilled' checkpoint_path='s3://bucket/cosmos_interactive_fastgen/cosmos_interactive/cosmos_fastgen_dmd2_trigflow_distill_cosmos_transfer2p5_2B_bidirectional_edge_bugfix_v2/checkpoints/iter_000010000' experiment='dmd2_trigflow_distill_cosmos_transfer2p5_2B_bidirectional_edge' config_file='cosmos_transfer2/_src/interactive/configs/registry_transfer2p5.py' context_parallel_size=1 disable_guardrails=True offload_guardrail_models=False keep_going=True profile=False benchmark=False compile_tokenizer= enable_parallel_tokenizer=False parallel_tokenizer_grid=(-1, -1))(['edge']) +[06-21 23:14:31|DEBUG|cosmos_transfer2/_src/imaginaire/attention/utils/safe_log.py:35:debug] Flash Attention v3 is not supported because the Python package ('flash_attn_3_nv'_) was not found. +[06-21 23:14:31|DEBUG|cosmos_transfer2/_src/imaginaire/attention/utils/safe_log.py:35:debug] NATTEN Attention is not supported due to insufficient NATTEN version 0.21.0, expected at least 0.21.5.dev9. +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering video_basic_augmentor_v1... +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering video_basic_augmentor_v2... +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering image_basic_augmentor... +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering video_basic_augmentor_v1_with_control... +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering video_basic_augmentor_v2_with_control... +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering image_basic_augmentor_with_control... +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering video_basic_augmentor_v2_with_control_and_image_context... +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering video_basic_augmentor_with_control_input... +[06-21 23:14:31|INFO|cosmos_transfer2/_src/transfer2/datasets/augmentor_provider.py:35:augmentor_register] registering hdmap_augmentor_for_local_datasets... +[06-21 23:14:31|CRITICAL|cosmos_transfer2/_src/imaginaire/utils/config_helper.py:192:import_all_modules_from_package] Reloading all modules from package cosmos_transfer2._src.interactive.configs.registry_experiment +[06-21 23:14:31|DEBUG|cosmos_transfer2/_src/imaginaire/utils/config_helper.py:210:import_modules_recursively] Reloading module cosmos_transfer2._src.interactive.configs.registry_experiment.experiment_list +[06-21 23:14:31|DEBUG|cosmos_transfer2/_src/imaginaire/utils/config_helper.py:210:import_modules_recursively] Reloading module cosmos_transfer2._src.interactive.configs.registry_experiment.experiments_dmd2_ac_predict2p5 +[06-21 23:14:31|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:297:download] Downloading checkpoint nvidia/Cosmos-Predict2.5-2B/robot/action-cond(38c6c645-7d41-4560-8eeb-6f4ddc0e6574) +[06-21 23:14:31|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:165:_hf_download] uvx 'hf>=1.3.5' download nvidia/Cosmos-Predict2.5-2B --repo-type model --revision main robot/action-cond/38c6c645-7d41-4560-8eeb-6f4ddc0e6574_ema_bf16.pt +[06-21 23:14:32|DEBUG|cosmos_transfer2/_src/imaginaire/utils/config_helper.py:210:import_modules_recursively] Reloading module cosmos_transfer2._src.interactive.configs.registry_experiment.experiments_dmd2_predict2p5 +[06-21 23:14:32|DEBUG|cosmos_transfer2/_src/imaginaire/utils/config_helper.py:210:import_modules_recursively] Reloading module cosmos_transfer2._src.interactive.configs.registry_experiment.experiments_dmd2_transfer2p5 +[06-21 23:14:32|DEBUG|cosmos_transfer2/_src/imaginaire/utils/config_helper.py:210:import_modules_recursively] Reloading module cosmos_transfer2._src.interactive.configs.registry_experiment.experiments_knowledge_distill_ac_predict2p5 +[06-21 23:14:32|DEBUG|cosmos_transfer2/_src/imaginaire/utils/config_helper.py:210:import_modules_recursively] Reloading module cosmos_transfer2._src.interactive.configs.registry_experiment.experiments_self_forcing_ac_predict2p5 +[06-21 23:14:32|INFO|cosmos_transfer2/_src/imaginaire/utils/misc.py:151:set_random_seed] Using random seed 0. +[06-21 23:14:32|INFO|cosmos_transfer2/_src/interactive/utils/model_loader.py:81:load_model_from_checkpoint] Loading model from s3://bucket/cosmos_interactive_fastgen/cosmos_interactive/cosmos_fastgen_dmd2_trigflow_distill_cosmos_transfer2p5_2B_bidirectional_edge_bugfix_v2/checkpoints/iter_000010000 +[06-21 23:14:32|INFO|cosmos_transfer2/_src/interactive/utils/model_loader.py:109:load_model_from_checkpoint] Setting load_teacher_weights=False for inference to skip teacher checkpoint download +[06-21 23:14:32|INFO|cosmos_transfer2/_src/interactive/methods/cosmos2_interactive_model.py:94:__init__] Setting precision to torch.bfloat16 +[06-21 23:14:32|INFO|cosmos_transfer2/_src/predict2/text_encoders/text_encoder.py:74:__init__] Instantiating text encoder model... +[06-21 23:14:32|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:297:download] Downloading checkpoint Qwen/Qwen2.5-VL-7B-Instruct(7219c6c7-f878-4137-bbdb-76842ea85e70) +[06-21 23:14:32|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:165:_hf_download] uvx 'hf>=1.3.5' download nvidia/Cosmos-Reason1-7B --repo-type model --revision 3210bec0495fdc7a8d3dbb8d58da5711eab4b423 --include '*' +[06-21 23:14:33|INFO|cosmos_transfer2/_src/reason1/tokenizer/processor.py:75:__init__] Successfully loaded processor from local cache +[06-21 23:14:33|INFO|cosmos_transfer2/_src/reason1/models/vlm_base.py:116:__init__] Setting torch default dtype from torch.float32 to torch.bfloat16 +[06-21 23:14:33|INFO|cosmos_transfer2/_src/reason1/models/vlm_base.py:123:__init__] Reset torch default dtype to torch.float32 +[06-21 23:14:33|INFO|cosmos_transfer2/_src/predict2/text_encoders/text_encoder.py:84:__init__] Loading checkpoint from s3://bucket/cosmos_reasoning1/sft_exp700/sft_exp721-1_qwen7b_tl_721_5vs5_s3_balanced_n32_resume_16k/checkpoints/iter_000016000/model/. +[06-21 23:14:33|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:297:download] Downloading checkpoint nvidia/Cosmos-Reason1.1-7B(cb3e3ffa-7b08-4c34-822d-61c7aa31a14f) +[06-21 23:14:33|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:165:_hf_download] uvx 'hf>=1.3.5' download nvidia/Cosmos-Reason1-7B --repo-type model --revision 3210bec0495fdc7a8d3dbb8d58da5711eab4b423 --include '*' +[06-21 23:14:37|INFO|cosmos_transfer2/_src/predict2/text_encoders/text_encoder.py:113:__init__] Finished loading checkpoint from /scratch/ssaha/.cache/huggingface/hub/models--nvidia--Cosmos-Reason1-7B/snapshots/3210bec0495fdc7a8d3dbb8d58da5711eab4b423. +[06-21 23:14:37|INFO|cosmos_transfer2/_src/predict2/text_encoders/text_encoder.py:116:__init__] Text encoder model instantiated +[06-21 23:14:37|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:297:download] Downloading checkpoint Wan2.1/vae(685afcaa-4de2-42fe-b7b9-69f7a2dee4d8) +[06-21 23:14:37|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:165:_hf_download] uvx 'hf>=1.3.5' download nvidia/Cosmos-Predict2.5-2B --repo-type model --revision f176dc95b4a70f53ce01c4b302851595e7322b00 tokenizer.pth +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/tokenizers/wan2pt1.py:669:_video_vae] loading /scratch/ssaha/.cache/huggingface/hub/models--nvidia--Cosmos-Predict2.5-2B/snapshots/f176dc95b4a70f53ce01c4b302851595e7322b00/tokenizer.pth +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1624:build_pos_embed] Building positional embedding with rope3d class, impl +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:886:__init__] Using AdaLN LoRA Flag: True. We enable bias if no AdaLN LoRA for backward compatibility. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 28 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 4 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 5 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 6 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 7 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 8 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 9 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 10 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 11 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 12 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 13 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 14 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 15 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 16 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 17 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 18 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 19 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 20 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 21 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 22 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 23 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 24 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 25 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 26 +[06-21 23:14:38|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 27 +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:38|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 28 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 4 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 5 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 6 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 7 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 8 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 9 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 10 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 11 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 12 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 13 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 14 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 15 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 16 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 17 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 18 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 19 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 20 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 21 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 22 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 23 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 24 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 25 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 26 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 27 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 4 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/imaginaire/utils/timer.py:150:_log] Time spent on Cosmos2InteractiveModel: build_net: 0.67 s +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #0-fps: + ReMapkey + input key: fps + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: fps + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #1-padding_mask: + ReMapkey + input key: padding_mask + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: padding_mask + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #2-text: + TextAttr + input key: ['t5_text_embeddings'] + Param count: 0 + Trainable: None + Dropout rate: 0.2 + Output key: [crossattn_emb] +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #3-use_video_condition: + BooleanFlag + input key: fps + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: use_video_condition + This is a boolean flag +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #4-control_input_edge: + ReMapkey + input key: control_input_edge + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_edge + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #5-control_input_vis: + ReMapkey + input key: control_input_vis + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_vis + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #6-control_input_depth: + ReMapkey + input key: control_input_depth + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_depth + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #7-control_input_seg: + ReMapkey + input key: control_input_seg + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_seg + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #8-control_input_inpaint: + ReMapkey + input key: control_input_inpaint + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_inpaint + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #9-control_input_edge_mask: + ReMapkey + input key: control_input_edge_mask + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_edge_mask + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #10-control_input_vis_mask: + ReMapkey + input key: control_input_vis_mask + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_vis_mask + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #11-control_input_depth_mask: + ReMapkey + input key: control_input_depth_mask + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_depth_mask + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #12-control_input_seg_mask: + ReMapkey + input key: control_input_seg_mask + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_seg_mask + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/conditioner.py:434:__init__] Initialized embedder #13-control_input_inpaint_mask: + ReMapkey + input key: control_input_inpaint_mask + Param count: 0 + Trainable: None + Dropout rate: 0.0 + Output key: control_input_inpaint_mask + Dtype: None +[06-21 23:14:39|INFO|cosmos_transfer2/_src/interactive/methods/cosmos2_interactive_model.py:231:build_model] + +==============config condition_postprocessor: {'hint_keys': ['edge'], '_target_': } +[06-21 23:14:39|INFO|cosmos_transfer2/_src/interactive/methods/cosmos2_interactive_model.py:232:build_model] + +==============condition_postprocessor: +[06-21 23:14:39|INFO|cosmos_transfer2/_src/imaginaire/utils/timer.py:150:_log] Time spent on Cosmos2InteractiveModel: build_model: 6.54 s +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1624:build_pos_embed] Building positional embedding with rope3d class, impl +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:886:__init__] Using AdaLN LoRA Flag: True. We enable bias if no AdaLN LoRA for backward compatibility. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 28 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 4 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 5 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 6 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 7 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 8 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 9 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 10 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 11 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 12 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 13 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 14 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 15 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 16 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 17 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 18 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 19 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 20 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 21 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 22 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 23 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 24 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 25 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 26 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 27 +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 28 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 4 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 5 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 6 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 7 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 8 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 9 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 10 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 11 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 12 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 13 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 14 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 15 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 16 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 17 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 18 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 19 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 20 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 21 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 22 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 23 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 24 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 25 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 26 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 27 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 4 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:39|INFO|cosmos_transfer2/_src/imaginaire/utils/timer.py:150:_log] Time spent on Cosmos2InteractiveModel: build_net: 0.65 s +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1624:build_pos_embed] Building positional embedding with rope3d class, impl +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:886:__init__] Using AdaLN LoRA Flag: True. We enable bias if no AdaLN LoRA for backward compatibility. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:39|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 28 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 4 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 5 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 6 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 7 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 8 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 9 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 10 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 11 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 12 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 13 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 14 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 15 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 16 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 17 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 18 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 19 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 20 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 21 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 22 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 23 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 24 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 25 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 26 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 27 +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is None and using 16 heads with a dimension of 128. +[06-21 23:14:40|DEBUG|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:432:__init__] Setting up Attention. Query dim is 2048, context_dim is 1024 and using 16 heads with a dimension of 128. +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 28 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 4 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 5 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 6 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 7 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 8 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 9 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 10 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 11 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 12 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 13 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 14 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 15 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 16 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 17 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 18 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 19 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 20 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 21 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 22 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 23 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 24 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 25 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 26 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 27 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1804:enable_selective_checkpoint] Enable selective checkpoint with predict2_2b_720_aggressive, for every 1 blocks. Total blocks: 4 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 0 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 1 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 2 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/predict2/networks/minimal_v4_dit.py:1810:enable_selective_checkpoint] Enable selective checkpoint for block 3 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/imaginaire/utils/timer.py:150:_log] Time spent on Cosmos2InteractiveModel: build_net: 0.65 s +[06-21 23:14:40|INFO|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:86:build_model] ==========Instantiating networks...========== +[06-21 23:14:40|INFO|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:87:build_model] net: MinimalV4LVGControlVaceDiT( + (x_embedder): PatchEmbed( + (proj): Sequential( + (0): Rearrange('b c (t r) (h m) (w n) -> b t h w (c r m n)', r=1, m=2, n=2) + (1): Linear(in_features=72, out_features=2048, bias=False) + ) + ) + (pos_embedder): VideoRopePosition3DEmb() + (t_embedder): Sequential( + (0): Timesteps() + (1): TimestepEmbedding( + (linear_1): Linear(in_features=2048, out_features=2048, bias=False) + (activation): SiLU() + (linear_2): Linear(in_features=2048, out_features=6144, bias=False) + ) + ) + (blocks): ModuleList( + (0-27): 28 x CheckpointWrapper( + (_checkpoint_wrapped_module): ControlAwareDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + ) + ) + ) + (final_layer): CheckpointWrapper( + (_checkpoint_wrapped_module): CheckpointWrapper( + (_checkpoint_wrapped_module): CheckpointWrapper( + (_checkpoint_wrapped_module): FinalLayer( + (layer_norm): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (linear): Linear(in_features=2048, out_features=64, bias=False) + (adaln_modulation): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=4096, bias=False) + ) + ) + ) + ) + ) + (t_embedding_norm): RMSNorm() + (crossattn_proj): Sequential( + (0): Linear(in_features=100352, out_features=1024, bias=True) + (1): GELU(approximate='none') + ) + (control_embedder): PatchEmbed( + (proj): Sequential( + (0): Rearrange('b c (t r) (h m) (w n) -> b t h w (c r m n)', r=1, m=2, n=2) + (1): Linear(in_features=520, out_features=2048, bias=False) + ) + ) + (control_blocks): ModuleList( + (0): CheckpointWrapper( + (_checkpoint_wrapped_module): ControlEncoderDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (before_proj): Linear(in_features=2048, out_features=2048, bias=True) + (after_proj): Linear(in_features=2048, out_features=2048, bias=True) + ) + ) + (1-3): 3 x CheckpointWrapper( + (_checkpoint_wrapped_module): ControlEncoderDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (after_proj): Linear(in_features=2048, out_features=2048, bias=True) + ) + ) + ) +) +[06-21 23:14:40|INFO|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:88:build_model] net_teacher: MinimalV4LVGControlVaceDiT( + (x_embedder): PatchEmbed( + (proj): Sequential( + (0): Rearrange('b c (t r) (h m) (w n) -> b t h w (c r m n)', r=1, m=2, n=2) + (1): Linear(in_features=72, out_features=2048, bias=False) + ) + ) + (pos_embedder): VideoRopePosition3DEmb() + (t_embedder): Sequential( + (0): Timesteps() + (1): TimestepEmbedding( + (linear_1): Linear(in_features=2048, out_features=2048, bias=False) + (activation): SiLU() + (linear_2): Linear(in_features=2048, out_features=6144, bias=False) + ) + ) + (blocks): ModuleList( + (0-27): 28 x CheckpointWrapper( + (_checkpoint_wrapped_module): ControlAwareDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + ) + ) + ) + (final_layer): CheckpointWrapper( + (_checkpoint_wrapped_module): CheckpointWrapper( + (_checkpoint_wrapped_module): CheckpointWrapper( + (_checkpoint_wrapped_module): FinalLayer( + (layer_norm): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (linear): Linear(in_features=2048, out_features=64, bias=False) + (adaln_modulation): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=4096, bias=False) + ) + ) + ) + ) + ) + (t_embedding_norm): RMSNorm() + (crossattn_proj): Sequential( + (0): Linear(in_features=100352, out_features=1024, bias=True) + (1): GELU(approximate='none') + ) + (control_embedder): PatchEmbed( + (proj): Sequential( + (0): Rearrange('b c (t r) (h m) (w n) -> b t h w (c r m n)', r=1, m=2, n=2) + (1): Linear(in_features=520, out_features=2048, bias=False) + ) + ) + (control_blocks): ModuleList( + (0): CheckpointWrapper( + (_checkpoint_wrapped_module): ControlEncoderDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (before_proj): Linear(in_features=2048, out_features=2048, bias=True) + (after_proj): Linear(in_features=2048, out_features=2048, bias=True) + ) + ) + (1-3): 3 x CheckpointWrapper( + (_checkpoint_wrapped_module): ControlEncoderDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (after_proj): Linear(in_features=2048, out_features=2048, bias=True) + ) + ) + ) +) +[06-21 23:14:40|INFO|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:89:build_model] net_fake_score: MinimalV4LVGControlVaceDiT( + (x_embedder): PatchEmbed( + (proj): Sequential( + (0): Rearrange('b c (t r) (h m) (w n) -> b t h w (c r m n)', r=1, m=2, n=2) + (1): Linear(in_features=72, out_features=2048, bias=False) + ) + ) + (pos_embedder): VideoRopePosition3DEmb() + (t_embedder): Sequential( + (0): Timesteps() + (1): TimestepEmbedding( + (linear_1): Linear(in_features=2048, out_features=2048, bias=False) + (activation): SiLU() + (linear_2): Linear(in_features=2048, out_features=6144, bias=False) + ) + ) + (blocks): ModuleList( + (0-27): 28 x CheckpointWrapper( + (_checkpoint_wrapped_module): ControlAwareDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + ) + ) + ) + (final_layer): CheckpointWrapper( + (_checkpoint_wrapped_module): CheckpointWrapper( + (_checkpoint_wrapped_module): CheckpointWrapper( + (_checkpoint_wrapped_module): FinalLayer( + (layer_norm): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (linear): Linear(in_features=2048, out_features=64, bias=False) + (adaln_modulation): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=4096, bias=False) + ) + ) + ) + ) + ) + (t_embedding_norm): RMSNorm() + (crossattn_proj): Sequential( + (0): Linear(in_features=100352, out_features=1024, bias=True) + (1): GELU(approximate='none') + ) + (control_embedder): PatchEmbed( + (proj): Sequential( + (0): Rearrange('b c (t r) (h m) (w n) -> b t h w (c r m n)', r=1, m=2, n=2) + (1): Linear(in_features=520, out_features=2048, bias=False) + ) + ) + (control_blocks): ModuleList( + (0): CheckpointWrapper( + (_checkpoint_wrapped_module): ControlEncoderDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (before_proj): Linear(in_features=2048, out_features=2048, bias=True) + (after_proj): Linear(in_features=2048, out_features=2048, bias=True) + ) + ) + (1-3): 3 x CheckpointWrapper( + (_checkpoint_wrapped_module): ControlEncoderDiTBlock( + (layer_norm_self_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (self_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=2048, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=2048, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_cross_attn): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (cross_attn): Attention( + (q_proj): Linear(in_features=2048, out_features=2048, bias=False) + (q_norm): RMSNorm() + (k_proj): Linear(in_features=1024, out_features=2048, bias=False) + (k_norm): RMSNorm() + (v_proj): Linear(in_features=1024, out_features=2048, bias=False) + (v_norm): Identity() + (output_proj): Linear(in_features=2048, out_features=2048, bias=False) + (output_dropout): Identity() + (attn_op): MinimalA2AAttnOp() + ) + (layer_norm_mlp): LayerNorm((2048,), eps=1e-06, elementwise_affine=False) + (mlp): GPT2FeedForward( + (activation): GELU(approximate='none') + (layer1): Linear(in_features=2048, out_features=8192, bias=False) + (layer2): Linear(in_features=8192, out_features=2048, bias=False) + ) + (adaln_modulation_self_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_cross_attn): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (adaln_modulation_mlp): Sequential( + (0): SiLU() + (1): Linear(in_features=2048, out_features=256, bias=False) + (2): Linear(in_features=256, out_features=6144, bias=False) + ) + (after_proj): Linear(in_features=2048, out_features=2048, bias=True) + ) + ) + ) +) +[06-21 23:14:40|INFO|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:128:build_model] Freezing the CR1 embedding projection layer in student net.. +[06-21 23:14:40|INFO|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:132:build_model] Freezing the CR1 embedding projection layer in fake score net.. +[06-21 23:14:40|INFO|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:135:build_model] Freezing teacher net.. +[06-21 23:14:40|CRITICAL|cosmos_transfer2/_src/interactive/methods/cosmos2_interactive_model.py:119:setup_loss_specs] Using mean loss reduce with loss scale 1.0 +[06-21 23:14:40|INFO|cosmos_transfer2/_src/imaginaire/utils/timer.py:150:_log] Time spent on instantiate model: 7.88 s +[06-21 23:14:40|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:297:download] Downloading checkpoint nvidia/Cosmos-Transfer2.5-2B/distilled/edge(41f07f13-f2e4-4e34-ba4c-86f595acbc20) +[06-21 23:14:40|INFO|cosmos_transfer2/_src/imaginaire/utils/checkpoint_db.py:165:_hf_download] uvx 'hf>=1.3.5' download nvidia/Cosmos-Transfer2.5-2B --repo-type model --revision bbaeedb2b57cc8b14a44653099e2551adb69dcc7 distilled/general/edge/41f07f13-f2e4-4e34-ba4c-86f595acbc20_ema_bf16.pt +[06-21 23:14:41|INFO|cosmos_transfer2/_src/interactive/utils/model_loader.py:214:load_model_state_dict_from_checkpoint] Loading model cached locally from /scratch/ssaha/.cache/huggingface/hub/models--nvidia--Cosmos-Transfer2.5-2B/snapshots/bbaeedb2b57cc8b14a44653099e2551adb69dcc7/distilled/general/edge/41f07f13-f2e4-4e34-ba4c-86f595acbc20_ema_bf16.pt +[06-21 23:14:41|INFO|cosmos_transfer2/_src/interactive/utils/model_loader.py:173:_process_pt_state_dict_for_ema] Loading EMA-exported checkpoint with net.* keys directly +[06-21 23:14:41|CRITICAL|cosmos_transfer2/_src/interactive/methods/cosmos2_interactive_model.py:408:load_state_dict] load model in non-strict mode +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.0.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.0.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.0.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.0.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.1.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.1.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.1.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.1.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.2.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.2.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.2.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.2.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.3.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.3.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.3.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.3.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.4.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.4.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.4.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.4.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.5.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.5.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.5.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.5.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.6.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.6.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.6.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.6.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.7.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.7.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.7.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.7.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.8.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.8.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.8.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.8.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.9.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.9.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.9.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.9.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.10.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.10.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.10.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.10.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.11.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.11.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.11.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.11.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.12.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.12.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.12.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.12.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.13.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.13.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.13.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.13.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.14.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.14.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.14.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.14.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.15.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.15.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.15.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.15.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.16.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.16.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.16.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.16.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.17.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.17.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.17.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.17.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.18.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.18.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.18.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.18.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.19.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.19.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.19.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.19.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.20.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.20.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.20.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.20.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.21.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.21.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.21.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.21.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.22.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.22.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.22.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.22.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.23.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.23.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.23.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.23.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.24.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.24.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.24.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.24.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.25.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.25.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.25.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.25.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.26.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.26.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.26.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.26.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.27.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.27.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.27.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key blocks.27.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key t_embedding_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.0.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.0.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.0.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.0.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.1.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.1.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.1.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.1.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.2.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.2.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.2.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.2.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.3.self_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.3.self_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.3.cross_attn.q_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|WARNING|cosmos_transfer2/_src/imaginaire/utils/checkpointer.py:451:non_strict_load_model] Skipping key control_blocks.3.cross_attn.k_norm._extra_state introduced by TransformerEngine for FP8 in the checkpoint. +[06-21 23:14:41|CRITICAL|cosmos_transfer2/_src/interactive/methods/cosmos2_interactive_model.py:409:load_state_dict] [RANK 0] _IncompatibleKeys(missing_keys=[], unexpected_keys=[], incorrect_shapes=[]) +[06-21 23:14:41|CRITICAL|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:819:load_state_dict] load fake score model in non-strict mode +[06-21 23:14:41|CRITICAL|cosmos_transfer2/_src/interactive/methods/distribution_matching/dmd2.py:820:load_state_dict] [RANK 0] _IncompatibleKeys(missing_keys=['accum_video_sample_counter', 'accum_image_sample_counter', 'accum_iteration', 'accum_train_in_hours', 'x_embedder.proj.1.weight', 'pos_embedder.seq', 'pos_embedder.dim_spatial_range', 'pos_embedder.dim_temporal_range', 't_embedder.1.linear_1.weight', 't_embedder.1.linear_2.weight', 'blocks.0._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.0._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.0._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.0._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.0._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.0._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.0._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.0._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.0._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.0._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.0._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.0._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.0._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.0._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.0._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.0._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.0._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.0._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.0._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.0._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.1._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.1._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.1._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.1._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.1._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.1._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.1._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.1._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.1._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.1._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.1._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.1._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.1._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.1._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.1._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.1._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.1._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.1._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.1._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.1._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.2._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.2._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.2._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.2._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.2._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.2._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.2._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.2._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.2._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.2._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.2._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.2._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.2._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.2._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.2._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.2._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.2._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.2._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.2._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.2._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.3._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.3._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.3._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.3._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.3._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.3._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.3._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.3._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.3._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.3._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.3._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.3._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.3._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.3._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.3._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.3._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.3._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.3._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.3._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.3._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.4._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.4._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.4._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.4._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.4._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.4._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.4._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.4._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.4._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.4._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.4._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.4._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.4._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.4._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.4._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.4._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.4._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.4._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.4._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.4._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.5._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.5._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.5._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.5._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.5._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.5._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.5._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.5._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.5._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.5._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.5._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.5._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.5._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.5._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.5._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.5._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.5._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.5._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.5._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.5._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.6._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.6._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.6._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.6._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.6._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.6._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.6._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.6._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.6._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.6._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.6._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.6._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.6._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.6._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.6._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.6._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.6._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.6._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.6._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.6._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.7._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.7._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.7._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.7._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.7._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.7._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.7._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.7._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.7._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.7._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.7._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.7._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.7._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.7._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.7._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.7._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.7._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.7._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.7._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.7._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.8._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.8._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.8._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.8._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.8._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.8._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.8._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.8._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.8._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.8._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.8._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.8._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.8._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.8._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.8._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.8._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.8._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.8._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.8._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.8._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.9._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.9._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.9._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.9._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.9._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.9._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.9._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.9._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.9._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.9._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.9._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.9._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.9._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.9._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.9._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.9._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.9._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.9._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.9._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.9._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.10._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.10._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.10._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.10._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.10._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.10._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.10._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.10._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.10._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.10._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.10._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.10._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.10._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.10._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.10._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.10._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.10._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.10._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.10._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.10._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.11._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.11._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.11._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.11._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.11._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.11._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.11._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.11._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.11._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.11._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.11._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.11._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.11._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.11._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.11._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.11._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.11._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.11._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.11._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.11._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.12._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.12._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.12._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.12._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.12._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.12._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.12._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.12._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.12._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.12._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.12._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.12._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.12._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.12._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.12._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.12._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.12._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.12._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.12._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.12._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.13._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.13._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.13._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.13._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.13._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.13._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.13._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.13._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.13._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.13._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.13._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.13._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.13._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.13._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.13._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.13._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.13._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.13._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.13._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.13._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.14._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.14._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.14._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.14._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.14._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.14._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.14._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.14._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.14._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.14._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.14._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.14._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.14._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.14._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.14._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.14._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.14._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.14._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.14._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.14._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.15._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.15._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.15._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.15._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.15._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.15._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.15._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.15._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.15._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.15._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.15._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.15._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.15._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.15._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.15._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.15._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.15._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.15._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.15._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.15._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.16._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.16._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.16._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.16._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.16._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.16._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.16._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.16._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.16._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.16._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.16._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.16._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.16._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.16._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.16._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.16._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.16._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.16._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.16._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.16._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.17._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.17._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.17._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.17._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.17._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.17._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.17._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.17._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.17._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.17._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.17._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.17._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.17._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.17._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.17._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.17._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.17._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.17._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.17._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.17._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.18._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.18._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.18._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.18._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.18._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.18._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.18._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.18._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.18._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.18._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.18._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.18._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.18._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.18._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.18._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.18._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.18._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.18._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.18._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.18._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.19._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.19._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.19._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.19._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.19._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.19._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.19._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.19._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.19._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.19._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.19._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.19._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.19._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.19._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.19._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.19._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.19._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.19._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.19._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.19._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.20._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.20._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.20._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.20._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.20._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.20._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.20._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.20._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.20._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.20._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.20._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.20._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.20._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.20._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.20._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.20._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.20._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.20._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.20._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.20._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.21._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.21._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.21._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.21._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.21._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.21._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.21._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.21._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.21._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.21._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.21._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.21._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.21._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.21._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.21._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.21._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.21._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.21._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.21._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.21._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.22._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.22._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.22._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.22._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.22._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.22._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.22._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.22._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.22._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.22._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.22._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.22._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.22._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.22._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.22._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.22._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.22._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.22._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.22._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.22._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.23._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.23._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.23._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.23._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.23._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.23._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.23._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.23._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.23._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.23._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.23._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.23._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.23._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.23._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.23._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.23._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.23._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.23._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.23._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.23._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.24._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.24._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.24._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.24._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.24._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.24._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.24._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.24._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.24._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.24._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.24._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.24._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.24._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.24._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.24._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.24._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.24._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.24._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.24._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.24._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.25._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.25._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.25._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.25._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.25._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.25._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.25._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.25._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.25._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.25._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.25._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.25._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.25._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.25._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.25._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.25._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.25._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.25._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.25._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.25._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.26._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.26._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.26._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.26._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.26._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.26._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.26._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.26._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.26._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.26._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.26._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.26._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.26._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.26._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.26._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.26._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.26._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.26._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.26._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.26._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'blocks.27._checkpoint_wrapped_module.self_attn.q_proj.weight', 'blocks.27._checkpoint_wrapped_module.self_attn.q_norm.weight', 'blocks.27._checkpoint_wrapped_module.self_attn.k_proj.weight', 'blocks.27._checkpoint_wrapped_module.self_attn.k_norm.weight', 'blocks.27._checkpoint_wrapped_module.self_attn.v_proj.weight', 'blocks.27._checkpoint_wrapped_module.self_attn.output_proj.weight', 'blocks.27._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'blocks.27._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'blocks.27._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'blocks.27._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'blocks.27._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'blocks.27._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'blocks.27._checkpoint_wrapped_module.mlp.layer1.weight', 'blocks.27._checkpoint_wrapped_module.mlp.layer2.weight', 'blocks.27._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'blocks.27._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'blocks.27._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'blocks.27._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'blocks.27._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'blocks.27._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'final_layer._checkpoint_wrapped_module._checkpoint_wrapped_module._checkpoint_wrapped_module.linear.weight', 'final_layer._checkpoint_wrapped_module._checkpoint_wrapped_module._checkpoint_wrapped_module.adaln_modulation.1.weight', 'final_layer._checkpoint_wrapped_module._checkpoint_wrapped_module._checkpoint_wrapped_module.adaln_modulation.2.weight', 't_embedding_norm.weight', 'crossattn_proj.0.weight', 'crossattn_proj.0.bias', 'control_embedder.proj.1.weight', 'control_blocks.0._checkpoint_wrapped_module.self_attn.q_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.self_attn.q_norm.weight', 'control_blocks.0._checkpoint_wrapped_module.self_attn.k_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.self_attn.k_norm.weight', 'control_blocks.0._checkpoint_wrapped_module.self_attn.v_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.self_attn.output_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'control_blocks.0._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'control_blocks.0._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.mlp.layer1.weight', 'control_blocks.0._checkpoint_wrapped_module.mlp.layer2.weight', 'control_blocks.0._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'control_blocks.0._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'control_blocks.0._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'control_blocks.0._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'control_blocks.0._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'control_blocks.0._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'control_blocks.0._checkpoint_wrapped_module.before_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.before_proj.bias', 'control_blocks.0._checkpoint_wrapped_module.after_proj.weight', 'control_blocks.0._checkpoint_wrapped_module.after_proj.bias', 'control_blocks.1._checkpoint_wrapped_module.self_attn.q_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.self_attn.q_norm.weight', 'control_blocks.1._checkpoint_wrapped_module.self_attn.k_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.self_attn.k_norm.weight', 'control_blocks.1._checkpoint_wrapped_module.self_attn.v_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.self_attn.output_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'control_blocks.1._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'control_blocks.1._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.mlp.layer1.weight', 'control_blocks.1._checkpoint_wrapped_module.mlp.layer2.weight', 'control_blocks.1._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'control_blocks.1._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'control_blocks.1._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'control_blocks.1._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'control_blocks.1._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'control_blocks.1._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'control_blocks.1._checkpoint_wrapped_module.after_proj.weight', 'control_blocks.1._checkpoint_wrapped_module.after_proj.bias', 'control_blocks.2._checkpoint_wrapped_module.self_attn.q_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.self_attn.q_norm.weight', 'control_blocks.2._checkpoint_wrapped_module.self_attn.k_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.self_attn.k_norm.weight', 'control_blocks.2._checkpoint_wrapped_module.self_attn.v_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.self_attn.output_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'control_blocks.2._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'control_blocks.2._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.mlp.layer1.weight', 'control_blocks.2._checkpoint_wrapped_module.mlp.layer2.weight', 'control_blocks.2._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'control_blocks.2._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'control_blocks.2._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'control_blocks.2._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'control_blocks.2._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'control_blocks.2._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'control_blocks.2._checkpoint_wrapped_module.after_proj.weight', 'control_blocks.2._checkpoint_wrapped_module.after_proj.bias', 'control_blocks.3._checkpoint_wrapped_module.self_attn.q_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.self_attn.q_norm.weight', 'control_blocks.3._checkpoint_wrapped_module.self_attn.k_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.self_attn.k_norm.weight', 'control_blocks.3._checkpoint_wrapped_module.self_attn.v_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.self_attn.output_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.cross_attn.q_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.cross_attn.q_norm.weight', 'control_blocks.3._checkpoint_wrapped_module.cross_attn.k_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.cross_attn.k_norm.weight', 'control_blocks.3._checkpoint_wrapped_module.cross_attn.v_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.cross_attn.output_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.mlp.layer1.weight', 'control_blocks.3._checkpoint_wrapped_module.mlp.layer2.weight', 'control_blocks.3._checkpoint_wrapped_module.adaln_modulation_self_attn.1.weight', 'control_blocks.3._checkpoint_wrapped_module.adaln_modulation_self_attn.2.weight', 'control_blocks.3._checkpoint_wrapped_module.adaln_modulation_cross_attn.1.weight', 'control_blocks.3._checkpoint_wrapped_module.adaln_modulation_cross_attn.2.weight', 'control_blocks.3._checkpoint_wrapped_module.adaln_modulation_mlp.1.weight', 'control_blocks.3._checkpoint_wrapped_module.adaln_modulation_mlp.2.weight', 'control_blocks.3._checkpoint_wrapped_module.after_proj.weight', 'control_blocks.3._checkpoint_wrapped_module.after_proj.bias'], unexpected_keys=[], incorrect_shapes=[]) +[06-21 23:14:41|INFO|cosmos_transfer2/inference.py:139:__init__] Setting net_fake_score to None for distilled model inference +[06-21 23:14:41|INFO|cosmos_transfer2/_src/transfer2/inference/utils.py:1015:compile_tokenizer_if_enabled] Tokenizer compilation disabled +[06-21 23:14:41|INFO|cosmos_transfer2/inference.py:146:__init__] Found 1 hint keys across all samples +[06-21 23:14:42|INFO|cosmos_transfer2/inference.py:156:__init__] Saved config to /scratch/ssaha/cosmos-transfer2.5/outputs/edge_threshold_grid/rainy/weather/distilled_step4_bg0.1_rm1/config.yaml +[06-21 23:14:42|INFO|cosmos_transfer2/inference.py:163:generate] Generating 1 samples: ['rainy_weather_distilled_step4_bg0.1_rm1'] +[06-21 23:14:42|INFO|cosmos_transfer2/inference.py:167:generate] [1/1] Processing sample rainy_weather_distilled_step4_bg0.1_rm1 +[06-21 23:14:42|DEBUG|cosmos_transfer2/inference.py:186:_generate_sample] InferenceArguments(name='rainy_weather_distilled_step4_bg0.1_rm1' prompt_path=None prompt='The video shows an urban intersection captured from a vehicle-mounted camera during heavy rain. A flat overcast grey sky hangs low over the scene, the asphalt is dark, wet and mirror-like with rippling puddles and reflections of the surroundings, rain streaks the air and droplets bead on surfaces. The painted lane markings, stop lines and pedestrian crosswalks show through the glossy wet road. Multiple vehicles move through the junction with headlights on, low-rise buildings, sidewalks and street lamps line the streets, and the camera follows the moving ego vehicle smoothly. Photorealistic, moody overcast lighting, detail.' negative_prompt='The video captures a game playing, with bad crappy graphics and cartoonish frames. It represents a recording of old outdated games. The lighting looks very fake. The textures are very raw and basic. The geometries are very primitive. The images are very pixelated and of poor CG quality. There are many subtitles in the footage. Overall, the video is unrealistic at all. Bright sunlight, blue sky, dry road, hard crisp shadows.' seed=2025 guidance=3 video_path=PosixPath('/scratch/ssaha/cosmos-transfer2.5/SceneDriveBench/output/priority_from_right/run_000_rainy/videos/rgb_front.mp4') max_frames=None context_frame_index=None image_context_path=None num_conditional_frames=1 resolution='720' sigma_max=None num_video_frames_per_chunk=93 num_steps=4 show_control_condition=False show_input=False keep_input_resolution=True edge=EdgeConfig(control_path=PosixPath('/scratch/ssaha/cosmos-transfer2.5/outputs/edge_threshold_grid/_edges/rainy/edge_bg0.1_rm1.mp4'), control_weight=1.0, mask_path=None, mask_prompt=None, preset_edge_threshold='medium') depth=None vis=None seg=None guided_generation_mask=None guided_generation_step_threshold=25 guided_generation_foreground_labels=None) +[06-21 23:14:42|INFO|cosmos_transfer2/inference.py:208:_generate_sample] Saved arguments to /scratch/ssaha/cosmos-transfer2.5/outputs/edge_threshold_grid/rainy/weather/distilled_step4_bg0.1_rm1/rainy_weather_distilled_step4_bg0.1_rm1.json +[06-21 23:14:42|WARNING|cosmos_transfer2/inference.py:239:_generate_sample] Guardrail checks on prompt are disabled +[06-21 23:14:42|INFO|cosmos_transfer2/inference.py:242:_generate_sample] Processing the following paths: {'edge': '/scratch/ssaha/cosmos-transfer2.5/outputs/edge_threshold_grid/_edges/rainy/edge_bg0.1_rm1.mp4', 'edge_mask': None, 'edge_mask_prompt': None} +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:472:generate_img2world] Loading input video... +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/utils.py:204:read_video_or_image_into_frames_BCTHW] Reading video from /scratch/ssaha/cosmos-transfer2.5/SceneDriveBench/output/priority_from_right/run_000_rainy/videos/rgb_front.mp4 +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/utils.py:243:read_video_or_image_into_frames_BCTHW] Loaded input tensor with shape (1, 3, 120, 720, 1280) value 0, 255 +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:492:generate_img2world] Computing prompt text embeddings... +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:513:generate_img2world] Processing image context if available... +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/utils.py:330:read_and_process_image_context] No image context provided. +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:525:generate_img2world] Loading control inputs... +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/utils.py:204:read_video_or_image_into_frames_BCTHW] Reading video from /scratch/ssaha/cosmos-transfer2.5/outputs/edge_threshold_grid/_edges/rainy/edge_bg0.1_rm1.mp4 +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/utils.py:243:read_video_or_image_into_frames_BCTHW] Loaded input tensor with shape (1, 3, 120, 720, 1280) value 0, 255 +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:554:generate_img2world] Generating chunk 1/2 +[06-21 23:14:42|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:640:generate_img2world] Seed: 584845 +[06-21 23:15:20|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:554:generate_img2world] Generating chunk 2/2 +[06-21 23:15:20|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:640:generate_img2world] Seed: 643774 +[06-21 23:15:48|INFO|cosmos_transfer2/_src/transfer2/inference/inference_pipeline.py:736:generate_img2world] Average time per chunk: 31.66861143137794 +[06-21 23:15:48|WARNING|cosmos_transfer2/inference.py:299:_generate_sample] Generated output has 120 frames (> 93). The distilled Transfer 2.5 model is not trained to support auto-regressive generation +[06-21 23:15:48|DEBUG|cosmos_transfer2/_src/imaginaire/utils/easy_io/handlers/imageio_video_handler.py:163:dump_to_fileobj] mimsave_kwargs: {'fps': 10, 'quality': 5, 'macro_block_size': 1, 'ffmpeg_params': ['-s', '1280x720'], 'output_params': ['-f', 'mp4']} +[06-21 23:15:49|INFO|cosmos_transfer2/inference.py:311:_generate_sample] edge control video saved to /scratch/ssaha/cosmos-transfer2.5/outputs/edge_threshold_grid/rainy/weather/distilled_step4_bg0.1_rm1/rainy_weather_distilled_step4_bg0.1_rm1_control_edge.mp4 +[06-21 23:15:49|WARNING|cosmos_transfer2/inference.py:335:_generate_sample] Guardrail checks on video are disabled +[06-21 23:15:49|DEBUG|cosmos_transfer2/_src/imaginaire/utils/easy_io/handlers/imageio_video_handler.py:163:dump_to_fileobj] mimsave_kwargs: {'fps': 10, 'quality': 5, 'macro_block_size': 1, 'ffmpeg_params': ['-s', '1280x720'], 'output_params': ['-f', 'mp4']} +[06-21 23:15:50|SUCCESS|cosmos_transfer2/inference.py:343:_generate_sample] Generated video saved to /scratch/ssaha/cosmos-transfer2.5/outputs/edge_threshold_grid/rainy/weather/distilled_step4_bg0.1_rm1/rainy_weather_distilled_step4_bg0.1_rm1.mp4