diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/action.npy b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/action.npy new file mode 100644 index 0000000000000000000000000000000000000000..8319741aafcf6b67c57cad5b1a87d8d32b52f96c --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/action.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85897aa9b515f6b65cc4a63fe9ba29617af04399581c900d42499852ee9f297c +size 15504 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/ijkl_action.npy b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/ijkl_action.npy new file mode 100644 index 0000000000000000000000000000000000000000..a6edd16763b8613ea356eb33519879de6c9ec3cd --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/ijkl_action.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dee2a3de69c50600ee9f46e3ffbf17356c45929f173b604ab1d0a9b8c0c4d156 +size 15504 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/image.jpg b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/image.jpg new file mode 100644 index 0000000000000000000000000000000000000000..c38238ccdb6bd3f7034c4f0d856b6b25c8f702e7 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/image.jpg @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c17c56aa08a20579c55307085d79b01587a500fd030fac600d808058daa2dccb +size 154078 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/intrinsics.npy b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/intrinsics.npy new file mode 100644 index 0000000000000000000000000000000000000000..c9b53827d5675394134b7c68b7e88f07010f0ac3 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/intrinsics.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0737afdfc05bdc313494110496d4cbb04e571875bdc0a6ffbc9a6d3d334dbb89 +size 15504 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/poses.npy b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/poses.npy new file mode 100644 index 0000000000000000000000000000000000000000..64291e83d54c53ff9f76f3b72697de31b0632d14 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/poses.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:795670d055bdfd9953fe5b015902170eba8e915e4417a0dcc62bfc1a5437d6cc +size 61632 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/prompt.txt b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/prompt.txt new file mode 100644 index 0000000000000000000000000000000000000000..516272f0f2bac4ba3b0b2fd902ab229689b98d25 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/prompt.txt @@ -0,0 +1 @@ +The video presents a soaring journey through a fantasy jungle. The wind whips past the rider's blue hands gripping the reins, causing the leather straps to vibrate. The ancient gothic castle approaches steadily, its stone details becoming clearer against the backdrop of floating islands and distant waterfalls. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/wasd_action.npy b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/wasd_action.npy new file mode 100644 index 0000000000000000000000000000000000000000..9b663da2250c5c6e3f9ad3ca39a9fda3440351e4 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/controls/example05_pose_ui_corrected/wasd_action.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2a1aecf1e3185eb2414e6b49e0e9a093c6ffadd84a552b158a12474afbd0bfa +size 15504 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/logs/example05_chunk01_seed6200.log b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/logs/example05_chunk01_seed6200.log new file mode 100644 index 0000000000000000000000000000000000000000..ca93c0b19676b77c65201f0e3a609f650349ef71 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/logs/example05_chunk01_seed6200.log @@ -0,0 +1,23 @@ +Command: torchrun --nproc_per_node 4 --master_port 37470 generate_fast.py --task i2v-A14B --size '480*832' --ckpt_dir models/lingbot-world-base-cam --image experiments/fast_chunk_ablation_20260528_pose_ui_corrected/controls/example05_pose_ui_corrected/image.jpg --action_path experiments/fast_chunk_ablation_20260528_pose_ui_corrected/controls/example05_pose_ui_corrected --dit_fsdp --t5_fsdp --ulysses_size 4 --frame_num 141 --base_seed 6200 --save_file experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200.mp4 --prompt 'A first-person cinematic flight through a lush fantasy jungle, blue hands gripping vibrating leather reins in the foreground as an ancient gothic castle grows clearer ahead, with floating islands and distant waterfalls in the background.' --chunk_size 1 +W0529 00:30:04.355000 2200098 /lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/miniconda3/envs/lingbot_world_repro/lib/python3.10/site-packages/torch/distributed/run.py:793] +W0529 00:30:04.355000 2200098 /lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/miniconda3/envs/lingbot_world_repro/lib/python3.10/site-packages/torch/distributed/run.py:793] ***************************************** +W0529 00:30:04.355000 2200098 /lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/miniconda3/envs/lingbot_world_repro/lib/python3.10/site-packages/torch/distributed/run.py:793] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. +W0529 00:30:04.355000 2200098 /lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/miniconda3/envs/lingbot_world_repro/lib/python3.10/site-packages/torch/distributed/run.py:793] ***************************************** +[2026-05-29 00:30:26,547] INFO: offload_model is not specified, set to False. +[2026-05-29 00:30:26,580] INFO: Generation job args: Namespace(task='i2v-A14B', size='480*832', frame_num=141, ckpt_dir='models/lingbot-world-base-cam', offload_model=False, ulysses_size=4, t5_fsdp=True, t5_cpu=False, dit_fsdp=True, save_file='experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200.mp4', prompt='A first-person cinematic flight through a lush fantasy jungle, blue hands gripping vibrating leather reins in the foreground as an ancient gothic castle grows clearer ahead, with floating islands and distant waterfalls in the background.', prompt_b=None, switch_frame=None, prompt_switch_case=None, use_prompt_extend=False, base_seed=6200, image='experiments/fast_chunk_ablation_20260528_pose_ui_corrected/controls/example05_pose_ui_corrected/image.jpg', action_path='experiments/fast_chunk_ablation_20260528_pose_ui_corrected/controls/example05_pose_ui_corrected', sample_shift=10.0, convert_model_dtype=False, local_attn_size=-1, sink_size=0, max_attention_size=None, chunk_size=1, save_dir='output') +[2026-05-29 00:30:26,581] INFO: Generation model config: {'__name__': 'Config: Wan I2V A14B', 't5_model': 'umt5_xxl', 't5_dtype': torch.bfloat16, 'text_len': 512, 'param_dtype': torch.bfloat16, 'num_train_timesteps': 1000, 'sample_fps': 16, 'sample_neg_prompt': '画面突变，色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走，镜头晃动，画面闪烁，模糊，噪点，水印，签名，文字，变形，扭曲，液化，不合逻辑的结构，卡顿，PPT幻灯片感，过暗，欠曝，低对比度，霓虹灯光感，过度锐化，3D渲染感，人物，行人，游客，身体，皮肤，肢体，面部特征，汽车，电线', 'frame_num': 81, 't5_checkpoint': 'models_t5_umt5-xxl-enc-bf16.pth', 't5_tokenizer': 'google/umt5-xxl', 'vae_checkpoint': 'Wan2.1_VAE.pth', 'vae_stride': (4, 8, 8), 'patch_size': (1, 2, 2), 'dim': 5120, 'ffn_dim': 13824, 'freq_dim': 256, 'num_heads': 40, 'num_layers': 40, 'window_size': (-1, -1), 'qk_norm': True, 'cross_attn_norm': True, 'eps': 1e-06, 'low_noise_checkpoint': 'low_noise_model', 'high_noise_checkpoint': 'high_noise_model', 'fast_noise_checkpoint': 'lingbot_world_fast', 'sample_shift': 10.0, 'sample_steps': 70, 'boundary': 0.947, 'sample_guide_scale': (5.0, 5.0)} +NCCL version 2.21.5+cuda12.4 +[2026-05-29 00:30:28,259] INFO: Input prompt: A first-person cinematic flight through a lush fantasy jungle, blue hands gripping vibrating leather reins in the foreground as an ancient gothic castle grows clearer ahead, with floating islands and distant waterfalls in the background. +[2026-05-29 00:30:28,317] INFO: Input image: experiments/fast_chunk_ablation_20260528_pose_ui_corrected/controls/example05_pose_ui_corrected/image.jpg +[2026-05-29 00:30:28,317] INFO: Creating WanI2VFast pipeline. +[2026-05-29 00:31:19,770] INFO: loading models/lingbot-world-base-cam/models_t5_umt5-xxl-enc-bf16.pth +[2026-05-29 00:31:45,191] INFO: loading models/lingbot-world-base-cam/Wan2.1_VAE.pth +[2026-05-29 00:31:46,600] INFO: Creating WanModelFast from models/lingbot-world-base-cam + Loading checkpoint shards: 0%| | 0/16 [00:00 00:00:08,812 +第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮 +革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。 +A first-person cinematic flight through a lush fantasy jungle, blue hands +gripping vibrating leather reins in the foreground as an ancient gothic castle +grows clearer ahead, with floating islands and distant waterfalls in the +background. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..b456d3196a9a49b836b35ac7a63eaf5c52f2c910 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:248bedc4794ff0370794c5127dc3c4495bd6b370d3696e608bb9518a448145a3 +size 12951727 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.ass b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.ass new file mode 100644 index 0000000000000000000000000000000000000000..66e10b2d5969c44d8e33f33ffbfc9fc96302bd05 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.ass @@ -0,0 +1,14 @@ +[Script Info] +ScriptType: v4.00+ +PlayResX: 832 +PlayResY: 464 +WrapStyle: 0 +ScaledBorderAndShadow: yes + +[V4+ Styles] +Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding +Style: Default,Arial Unicode MS,15,&H00FFFFFF,&H000000FF,&H00000000,&H98000000,0,0,0,0,100,100,0,0,3,2,0,2,30,30,24,1 + +[Events] +Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text +Dialogue: 0,0:00:00.00,0:00:08.81,Default,,0,0,0,,第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮\N革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。\NA first-person cinematic flight through a lush fantasy jungle, blue hands\Ngripping vibrating leather reins in the foreground as an ancient gothic castle\Ngrows clearer ahead, with floating islands and distant waterfalls in the\Nbackground. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..faf3b90ef5b1610662b90db416461e741682e015 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01ccdf3fc579c5749fc2e5a68542a0fe850604614ae87c87489f32b2533301d4 +size 4795234 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.srt b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.srt new file mode 100644 index 0000000000000000000000000000000000000000..4b6e34f9a4cad8b3f5b8ca0291cb86b7ec49e843 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.srt @@ -0,0 +1,8 @@ +1 +00:00:00,000 --> 00:00:08,812 +第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮 +革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。 +A first-person cinematic flight through a lush fantasy jungle, blue hands +gripping vibrating leather reins in the foreground as an ancient gothic castle +grows clearer ahead, with floating islands and distant waterfalls in the +background. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..fb664c04624f349899b992c44308f61306c5bafe --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:19d9e633735eee91209a66a1b56a675e5970832ac17b766f34a16f534a7a2b58 +size 14601355 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.ass b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.ass new file mode 100644 index 0000000000000000000000000000000000000000..66e10b2d5969c44d8e33f33ffbfc9fc96302bd05 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.ass @@ -0,0 +1,14 @@ +[Script Info] +ScriptType: v4.00+ +PlayResX: 832 +PlayResY: 464 +WrapStyle: 0 +ScaledBorderAndShadow: yes + +[V4+ Styles] +Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding +Style: Default,Arial Unicode MS,15,&H00FFFFFF,&H000000FF,&H00000000,&H98000000,0,0,0,0,100,100,0,0,3,2,0,2,30,30,24,1 + +[Events] +Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text +Dialogue: 0,0:00:00.00,0:00:08.81,Default,,0,0,0,,第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮\N革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。\NA first-person cinematic flight through a lush fantasy jungle, blue hands\Ngripping vibrating leather reins in the foreground as an ancient gothic castle\Ngrows clearer ahead, with floating islands and distant waterfalls in the\Nbackground. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..8142ab670ea3d9b536c27787abfad123ce28d028 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:edfc84685d1279de9724f839636249666caf48bd8cff141e271fec3d9e396029 +size 5797163 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.srt b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.srt new file mode 100644 index 0000000000000000000000000000000000000000..4b6e34f9a4cad8b3f5b8ca0291cb86b7ec49e843 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.srt @@ -0,0 +1,8 @@ +1 +00:00:00,000 --> 00:00:08,812 +第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮 +革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。 +A first-person cinematic flight through a lush fantasy jungle, blue hands +gripping vibrating leather reins in the foreground as an ancient gothic castle +grows clearer ahead, with floating islands and distant waterfalls in the +background. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..8a819ba491d3f4ff8b55647583bfbe76008e796a --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4973486fd11cf34153bdf1441c527de99b02d46660ca5ead6100128ada55557c +size 15540042 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.ass b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.ass new file mode 100644 index 0000000000000000000000000000000000000000..66e10b2d5969c44d8e33f33ffbfc9fc96302bd05 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.ass @@ -0,0 +1,14 @@ +[Script Info] +ScriptType: v4.00+ +PlayResX: 832 +PlayResY: 464 +WrapStyle: 0 +ScaledBorderAndShadow: yes + +[V4+ Styles] +Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding +Style: Default,Arial Unicode MS,15,&H00FFFFFF,&H000000FF,&H00000000,&H98000000,0,0,0,0,100,100,0,0,3,2,0,2,30,30,24,1 + +[Events] +Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text +Dialogue: 0,0:00:00.00,0:00:08.81,Default,,0,0,0,,第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮\N革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。\NA first-person cinematic flight through a lush fantasy jungle, blue hands\Ngripping vibrating leather reins in the foreground as an ancient gothic castle\Ngrows clearer ahead, with floating islands and distant waterfalls in the\Nbackground. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..cfc2b963e1d88f53739fb87d3a173bff2c4fd1b5 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c0806a76bbce768629e38ae531439ba0d16472dcf6cd8ab166b7e5503937c38c +size 6195788 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.srt b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.srt new file mode 100644 index 0000000000000000000000000000000000000000..4b6e34f9a4cad8b3f5b8ca0291cb86b7ec49e843 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.srt @@ -0,0 +1,8 @@ +1 +00:00:00,000 --> 00:00:08,812 +第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮 +革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。 +A first-person cinematic flight through a lush fantasy jungle, blue hands +gripping vibrating leather reins in the foreground as an ancient gothic castle +grows clearer ahead, with floating islands and distant waterfalls in the +background. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..99a8516763263c2e22b796948b4227f3083220c7 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a952bf1f79a9859a24238fa3f012e5b1cd94bb3fca325dc64e687fd2a0bcb1d4 +size 12878559 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.ass b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.ass new file mode 100644 index 0000000000000000000000000000000000000000..66e10b2d5969c44d8e33f33ffbfc9fc96302bd05 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.ass @@ -0,0 +1,14 @@ +[Script Info] +ScriptType: v4.00+ +PlayResX: 832 +PlayResY: 464 +WrapStyle: 0 +ScaledBorderAndShadow: yes + +[V4+ Styles] +Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding +Style: Default,Arial Unicode MS,15,&H00FFFFFF,&H000000FF,&H00000000,&H98000000,0,0,0,0,100,100,0,0,3,2,0,2,30,30,24,1 + +[Events] +Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text +Dialogue: 0,0:00:00.00,0:00:08.81,Default,,0,0,0,,第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮\N革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。\NA first-person cinematic flight through a lush fantasy jungle, blue hands\Ngripping vibrating leather reins in the foreground as an ancient gothic castle\Ngrows clearer ahead, with floating islands and distant waterfalls in the\Nbackground. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..52b8a4d25cbd1ea789dbe6d645affe483e102701 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:539a5e0276da8d2d502f82464eb8743ab4b9e219863d02e500004efe7d6eb8f8 +size 4771758 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.srt b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.srt new file mode 100644 index 0000000000000000000000000000000000000000..4b6e34f9a4cad8b3f5b8ca0291cb86b7ec49e843 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.srt @@ -0,0 +1,8 @@ +1 +00:00:00,000 --> 00:00:08,812 +第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮 +革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。 +A first-person cinematic flight through a lush fantasy jungle, blue hands +gripping vibrating leather reins in the foreground as an ancient gothic castle +grows clearer ahead, with floating islands and distant waterfalls in the +background. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..fd23f91829f26dc99149ecb5162d887f4b7cff10 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:51fe5e75c2c95f6e7494b97abb0d194045c0fd55a32d447e4153a16798920f6b +size 13584138 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.ass b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.ass new file mode 100644 index 0000000000000000000000000000000000000000..66e10b2d5969c44d8e33f33ffbfc9fc96302bd05 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.ass @@ -0,0 +1,14 @@ +[Script Info] +ScriptType: v4.00+ +PlayResX: 832 +PlayResY: 464 +WrapStyle: 0 +ScaledBorderAndShadow: yes + +[V4+ Styles] +Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding +Style: Default,Arial Unicode MS,15,&H00FFFFFF,&H000000FF,&H00000000,&H98000000,0,0,0,0,100,100,0,0,3,2,0,2,30,30,24,1 + +[Events] +Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text +Dialogue: 0,0:00:00.00,0:00:08.81,Default,,0,0,0,,第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮\N革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。\NA first-person cinematic flight through a lush fantasy jungle, blue hands\Ngripping vibrating leather reins in the foreground as an ancient gothic castle\Ngrows clearer ahead, with floating islands and distant waterfalls in the\Nbackground. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.mp4 b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.mp4 new file mode 100644 index 0000000000000000000000000000000000000000..486f30d4f75846193c4e231025b0603ce4499de1 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.mp4 @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:600b6091d18bee2d0dee93d4647d930d061730fae46c696664359c77c6e9b3c4 +size 5154629 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.srt b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.srt new file mode 100644 index 0000000000000000000000000000000000000000..4b6e34f9a4cad8b3f5b8ca0291cb86b7ec49e843 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.srt @@ -0,0 +1,8 @@ +1 +00:00:00,000 --> 00:00:08,812 +第一人称电影镜头飞越茂密的幻想丛林，前景中蓝色双手握着震动的皮 +革缰绳，前方古老哥特城堡逐渐清晰，背景有漂浮岛屿和远处瀑布。 +A first-person cinematic flight through a lush fantasy jungle, blue hands +gripping vibrating leather reins in the foreground as an ancient gothic castle +grows clearer ahead, with floating islands and distant waterfalls in the +background. diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/report.md b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/report.md new file mode 100644 index 0000000000000000000000000000000000000000..e8696fccaac05c39c9be41835e983baf78b5e033 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/report.md @@ -0,0 +1,19 @@ +# LingBot Fast Chunk Size Ablation + +- Description: Same-prompt LingBot fast inference ablation over latent-frame chunk_size, rerun with pose-derived corrected WASD/IJKL visualization controls. +- Prompt: A first-person cinematic flight through a lush fantasy jungle, blue hands gripping vibrating leather reins in the foreground as an ancient gothic castle grows clearer ahead, with floating islands and distant waterfalls in the background. +- Image: `experiments/fast_chunk_ablation_20260528_pose_ui_corrected/controls/example05_pose_ui_corrected/image.jpg` +- Camera/action path: `experiments/fast_chunk_ablation_20260528_pose_ui_corrected/controls/example05_pose_ui_corrected` +- Visualization controls: wasd_action.npy and ijkl_action.npy regenerated from the same poses.npy used for camera conditioning +- Seed: `6200` +- Frame num / FPS: `141` / `16` +- Effective duration: `8.8125` seconds + +| chunk_size | chunk seconds | status | wall time s | peak single GPU MiB | mean GPU util % | output | +| ---: | ---: | --- | ---: | ---: | ---: | --- | +| 1 | 0.25 | ok | 624.4 | 42219 | 12.039 | `experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200.mp4` | +| 2 | 0.5 | ok | 584.3 | 42193 | 10.703 | `experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200.mp4` | +| 3 | 0.75 | ok | 508.1 | 41379 | 11.483 | `experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200.mp4` | +| 4 | 1.0 | ok | 181.0 | 42135 | 32.871 | `experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200.mp4` | +| 6 | 1.5 | ok | 171.1 | 42227 | 31.723 | `experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200.mp4` | +| 12 | 3.0 | ok | 164.8 | 43305 | 35.14 | `experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200.mp4` | diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_fast_chunk_ablation_4gpu.sbatch b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_fast_chunk_ablation_4gpu.sbatch new file mode 100644 index 0000000000000000000000000000000000000000..3cf6c79ab2bb4ad6eb8606f4efa4560f84b27e96 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_fast_chunk_ablation_4gpu.sbatch @@ -0,0 +1,35 @@ +#!/bin/bash +#SBATCH -A nvr_elm_llm +#SBATCH -p batch_block1,batch_block2,batch_block3,batch_block4 +#SBATCH -t 04:00:00 +#SBATCH -N 1 +#SBATCH --gpus-per-node=4 +#SBATCH -J nvr_elm_llm:fast-chunk-corrected +#SBATCH --array=0-5%2 +#SBATCH -o experiments/fast_chunk_ablation_20260528_pose_ui_corrected/slurm/%x_%A_%a.out +#SBATCH -e experiments/fast_chunk_ablation_20260528_pose_ui_corrected/slurm/%x_%A_%a.err + +set -euo pipefail + +REPO_ROOT="/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot" +CONDA_SH="/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/etc/profile.d/conda.sh" +LOCAL_ROOT="${LOCAL_ROOT:-experiments/fast_chunk_ablation_20260528_pose_ui_corrected}" +NPROC="${NPROC_PER_NODE:-4}" +MASTER_PORT=$((20000 + (SLURM_ARRAY_JOB_ID % 20000) + SLURM_ARRAY_TASK_ID * 100)) + +cd "${REPO_ROOT}" +mkdir -p "${LOCAL_ROOT}/slurm" +source "${CONDA_SH}" +conda activate lingbot_world_repro + +python "${LOCAL_ROOT}/run_fast_chunk_case.py" \ + --manifest "${LOCAL_ROOT}/manifest.json" \ + --case-index "${SLURM_ARRAY_TASK_ID}" \ + --output-root "${LOCAL_ROOT}/outputs" \ + --metrics-dir "${LOCAL_ROOT}/metrics" \ + --log-dir "${LOCAL_ROOT}/logs" \ + --ckpt-dir models/lingbot-world-base-cam \ + --nproc-per-node "${NPROC}" \ + --master-port "${MASTER_PORT}" \ + --monitor-interval "${MONITOR_INTERVAL:-1.0}" \ + --resume diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_fast_chunk_case.py b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_fast_chunk_case.py new file mode 100644 index 0000000000000000000000000000000000000000..e04e837ae768524287eec6984e9c666a32ce2ab7 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_fast_chunk_case.py @@ -0,0 +1,319 @@ +#!/usr/bin/env python3 +from __future__ import annotations + +import argparse +import csv +import json +import os +import shlex +import subprocess +import sys +import threading +import time +from pathlib import Path + + +OOM_MARKERS = [ + "CUDA out of memory", + "torch.OutOfMemoryError", + "out of memory", + "CUBLAS_STATUS_ALLOC_FAILED", +] + + +def load_manifest(path: Path) -> dict: + return json.loads(path.read_text(encoding="utf-8")) + + +def query_gpu(): + cmd = [ + "nvidia-smi", + "--query-gpu=index,utilization.gpu,memory.used", + "--format=csv,noheader,nounits", + ] + try: + result = subprocess.run( + cmd, + check=True, + text=True, + stdout=subprocess.PIPE, + stderr=subprocess.PIPE, + ) + except (FileNotFoundError, subprocess.CalledProcessError): + return [] + + rows = [] + for line in result.stdout.splitlines(): + parts = [part.strip() for part in line.split(",")] + if len(parts) != 3: + continue + try: + rows.append( + { + "index": int(parts[0]), + "utilization_gpu_percent": int(parts[1]), + "memory_used_mib": int(parts[2]), + } + ) + except ValueError: + continue + return rows + + +def monitor_gpu(stop_event: threading.Event, samples: list[dict], interval: float) -> None: + while not stop_event.is_set(): + readings = query_gpu() + if readings: + samples.append({"time": time.time(), "gpus": readings}) + stop_event.wait(interval) + + +def summarize_gpu(samples: list[dict]) -> dict: + if not samples: + return { + "peak_total_memory_used_mib": None, + "peak_single_gpu_memory_used_mib": None, + "peak_by_gpu_mib": {}, + "mean_gpu_utilization_percent": None, + "mean_nonzero_gpu_count": None, + } + + peak_total = 0 + peak_single = 0 + peak_by_gpu: dict[str, int] = {} + util_values = [] + nonzero_counts = [] + for sample in samples: + total = 0 + nonzero = 0 + for gpu in sample["gpus"]: + idx = str(gpu["index"]) + used = gpu["memory_used_mib"] + util = gpu["utilization_gpu_percent"] + total += used + peak_single = max(peak_single, used) + peak_by_gpu[idx] = max(peak_by_gpu.get(idx, 0), used) + util_values.append(util) + if util > 0: + nonzero += 1 + nonzero_counts.append(nonzero) + peak_total = max(peak_total, total) + return { + "peak_total_memory_used_mib": peak_total, + "peak_single_gpu_memory_used_mib": peak_single, + "peak_by_gpu_mib": peak_by_gpu, + "mean_gpu_utilization_percent": round(sum(util_values) / len(util_values), 3) + if util_values + else None, + "mean_nonzero_gpu_count": round(sum(nonzero_counts) / len(nonzero_counts), 3) + if nonzero_counts + else None, + } + + +def detect_oom(log_file: Path) -> bool: + if not log_file.exists(): + return False + text = log_file.read_text(encoding="utf-8", errors="ignore") + return any(marker in text for marker in OOM_MARKERS) + + +def build_command(args: argparse.Namespace, manifest: dict, case: dict, output_file: Path) -> list[str]: + return [ + "torchrun", + "--nproc_per_node", + str(args.nproc_per_node), + "--master_port", + str(args.master_port), + "generate_fast.py", + "--task", + "i2v-A14B", + "--size", + manifest["size"], + "--ckpt_dir", + args.ckpt_dir, + "--image", + manifest["image"], + "--action_path", + manifest["action_path"], + "--dit_fsdp", + "--t5_fsdp", + "--ulysses_size", + str(args.nproc_per_node), + "--frame_num", + str(manifest["frame_num"]), + "--base_seed", + str(manifest["seed"]), + "--save_file", + str(output_file), + "--prompt", + manifest["prompt_en"], + "--chunk_size", + str(case["chunk_size"]), + ] + + +def write_summary_csv(path: Path, rows: list[dict]) -> None: + if not rows: + return + path.parent.mkdir(parents=True, exist_ok=True) + fieldnames = list(rows[0].keys()) + with path.open("w", encoding="utf-8", newline="") as handle: + writer = csv.DictWriter(handle, fieldnames=fieldnames) + writer.writeheader() + writer.writerows(rows) + + +def main() -> int: + parser = argparse.ArgumentParser() + parser.add_argument("--manifest", type=Path, default=Path("experiments/fast_chunk_ablation_20260528_pose_ui_corrected/manifest.json")) + parser.add_argument("--case-index", type=int, default=None) + parser.add_argument("--case-id", default=None) + parser.add_argument("--ckpt-dir", default="models/lingbot-world-base-cam") + parser.add_argument("--output-root", type=Path, default=Path("experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs")) + parser.add_argument("--metrics-dir", type=Path, default=Path("experiments/fast_chunk_ablation_20260528_pose_ui_corrected/metrics")) + parser.add_argument("--log-dir", type=Path, default=Path("experiments/fast_chunk_ablation_20260528_pose_ui_corrected/logs")) + parser.add_argument("--nproc-per-node", type=int, default=4) + parser.add_argument("--master-port", type=int, default=None) + parser.add_argument("--monitor-interval", type=float, default=1.0) + parser.add_argument("--resume", action="store_true", default=False) + args = parser.parse_args() + + if args.master_port is None: + job_id = int(os.environ.get("SLURM_JOB_ID", os.getpid())) + task_id = int(os.environ.get("SLURM_ARRAY_TASK_ID", "0")) + args.master_port = 17000 + ((job_id * 97 + task_id * 7919 + os.getpid()) % 45000) + + manifest = load_manifest(args.manifest) + cases = manifest["cases"] + if args.case_index is not None: + case = cases[args.case_index] + else: + matches = [item for item in cases if item["case_id"] == args.case_id] + if len(matches) != 1: + raise SystemExit(f"Expected one matching case, found {len(matches)}") + case = matches[0] + + output_dir = args.output_root / f"chunk_size_{case['chunk_size']:02d}" + output_dir.mkdir(parents=True, exist_ok=True) + args.metrics_dir.mkdir(parents=True, exist_ok=True) + args.log_dir.mkdir(parents=True, exist_ok=True) + output_file = output_dir / f"fast_chunk_ablation_{case['case_id']}.mp4" + metrics_file = args.metrics_dir / f"{case['case_id']}.json" + log_file = args.log_dir / f"{case['case_id']}.log" + + if args.resume and metrics_file.exists() and output_file.exists(): + old = json.loads(metrics_file.read_text(encoding="utf-8")) + if old.get("status") == "ok" and old.get("output_exists"): + print(f"Case already completed: {output_file}") + return 0 + + cmd = build_command(args, manifest, case, output_file) + env = os.environ.copy() + env.setdefault("PYTHONUNBUFFERED", "1") + env.setdefault("TOKENIZERS_PARALLELISM", "false") + env.setdefault("NCCL_DEBUG", "WARN") + + samples: list[dict] = [] + stop_event = threading.Event() + monitor = threading.Thread( + target=monitor_gpu, + args=(stop_event, samples, args.monitor_interval), + daemon=True, + ) + + start = time.time() + monitor.start() + with log_file.open("w", encoding="utf-8") as log: + log.write("Command: " + " ".join(shlex.quote(part) for part in cmd) + "\n") + log.flush() + proc = subprocess.run( + cmd, + cwd=Path.cwd(), + env=env, + stdout=log, + stderr=subprocess.STDOUT, + text=True, + ) + stop_event.set() + monitor.join(timeout=5) + end = time.time() + + output_exists = output_file.exists() + status = "ok" if proc.returncode == 0 and output_exists else "failed" + if status == "failed" and detect_oom(log_file): + status = "oom" + summary = { + "case_id": case["case_id"], + "chunk_size": case["chunk_size"], + "chunk_frame_span": case["chunk_frame_span"], + "chunk_seconds": case["chunk_seconds"], + "frame_num": manifest["frame_num"], + "fps": manifest["fps"], + "effective_duration_seconds": manifest["effective_duration_seconds"], + "seed": manifest["seed"], + "status": status, + "returncode": proc.returncode, + "wall_time_seconds": round(end - start, 3), + "output_file": str(output_file), + "output_exists": output_exists, + "output_size_bytes": output_file.stat().st_size if output_exists else 0, + "log_file": str(log_file), + "metrics_file": str(metrics_file), + "nproc_per_node": args.nproc_per_node, + } + metrics = { + "case": case, + "manifest_settings": { + key: manifest[key] + for key in [ + "image", + "action_path", + "prompt_en", + "prompt_zh", + "fps", + "frame_num", + "effective_duration_seconds", + "seed", + "size", + ] + }, + "command": cmd, + "status": status, + "returncode": proc.returncode, + "start_time_unix": start, + "end_time_unix": end, + "wall_time_seconds": end - start, + "output_file": str(output_file), + "output_exists": output_exists, + "output_size_bytes": output_file.stat().st_size if output_exists else 0, + "summary": summary, + "gpu": summarize_gpu(samples), + "gpu_sample_count": len(samples), + } + metrics_file.write_text(json.dumps(metrics, indent=2, ensure_ascii=False) + "\n", encoding="utf-8") + + rows = [] + for path in sorted(args.metrics_dir.glob("*.json")): + try: + data = json.loads(path.read_text(encoding="utf-8")) + except Exception: + continue + row = dict(data.get("summary", {})) + gpu = data.get("gpu", {}) + row["peak_single_gpu_memory_used_mib"] = gpu.get("peak_single_gpu_memory_used_mib") + row["peak_total_memory_used_mib"] = gpu.get("peak_total_memory_used_mib") + row["mean_gpu_utilization_percent"] = gpu.get("mean_gpu_utilization_percent") + row["mean_nonzero_gpu_count"] = gpu.get("mean_nonzero_gpu_count") + rows.append(row) + write_summary_csv(args.manifest.parent / "run_manifest.csv", rows) + + if status != "ok": + print(f"Case {status}; see {log_file}", file=sys.stderr) + return proc.returncode if proc.returncode != 0 else 1 + print(f"Case completed: {output_file}") + return 0 + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_manifest.csv b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_manifest.csv new file mode 100644 index 0000000000000000000000000000000000000000..d57dcd9837db5e4a3adcb2a1700db5587b6e177c --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_manifest.csv @@ -0,0 +1,7 @@ +case_id,chunk_size,chunk_frame_span,chunk_seconds,frame_num,fps,effective_duration_seconds,seed,status,returncode,wall_time_seconds,peak_single_gpu_memory_used_mib,peak_total_memory_used_mib,mean_gpu_utilization_percent,mean_nonzero_gpu_count,output_file,output_size_bytes,nproc_per_node +example05_chunk01_seed6200,1,4,0.25,141,16,8.8125,6200,ok,0,624.449,42219,142748,12.039,0.573,experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200.mp4,12847550,4 +example05_chunk02_seed6200,2,8,0.5,141,16,8.8125,6200,ok,0,584.266,42193,142030,10.703,0.5,experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200.mp4,12951727,4 +example05_chunk03_seed6200,3,12,0.75,141,16,8.8125,6200,ok,0,508.122,41379,143140,11.483,0.555,experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200.mp4,14601355,4 +example05_chunk04_seed6200,4,16,1.0,141,16,8.8125,6200,ok,0,181.017,42135,144042,32.871,1.465,experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200.mp4,15540042,4 +example05_chunk06_seed6200,6,24,1.5,141,16,8.8125,6200,ok,0,171.114,42227,145994,31.723,1.523,experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200.mp4,12878559,4 +example05_chunk12_seed6200,12,48,3.0,141,16,8.8125,6200,ok,0,164.83,43305,152560,35.14,1.566,experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200.mp4,13584138,4 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_manifest.json b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..59405637ff8e94eea46f6343538715a581649b24 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/run_manifest.json @@ -0,0 +1,140 @@ +[ + { + "case_id": "example05_chunk01_seed6200", + "chunk_size": 1, + "chunk_frame_span": 4, + "chunk_seconds": 0.25, + "frame_num": 141, + "fps": 16, + "effective_duration_seconds": 8.8125, + "seed": 6200, + "status": "ok", + "returncode": 0, + "wall_time_seconds": 624.449, + "output_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200.mp4", + "output_exists": true, + "output_size_bytes": 12847550, + "log_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/logs/example05_chunk01_seed6200.log", + "metrics_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/metrics/example05_chunk01_seed6200.json", + "nproc_per_node": 4, + "peak_single_gpu_memory_used_mib": 42219, + "peak_total_memory_used_mib": 142748, + "mean_gpu_utilization_percent": 12.039, + "mean_nonzero_gpu_count": 0.573 + }, + { + "case_id": "example05_chunk02_seed6200", + "chunk_size": 2, + "chunk_frame_span": 8, + "chunk_seconds": 0.5, + "frame_num": 141, + "fps": 16, + "effective_duration_seconds": 8.8125, + "seed": 6200, + "status": "ok", + "returncode": 0, + "wall_time_seconds": 584.266, + "output_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200.mp4", + "output_exists": true, + "output_size_bytes": 12951727, + "log_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/logs/example05_chunk02_seed6200.log", + "metrics_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/metrics/example05_chunk02_seed6200.json", + "nproc_per_node": 4, + "peak_single_gpu_memory_used_mib": 42193, + "peak_total_memory_used_mib": 142030, + "mean_gpu_utilization_percent": 10.703, + "mean_nonzero_gpu_count": 0.5 + }, + { + "case_id": "example05_chunk03_seed6200", + "chunk_size": 3, + "chunk_frame_span": 12, + "chunk_seconds": 0.75, + "frame_num": 141, + "fps": 16, + "effective_duration_seconds": 8.8125, + "seed": 6200, + "status": "ok", + "returncode": 0, + "wall_time_seconds": 508.122, + "output_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200.mp4", + "output_exists": true, + "output_size_bytes": 14601355, + "log_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/logs/example05_chunk03_seed6200.log", + "metrics_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/metrics/example05_chunk03_seed6200.json", + "nproc_per_node": 4, + "peak_single_gpu_memory_used_mib": 41379, + "peak_total_memory_used_mib": 143140, + "mean_gpu_utilization_percent": 11.483, + "mean_nonzero_gpu_count": 0.555 + }, + { + "case_id": "example05_chunk04_seed6200", + "chunk_size": 4, + "chunk_frame_span": 16, + "chunk_seconds": 1.0, + "frame_num": 141, + "fps": 16, + "effective_duration_seconds": 8.8125, + "seed": 6200, + "status": "ok", + "returncode": 0, + "wall_time_seconds": 181.017, + "output_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200.mp4", + "output_exists": true, + "output_size_bytes": 15540042, + "log_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/logs/example05_chunk04_seed6200.log", + "metrics_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/metrics/example05_chunk04_seed6200.json", + "nproc_per_node": 4, + "peak_single_gpu_memory_used_mib": 42135, + "peak_total_memory_used_mib": 144042, + "mean_gpu_utilization_percent": 32.871, + "mean_nonzero_gpu_count": 1.465 + }, + { + "case_id": "example05_chunk06_seed6200", + "chunk_size": 6, + "chunk_frame_span": 24, + "chunk_seconds": 1.5, + "frame_num": 141, + "fps": 16, + "effective_duration_seconds": 8.8125, + "seed": 6200, + "status": "ok", + "returncode": 0, + "wall_time_seconds": 171.114, + "output_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200.mp4", + "output_exists": true, + "output_size_bytes": 12878559, + "log_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/logs/example05_chunk06_seed6200.log", + "metrics_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/metrics/example05_chunk06_seed6200.json", + "nproc_per_node": 4, + "peak_single_gpu_memory_used_mib": 42227, + "peak_total_memory_used_mib": 145994, + "mean_gpu_utilization_percent": 31.723, + "mean_nonzero_gpu_count": 1.523 + }, + { + "case_id": "example05_chunk12_seed6200", + "chunk_size": 12, + "chunk_frame_span": 48, + "chunk_seconds": 3.0, + "frame_num": 141, + "fps": 16, + "effective_duration_seconds": 8.8125, + "seed": 6200, + "status": "ok", + "returncode": 0, + "wall_time_seconds": 164.83, + "output_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200.mp4", + "output_exists": true, + "output_size_bytes": 13584138, + "log_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/logs/example05_chunk12_seed6200.log", + "metrics_file": "experiments/fast_chunk_ablation_20260528_pose_ui_corrected/metrics/example05_chunk12_seed6200.json", + "nproc_per_node": 4, + "peak_single_gpu_memory_used_mib": 43305, + "peak_total_memory_used_mib": 152560, + "mean_gpu_utilization_percent": 35.14, + "mean_nonzero_gpu_count": 1.566 + } +] diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_0.err b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_0.err new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_0.out b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_0.out new file mode 100644 index 0000000000000000000000000000000000000000..f0a311f789ba60eb87b381c6effab21eee588ba6 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_0.out @@ -0,0 +1 @@ +Case completed: experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200.mp4 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_1.err b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_1.err new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_1.out b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_1.out new file mode 100644 index 0000000000000000000000000000000000000000..8476aea388a546f437e81f48c954203e09c86f6d --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_1.out @@ -0,0 +1 @@ +Case completed: experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200.mp4 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_2.err b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_2.err new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_2.out b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_2.out new file mode 100644 index 0000000000000000000000000000000000000000..eb494878cb28d1019b9712aa8ce4d71ffb59a5c0 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_2.out @@ -0,0 +1 @@ +Case completed: experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200.mp4 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_3.err b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_3.err new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_3.out b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_3.out new file mode 100644 index 0000000000000000000000000000000000000000..93f76014ef8d35fee8e6fd7782935cdf89b84ac3 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_3.out @@ -0,0 +1 @@ +Case completed: experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200.mp4 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_4.err b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_4.err new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_4.out b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_4.out new file mode 100644 index 0000000000000000000000000000000000000000..404e7aa9acd88bab3a6586083fcb48b3cd261ce3 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_4.out @@ -0,0 +1 @@ +Case completed: experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200.mp4 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_5.err b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_5.err new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_5.out b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_5.out new file mode 100644 index 0000000000000000000000000000000000000000..8881450f6f995d3008b8a11e017be588508594de --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/slurm/nvr_elm_llm:fast-chunk-corrected_9937470_5.out @@ -0,0 +1 @@ +Case completed: experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200.mp4 diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/upload_20260529_005115.log b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/upload_20260529_005115.log new file mode 100644 index 0000000000000000000000000000000000000000..343325b045d012b0e44af26c4f38f1725f003490 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/upload_20260529_005115.log @@ -0,0 +1,1619 @@ +Wrote 6 chunk ablation rows +ffmpeg version 7.0.2-static https://johnvansickle.com/ffmpeg/ Copyright (c) 2000-2024 the FFmpeg developers + built with gcc 8 (Debian 8.3.0-6) + configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libdav1d --enable-libxvid --enable-libzvbi --enable-libzimg + libavutil 59. 8.100 / 59. 8.100 + libavcodec 61. 3.100 / 61. 3.100 + libavformat 61. 1.100 / 61. 1.100 + libavdevice 61. 1.100 / 61. 1.100 + libavfilter 10. 1.100 / 10. 1.100 + libswscale 8. 1.100 / 8. 1.100 + libswresample 5. 1.100 / 5. 1.100 + libpostproc 58. 1.100 / 58. 1.100 +Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Duration: 00:00:08.81, start: 0.000000, bitrate: 11663 kb/s + Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(progressive), 832x464, 11660 kb/s, 16 fps, 16 tbr, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 +[Parsed_ass_0 @ 0x68b5a80] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x68b5a80] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x68b5a80] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x68b5a80] Using font provider fontconfig +[Parsed_ass_0 @ 0x68b5a80] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200_subtitled.ass' (2 styles, 1 events) +Stream mapping: + Stream #0:0 -> #0:0 (h264 (native) -> h264 (libx264)) +Press [q] to stop, [?] for help +[Parsed_ass_0 @ 0x15554c001d80] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x15554c001d80] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x15554c001d80] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x15554c001d80] Using font provider fontconfig +[Parsed_ass_0 @ 0x15554c001d80] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200_subtitled.ass' (2 styles, 1 events) +[Parsed_ass_0 @ 0x15554c001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/dejavu/DejaVuSans.ttf, 0, DejaVuSans +[Parsed_ass_0 @ 0x15554c001d80] Glyph 0x7B2C not found, selecting one more font for (Arial Unicode MS, 400, 0) +[Parsed_ass_0 @ 0x15554c001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/droid/DroidSansFallbackFull.ttf, 0, DroidSansFallback +[libx264 @ 0x6887ac0] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 +[libx264 @ 0x6887ac0] profile High, level 3.0, 4:2:0, 8-bit +[libx264 @ 0x6887ac0] 264 - core 164 r3191 4613ac3 - H.264/MPEG-4 AVC codec - Copyleft 2003-2024 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=4 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=16 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=18.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00 +Output #0, mp4, to 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200_subtitled.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Stream #0:0(und): Video: h264 (avc1 / 0x31637661), yuv420p(progressive), 832x464, q=2-31, 16 fps, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 + Side data: + cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A +frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 4 fps=1.7 q=22.0 size= 0KiB time=00:00:00.12 bitrate= 3.1kbits/s speed=0.0543x frame= 6 fps=2.1 q=22.0 size= 0KiB time=00:00:00.18 bitrate= 2.0kbits/s speed=0.066x frame= 16 fps=4.7 q=22.0 size= 512KiB time=00:00:00.87 bitrate=4793.9kbits/s speed=0.259x frame= 28 fps=7.2 q=22.0 size= 1024KiB time=00:00:01.62 bitrate=5162.5kbits/s speed=0.419x frame= 36 fps=8.2 q=22.0 size= 1280KiB time=00:00:02.12 bitrate=4934.7kbits/s speed=0.486x frame= 46 fps=9.4 q=22.0 size= 1536KiB time=00:00:02.75 bitrate=4575.7kbits/s speed=0.564x frame= 56 fps= 10 q=22.0 size= 1792KiB time=00:00:03.37 bitrate=4349.8kbits/s speed=0.628x frame= 68 fps= 12 q=22.0 size= 2304KiB time=00:00:04.12 bitrate=4575.7kbits/s speed=0.702x frame= 80 fps= 13 q=22.0 size= 2560KiB time=00:00:04.87 bitrate=4301.9kbits/s speed=0.765x frame= 92 fps= 13 q=22.0 size= 2816KiB time=00:00:05.62 bitrate=4101.2kbits/s speed=0.818x frame= 108 fps= 15 q=22.0 size= 3328KiB time=00:00:06.62 bitrate=4115.2kbits/s speed=0.898x frame= 118 fps= 15 q=22.0 size= 3584KiB time=00:00:07.25 bitrate=4049.7kbits/s speed=0.92x frame= 128 fps= 15 q=22.0 size= 3840KiB time=00:00:07.87 bitrate=3994.6kbits/s speed=0.935x frame= 136 fps= 15 q=22.0 size= 4096KiB time=00:00:08.37 bitrate=4006.5kbits/s speed=0.939x [out#0/mp4 @ 0x6886080] video:4593KiB audio:0KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.055240% +frame= 141 fps= 15 q=-1.0 Lsize= 4595KiB time=00:00:08.68 bitrate=4333.3kbits/s speed=0.922x +[libx264 @ 0x6887ac0] frame I:1 Avg QP:20.90 size: 92741 +[libx264 @ 0x6887ac0] frame P:37 Avg QP:19.32 size: 55468 +[libx264 @ 0x6887ac0] frame B:103 Avg QP:22.73 size: 24829 +[libx264 @ 0x6887ac0] consecutive B-frames: 1.4% 2.8% 2.1% 93.6% +[libx264 @ 0x6887ac0] mb I I16..4: 0.9% 43.6% 55.6% +[libx264 @ 0x6887ac0] mb P I16..4: 0.5% 20.1% 15.7% P16..4: 17.1% 24.3% 16.5% 0.0% 0.0% skip: 6.0% +[libx264 @ 0x6887ac0] mb B I16..4: 0.1% 5.3% 6.0% B16..8: 26.6% 19.7% 10.2% direct: 8.8% skip:23.2% L0:40.5% L1:33.2% BI:26.2% +[libx264 @ 0x6887ac0] 8x8 transform intra:51.0% inter:39.2% +[libx264 @ 0x6887ac0] coded y,uvDC,uvAC intra: 97.3% 96.4% 69.9% inter: 51.3% 37.8% 3.5% +[libx264 @ 0x6887ac0] i16 v,h,dc,p: 9% 17% 4% 70% +[libx264 @ 0x6887ac0] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 13% 16% 12% 7% 10% 8% 13% 9% 13% +[libx264 @ 0x6887ac0] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 13% 16% 12% 8% 12% 8% 13% 8% 11% +[libx264 @ 0x6887ac0] i8c dc,h,v,p: 52% 24% 16% 8% +[libx264 @ 0x6887ac0] Weighted P-Frames: Y:10.8% UV:5.4% +[libx264 @ 0x6887ac0] ref P L0: 43.4% 27.9% 20.7% 7.4% 0.6% +[libx264 @ 0x6887ac0] ref B L0: 91.3% 7.1% 1.6% +[libx264 @ 0x6887ac0] ref B L1: 98.0% 2.0% +[libx264 @ 0x6887ac0] kb/s:4268.90 +ffmpeg version 7.0.2-static https://johnvansickle.com/ffmpeg/ Copyright (c) 2000-2024 the FFmpeg developers + built with gcc 8 (Debian 8.3.0-6) + configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libdav1d --enable-libxvid --enable-libzvbi --enable-libzimg + libavutil 59. 8.100 / 59. 8.100 + libavcodec 61. 3.100 / 61. 3.100 + libavformat 61. 1.100 / 61. 1.100 + libavdevice 61. 1.100 / 61. 1.100 + libavfilter 10. 1.100 / 10. 1.100 + libswscale 8. 1.100 / 8. 1.100 + libswresample 5. 1.100 / 5. 1.100 + libpostproc 58. 1.100 / 58. 1.100 +Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Duration: 00:00:08.81, start: 0.000000, bitrate: 11757 kb/s + Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(progressive), 832x464, 11755 kb/s, 16 fps, 16 tbr, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 +[Parsed_ass_0 @ 0x68b4fc0] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x68b4fc0] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x68b4fc0] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x68b4fc0] Using font provider fontconfig +[Parsed_ass_0 @ 0x68b4fc0] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.ass' (2 styles, 1 events) +Stream mapping: + Stream #0:0 -> #0:0 (h264 (native) -> h264 (libx264)) +Press [q] to stop, [?] for help +[Parsed_ass_0 @ 0x155544001d80] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x155544001d80] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x155544001d80] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x155544001d80] Using font provider fontconfig +[Parsed_ass_0 @ 0x155544001d80] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.ass' (2 styles, 1 events) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/dejavu/DejaVuSans.ttf, 0, DejaVuSans +[Parsed_ass_0 @ 0x155544001d80] Glyph 0x7B2C not found, selecting one more font for (Arial Unicode MS, 400, 0) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/droid/DroidSansFallbackFull.ttf, 0, DroidSansFallback +[libx264 @ 0x6886880] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 +[libx264 @ 0x6886880] profile High, level 3.0, 4:2:0, 8-bit +[libx264 @ 0x6886880] 264 - core 164 r3191 4613ac3 - H.264/MPEG-4 AVC codec - Copyleft 2003-2024 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=4 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=16 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=18.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00 +Output #0, mp4, to 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Stream #0:0(und): Video: h264 (avc1 / 0x31637661), yuv420p(progressive), 832x464, q=2-31, 16 fps, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 + Side data: + cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A +frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 2 fps=0.9 q=22.0 size= 0KiB time=00:00:00.00 bitrate=N/A speed= 0x frame= 5 fps=1.8 q=22.0 size= 0KiB time=00:00:00.18 bitrate= 2.0kbits/s speed=0.0679x frame= 13 fps=4.0 q=22.0 size= 512KiB time=00:00:00.68 bitrate=6101.4kbits/s speed=0.209x frame= 21 fps=5.5 q=22.0 size= 768KiB time=00:00:01.18 bitrate=5298.4kbits/s speed=0.312x frame= 30 fps=7.0 q=22.0 size= 1280KiB time=00:00:01.75 bitrate=5992.1kbits/s speed=0.406x frame= 41 fps=8.5 q=22.0 size= 1536KiB time=00:00:02.43 bitrate=5162.4kbits/s speed=0.507x frame= 50 fps=9.4 q=22.0 size= 2048KiB time=00:00:03.00 bitrate=5592.5kbits/s speed=0.565x frame= 61 fps= 10 q=22.0 size= 2304KiB time=00:00:03.68 bitrate=5118.6kbits/s speed=0.635x frame= 73 fps= 12 q=22.0 size= 2560KiB time=00:00:04.43 bitrate=4726.1kbits/s speed=0.703x frame= 85 fps= 12 q=22.0 size= 2816KiB time=00:00:05.18 bitrate=4447.0kbits/s speed=0.762x frame= 98 fps= 13 q=22.0 size= 3328KiB time=00:00:06.00 bitrate=4543.9kbits/s speed=0.821x frame= 114 fps= 15 q=22.0 size= 3584KiB time=00:00:07.00 bitrate=4194.4kbits/s speed=0.896x frame= 123 fps= 15 q=22.0 size= 3840KiB time=00:00:07.56 bitrate=4159.7kbits/s speed=0.91x frame= 132 fps= 15 q=22.0 size= 4096KiB time=00:00:08.12 bitrate=4129.8kbits/s speed=0.922x [out#0/mp4 @ 0x6886180] video:4680KiB audio:0KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.054375% +frame= 141 fps= 15 q=-1.0 Lsize= 4683KiB time=00:00:08.68 bitrate=4415.8kbits/s speed=0.939x +[libx264 @ 0x6886880] frame I:1 Avg QP:21.15 size: 89720 +[libx264 @ 0x6886880] frame P:36 Avg QP:19.18 size: 51180 +[libx264 @ 0x6886880] frame B:104 Avg QP:22.21 size: 27498 +[libx264 @ 0x6886880] consecutive B-frames: 1.4% 0.0% 2.1% 96.5% +[libx264 @ 0x6886880] mb I I16..4: 0.5% 46.4% 53.1% +[libx264 @ 0x6886880] mb P I16..4: 0.7% 20.0% 19.1% P16..4: 18.1% 22.2% 13.4% 0.0% 0.0% skip: 6.5% +[libx264 @ 0x6886880] mb B I16..4: 0.1% 5.1% 8.6% B16..8: 27.6% 17.1% 7.7% direct: 8.6% skip:25.1% L0:25.7% L1:47.8% BI:26.4% +[libx264 @ 0x6886880] 8x8 transform intra:43.7% inter:38.1% +[libx264 @ 0x6886880] coded y,uvDC,uvAC intra: 96.4% 96.4% 68.7% inter: 45.9% 34.5% 3.1% +[libx264 @ 0x6886880] i16 v,h,dc,p: 5% 24% 4% 67% +[libx264 @ 0x6886880] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 14% 15% 13% 7% 10% 9% 11% 9% 12% +[libx264 @ 0x6886880] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 14% 15% 13% 8% 11% 9% 11% 8% 11% +[libx264 @ 0x6886880] i8c dc,h,v,p: 53% 23% 17% 8% +[libx264 @ 0x6886880] Weighted P-Frames: Y:5.6% UV:2.8% +[libx264 @ 0x6886880] ref P L0: 37.9% 27.7% 22.4% 11.7% 0.3% +[libx264 @ 0x6886880] ref B L0: 88.7% 8.4% 2.8% +[libx264 @ 0x6886880] ref B L1: 97.8% 2.2% +[libx264 @ 0x6886880] kb/s:4350.13 +ffmpeg version 7.0.2-static https://johnvansickle.com/ffmpeg/ Copyright (c) 2000-2024 the FFmpeg developers + built with gcc 8 (Debian 8.3.0-6) + configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libdav1d --enable-libxvid --enable-libzvbi --enable-libzimg + libavutil 59. 8.100 / 59. 8.100 + libavcodec 61. 3.100 / 61. 3.100 + libavformat 61. 1.100 / 61. 1.100 + libavdevice 61. 1.100 / 61. 1.100 + libavfilter 10. 1.100 / 10. 1.100 + libswscale 8. 1.100 / 8. 1.100 + libswresample 5. 1.100 / 5. 1.100 + libpostproc 58. 1.100 / 58. 1.100 +Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Duration: 00:00:08.81, start: 0.000000, bitrate: 13255 kb/s + Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(progressive), 832x464, 13252 kb/s, 16 fps, 16 tbr, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 +[Parsed_ass_0 @ 0x68b5b00] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x68b5b00] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x68b5b00] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x68b5b00] Using font provider fontconfig +[Parsed_ass_0 @ 0x68b5b00] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.ass' (2 styles, 1 events) +Stream mapping: + Stream #0:0 -> #0:0 (h264 (native) -> h264 (libx264)) +Press [q] to stop, [?] for help +[Parsed_ass_0 @ 0x155544001d80] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x155544001d80] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x155544001d80] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x155544001d80] Using font provider fontconfig +[Parsed_ass_0 @ 0x155544001d80] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.ass' (2 styles, 1 events) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/dejavu/DejaVuSans.ttf, 0, DejaVuSans +[Parsed_ass_0 @ 0x155544001d80] Glyph 0x7B2C not found, selecting one more font for (Arial Unicode MS, 400, 0) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/droid/DroidSansFallbackFull.ttf, 0, DroidSansFallback +[libx264 @ 0x6887b00] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 +[libx264 @ 0x6887b00] profile High, level 3.0, 4:2:0, 8-bit +[libx264 @ 0x6887b00] 264 - core 164 r3191 4613ac3 - H.264/MPEG-4 AVC codec - Copyleft 2003-2024 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=4 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=16 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=18.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00 +Output #0, mp4, to 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Stream #0:0(und): Video: h264 (avc1 / 0x31637661), yuv420p(progressive), 832x464, q=2-31, 16 fps, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 + Side data: + cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A +frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 1 fps=0.7 q=22.0 size= 0KiB time=-00:00:00.06 bitrate=N/A speed=N/A frame= 2 fps=1.0 q=22.0 size= 0KiB time=00:00:00.00 bitrate=N/A speed= 0x frame= 6 fps=2.3 q=22.0 size= 256KiB time=00:00:00.25 bitrate=8390.1kbits/s speed=0.0955x frame= 14 fps=4.4 q=22.0 size= 512KiB time=00:00:00.75 bitrate=5592.9kbits/s speed=0.234x frame= 23 fps=6.2 q=22.0 size= 1024KiB time=00:00:01.31 bitrate=6391.6kbits/s speed=0.354x frame= 33 fps=7.7 q=22.0 size= 1536KiB time=00:00:01.93 bitrate=6494.6kbits/s speed=0.453x frame= 41 fps=8.6 q=22.0 size= 1792KiB time=00:00:02.43 bitrate=6022.7kbits/s speed=0.51x frame= 49 fps=9.3 q=22.0 size= 2048KiB time=00:00:02.93 bitrate=5711.5kbits/s speed=0.556x frame= 57 fps=9.9 q=22.0 size= 2560KiB time=00:00:03.43 bitrate=6100.9kbits/s speed=0.595x frame= 69 fps= 11 q=22.0 size= 2816KiB time=00:00:04.18 bitrate=5509.0kbits/s speed=0.667x frame= 81 fps= 12 q=22.0 size= 3328KiB time=00:00:04.93 bitrate=5521.7kbits/s speed=0.728x frame= 90 fps= 12 q=22.0 size= 3584KiB time=00:00:05.50 bitrate=5338.3kbits/s speed=0.756x frame= 102 fps= 13 q=22.0 size= 3840KiB time=00:00:06.25 bitrate=5033.2kbits/s speed=0.803x frame= 114 fps= 14 q=22.0 size= 4352KiB time=00:00:07.00 bitrate=5093.1kbits/s speed=0.845x frame= 123 fps= 14 q=22.0 size= 4608KiB time=00:00:07.56 bitrate=4991.6kbits/s speed=0.861x frame= 134 fps= 14 q=22.0 size= 5120KiB time=00:00:08.25 bitrate=5084.1kbits/s speed=0.889x [out#0/mp4 @ 0x68860c0] video:5659KiB audio:0KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.045388% +frame= 141 fps= 15 q=-1.0 Lsize= 5661KiB time=00:00:08.68 bitrate=5338.4kbits/s speed=0.894x +[libx264 @ 0x6887b00] frame I:1 Avg QP:20.89 size: 90353 +[libx264 @ 0x6887b00] frame P:35 Avg QP:19.58 size: 56840 +[libx264 @ 0x6887b00] frame B:105 Avg QP:22.17 size: 35372 +[libx264 @ 0x6887b00] consecutive B-frames: 0.7% 0.0% 0.0% 99.3% +[libx264 @ 0x6887b00] mb I I16..4: 0.9% 45.8% 53.4% +[libx264 @ 0x6887b00] mb P I16..4: 1.1% 24.7% 23.3% P16..4: 15.5% 17.9% 10.5% 0.0% 0.0% skip: 6.9% +[libx264 @ 0x6887b00] mb B I16..4: 0.1% 5.8% 10.9% B16..8: 27.3% 17.5% 8.2% direct: 8.1% skip:22.0% L0:23.9% L1:51.6% BI:24.5% +[libx264 @ 0x6887b00] 8x8 transform intra:42.4% inter:36.5% +[libx264 @ 0x6887b00] coded y,uvDC,uvAC intra: 96.7% 96.8% 75.2% inter: 47.4% 35.0% 3.4% +[libx264 @ 0x6887b00] i16 v,h,dc,p: 7% 21% 7% 66% +[libx264 @ 0x6887b00] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 15% 14% 15% 6% 10% 9% 11% 8% 11% +[libx264 @ 0x6887b00] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 14% 14% 14% 8% 11% 9% 11% 7% 11% +[libx264 @ 0x6887b00] i8c dc,h,v,p: 53% 22% 17% 7% +[libx264 @ 0x6887b00] Weighted P-Frames: Y:0.0% UV:0.0% +[libx264 @ 0x6887b00] ref P L0: 39.1% 24.0% 25.0% 11.9% +[libx264 @ 0x6887b00] ref B L0: 87.2% 9.5% 3.3% +[libx264 @ 0x6887b00] ref B L1: 97.7% 2.3% +[libx264 @ 0x6887b00] kb/s:5259.66 +ffmpeg version 7.0.2-static https://johnvansickle.com/ffmpeg/ Copyright (c) 2000-2024 the FFmpeg developers + built with gcc 8 (Debian 8.3.0-6) + configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libdav1d --enable-libxvid --enable-libzvbi --enable-libzimg + libavutil 59. 8.100 / 59. 8.100 + libavcodec 61. 3.100 / 61. 3.100 + libavformat 61. 1.100 / 61. 1.100 + libavdevice 61. 1.100 / 61. 1.100 + libavfilter 10. 1.100 / 10. 1.100 + libswscale 8. 1.100 / 8. 1.100 + libswresample 5. 1.100 / 5. 1.100 + libpostproc 58. 1.100 / 58. 1.100 +Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Duration: 00:00:08.81, start: 0.000000, bitrate: 14107 kb/s + Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(progressive), 832x464, 14104 kb/s, 16 fps, 16 tbr, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 +[Parsed_ass_0 @ 0x68b6f00] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x68b6f00] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x68b6f00] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x68b6f00] Using font provider fontconfig +[Parsed_ass_0 @ 0x68b6f00] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.ass' (2 styles, 1 events) +Stream mapping: + Stream #0:0 -> #0:0 (h264 (native) -> h264 (libx264)) +Press [q] to stop, [?] for help +[Parsed_ass_0 @ 0x155544001d80] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x155544001d80] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x155544001d80] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x155544001d80] Using font provider fontconfig +[Parsed_ass_0 @ 0x155544001d80] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.ass' (2 styles, 1 events) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/dejavu/DejaVuSans.ttf, 0, DejaVuSans +[Parsed_ass_0 @ 0x155544001d80] Glyph 0x7B2C not found, selecting one more font for (Arial Unicode MS, 400, 0) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/droid/DroidSansFallbackFull.ttf, 0, DroidSansFallback +[libx264 @ 0x6887dc0] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 +[libx264 @ 0x6887dc0] profile High, level 3.0, 4:2:0, 8-bit +[libx264 @ 0x6887dc0] 264 - core 164 r3191 4613ac3 - H.264/MPEG-4 AVC codec - Copyleft 2003-2024 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=4 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=16 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=18.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00 +Output #0, mp4, to 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Stream #0:0(und): Video: h264 (avc1 / 0x31637661), yuv420p(progressive), 832x464, q=2-31, 16 fps, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 + Side data: + cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A +frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 2 fps=1.0 q=22.0 size= 0KiB time=00:00:00.00 bitrate=N/A speed= 0x frame= 6 fps=2.3 q=22.0 size= 256KiB time=00:00:00.25 bitrate=8390.1kbits/s speed=0.0944x frame= 7 fps=2.2 q=22.0 size= 256KiB time=00:00:00.43 bitrate=4794.4kbits/s speed=0.137x frame= 16 fps=4.3 q=22.0 size= 512KiB time=00:00:00.87 bitrate=4793.9kbits/s speed=0.237x frame= 25 fps=6.0 q=22.0 size= 1024KiB time=00:00:01.43 bitrate=5835.8kbits/s speed=0.342x frame= 34 fps=7.2 q=22.0 size= 1280KiB time=00:00:02.00 bitrate=5243.1kbits/s speed=0.426x frame= 40 fps=7.7 q=22.0 size= 1536KiB time=00:00:02.37 bitrate=5298.2kbits/s speed=0.457x frame= 48 fps=8.4 q=22.0 size= 2048KiB time=00:00:02.87 bitrate=5835.7kbits/s speed=0.504x frame= 55 fps=8.9 q=22.0 size= 2304KiB time=00:00:03.31 bitrate=5698.0kbits/s speed=0.534x frame= 64 fps=9.6 q=22.0 size= 2560KiB time=00:00:03.87 bitrate=5412.1kbits/s speed=0.578x frame= 75 fps= 10 q=22.0 size= 3072KiB time=00:00:04.56 bitrate=5515.9kbits/s speed=0.634x frame= 84 fps= 11 q=22.0 size= 3328KiB time=00:00:05.12 bitrate=5319.7kbits/s speed=0.666x frame= 95 fps= 12 q=22.0 size= 3840KiB time=00:00:05.81 bitrate=5412.1kbits/s speed=0.709x frame= 107 fps= 12 q=22.0 size= 4096KiB time=00:00:06.56 bitrate=5113.1kbits/s speed=0.754x frame= 117 fps= 13 q=22.0 size= 4608KiB time=00:00:07.18 bitrate=5252.1kbits/s speed=0.781x frame= 126 fps= 13 q=22.0 size= 5120KiB time=00:00:07.75 bitrate=5412.1kbits/s speed=0.799x frame= 136 fps= 13 q=22.0 size= 5632KiB time=00:00:08.37 bitrate=5509.0kbits/s speed=0.821x [out#0/mp4 @ 0x6889d00] video:6048KiB audio:0KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.042466% +frame= 141 fps= 13 q=-1.0 Lsize= 6051KiB time=00:00:08.68 bitrate=5705.5kbits/s speed=0.83x +[libx264 @ 0x6887dc0] frame I:1 Avg QP:20.83 size: 89579 +[libx264 @ 0x6887dc0] frame P:36 Avg QP:19.69 size: 59344 +[libx264 @ 0x6887dc0] frame B:104 Avg QP:22.42 size: 38139 +[libx264 @ 0x6887dc0] consecutive B-frames: 0.7% 2.8% 0.0% 96.5% +[libx264 @ 0x6887dc0] mb I I16..4: 1.3% 46.0% 52.8% +[libx264 @ 0x6887dc0] mb P I16..4: 1.1% 28.5% 25.2% P16..4: 13.1% 16.0% 9.1% 0.0% 0.0% skip: 6.9% +[libx264 @ 0x6887dc0] mb B I16..4: 0.2% 9.8% 12.6% B16..8: 24.3% 17.9% 8.6% direct: 7.8% skip:18.8% L0:40.0% L1:34.2% BI:25.8% +[libx264 @ 0x6887dc0] 8x8 transform intra:47.3% inter:37.3% +[libx264 @ 0x6887dc0] coded y,uvDC,uvAC intra: 97.3% 95.1% 66.0% inter: 50.9% 35.2% 2.7% +[libx264 @ 0x6887dc0] i16 v,h,dc,p: 16% 19% 7% 58% +[libx264 @ 0x6887dc0] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 20% 12% 14% 6% 9% 9% 10% 9% 10% +[libx264 @ 0x6887dc0] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 17% 13% 14% 8% 10% 9% 10% 8% 10% +[libx264 @ 0x6887dc0] i8c dc,h,v,p: 50% 21% 22% 7% +[libx264 @ 0x6887dc0] Weighted P-Frames: Y:0.0% UV:0.0% +[libx264 @ 0x6887dc0] ref P L0: 44.9% 25.4% 20.5% 9.3% +[libx264 @ 0x6887dc0] ref B L0: 91.2% 7.0% 1.8% +[libx264 @ 0x6887dc0] ref B L1: 98.1% 1.9% +[libx264 @ 0x6887dc0] kb/s:5621.53 +ffmpeg version 7.0.2-static https://johnvansickle.com/ffmpeg/ Copyright (c) 2000-2024 the FFmpeg developers + built with gcc 8 (Debian 8.3.0-6) + configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libdav1d --enable-libxvid --enable-libzvbi --enable-libzimg + libavutil 59. 8.100 / 59. 8.100 + libavcodec 61. 3.100 / 61. 3.100 + libavformat 61. 1.100 / 61. 1.100 + libavdevice 61. 1.100 / 61. 1.100 + libavfilter 10. 1.100 / 10. 1.100 + libswscale 8. 1.100 / 8. 1.100 + libswresample 5. 1.100 / 5. 1.100 + libpostproc 58. 1.100 / 58. 1.100 +Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Duration: 00:00:08.81, start: 0.000000, bitrate: 11691 kb/s + Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(progressive), 832x464, 11688 kb/s, 16 fps, 16 tbr, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 +[Parsed_ass_0 @ 0x68b4fc0] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x68b4fc0] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x68b4fc0] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x68b4fc0] Using font provider fontconfig +[Parsed_ass_0 @ 0x68b4fc0] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.ass' (2 styles, 1 events) +Stream mapping: + Stream #0:0 -> #0:0 (h264 (native) -> h264 (libx264)) +Press [q] to stop, [?] for help +[Parsed_ass_0 @ 0x155544001d80] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x155544001d80] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x155544001d80] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x155544001d80] Using font provider fontconfig +[Parsed_ass_0 @ 0x155544001d80] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.ass' (2 styles, 1 events) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/dejavu/DejaVuSans.ttf, 0, DejaVuSans +[Parsed_ass_0 @ 0x155544001d80] Glyph 0x7B2C not found, selecting one more font for (Arial Unicode MS, 400, 0) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/droid/DroidSansFallbackFull.ttf, 0, DroidSansFallback +[libx264 @ 0x6886880] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 +[libx264 @ 0x6886880] profile High, level 3.0, 4:2:0, 8-bit +[libx264 @ 0x6886880] 264 - core 164 r3191 4613ac3 - H.264/MPEG-4 AVC codec - Copyleft 2003-2024 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=4 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=16 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=18.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00 +Output #0, mp4, to 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Stream #0:0(und): Video: h264 (avc1 / 0x31637661), yuv420p(progressive), 832x464, q=2-31, 16 fps, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 + Side data: + cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A +frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 1 fps=0.7 q=22.0 size= 0KiB time=-00:00:00.06 bitrate=N/A speed=N/A frame= 3 fps=1.4 q=22.0 size= 0KiB time=00:00:00.06 bitrate= 6.1kbits/s speed=0.0296x frame= 9 fps=3.4 q=22.0 size= 256KiB time=00:00:00.43 bitrate=4794.4kbits/s speed=0.164x frame= 18 fps=5.6 q=22.0 size= 768KiB time=00:00:01.00 bitrate=6291.8kbits/s speed=0.313x frame= 30 fps=8.1 q=22.0 size= 1024KiB time=00:00:01.75 bitrate=4793.7kbits/s speed=0.473x frame= 38 fps=9.0 q=22.0 size= 1536KiB time=00:00:02.25 bitrate=5592.6kbits/s speed=0.536x frame= 46 fps=9.8 q=22.0 size= 1792KiB time=00:00:02.75 bitrate=5338.3kbits/s speed=0.585x frame= 54 fps= 10 q=22.0 size= 2048KiB time=00:00:03.25 bitrate=5162.3kbits/s speed=0.625x frame= 64 fps= 11 q=22.0 size= 2304KiB time=00:00:03.87 bitrate=4870.9kbits/s speed=0.68x frame= 75 fps= 12 q=22.0 size= 2560KiB time=00:00:04.56 bitrate=4596.6kbits/s speed=0.736x frame= 84 fps= 13 q=22.0 size= 2816KiB time=00:00:05.12 bitrate=4501.3kbits/s speed=0.765x frame= 98 fps= 14 q=22.0 size= 3328KiB time=00:00:06.00 bitrate=4543.9kbits/s speed=0.833x frame= 114 fps= 15 q=22.0 size= 3584KiB time=00:00:07.00 bitrate=4194.4kbits/s speed=0.909x frame= 126 fps= 15 q=22.0 size= 3840KiB time=00:00:07.75 bitrate=4059.1kbits/s speed=0.945x frame= 136 fps= 16 q=22.0 size= 4352KiB time=00:00:08.37 bitrate=4257.0kbits/s speed=0.963x [out#0/mp4 @ 0x6886180] video:4657KiB audio:0KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.054643% +frame= 141 fps= 16 q=-1.0 Lsize= 4660KiB time=00:00:08.68 bitrate=4394.1kbits/s speed=0.974x +[libx264 @ 0x6886880] frame I:1 Avg QP:20.55 size: 90000 +[libx264 @ 0x6886880] frame P:36 Avg QP:18.95 size: 49558 +[libx264 @ 0x6886880] frame B:104 Avg QP:21.72 size: 27831 +[libx264 @ 0x6886880] consecutive B-frames: 1.4% 0.0% 2.1% 96.5% +[libx264 @ 0x6886880] mb I I16..4: 0.7% 48.5% 50.7% +[libx264 @ 0x6886880] mb P I16..4: 1.3% 22.0% 17.0% P16..4: 18.7% 21.4% 12.5% 0.0% 0.0% skip: 7.1% +[libx264 @ 0x6886880] mb B I16..4: 0.2% 5.6% 7.2% B16..8: 29.7% 18.6% 8.0% direct: 8.2% skip:22.5% L0:38.2% L1:39.3% BI:22.4% +[libx264 @ 0x6886880] 8x8 transform intra:49.1% inter:40.7% +[libx264 @ 0x6886880] coded y,uvDC,uvAC intra: 95.0% 95.0% 68.0% inter: 46.6% 34.1% 2.2% +[libx264 @ 0x6886880] i16 v,h,dc,p: 6% 24% 5% 65% +[libx264 @ 0x6886880] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 15% 16% 15% 6% 9% 8% 11% 8% 11% +[libx264 @ 0x6886880] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 15% 14% 14% 8% 11% 9% 10% 8% 10% +[libx264 @ 0x6886880] i8c dc,h,v,p: 53% 23% 17% 7% +[libx264 @ 0x6886880] Weighted P-Frames: Y:0.0% UV:0.0% +[libx264 @ 0x6886880] ref P L0: 46.0% 27.2% 18.3% 8.6% +[libx264 @ 0x6886880] ref B L0: 91.0% 6.9% 2.2% +[libx264 @ 0x6886880] ref B L1: 97.8% 2.2% +[libx264 @ 0x6886880] kb/s:4328.82 +ffmpeg version 7.0.2-static https://johnvansickle.com/ffmpeg/ Copyright (c) 2000-2024 the FFmpeg developers + built with gcc 8 (Debian 8.3.0-6) + configuration: --enable-gpl --enable-version3 --enable-static --disable-debug --disable-ffplay --disable-indev=sndio --disable-outdev=sndio --cc=gcc --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gmp --enable-libgme --enable-gray --enable-libaom --enable-libfribidi --enable-libass --enable-libvmaf --enable-libfreetype --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-librubberband --enable-libsoxr --enable-libspeex --enable-libsrt --enable-libvorbis --enable-libopus --enable-libtheora --enable-libvidstab --enable-libvo-amrwbenc --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libdav1d --enable-libxvid --enable-libzvbi --enable-libzimg + libavutil 59. 8.100 / 59. 8.100 + libavcodec 61. 3.100 / 61. 3.100 + libavformat 61. 1.100 / 61. 1.100 + libavdevice 61. 1.100 / 61. 1.100 + libavfilter 10. 1.100 / 10. 1.100 + libswscale 8. 1.100 / 8. 1.100 + libswresample 5. 1.100 / 5. 1.100 + libpostproc 58. 1.100 / 58. 1.100 +Input #0, mov,mp4,m4a,3gp,3g2,mj2, from 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Duration: 00:00:08.81, start: 0.000000, bitrate: 12331 kb/s + Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(progressive), 832x464, 12329 kb/s, 16 fps, 16 tbr, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 +[Parsed_ass_0 @ 0x68b6f00] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x68b6f00] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x68b6f00] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x68b6f00] Using font provider fontconfig +[Parsed_ass_0 @ 0x68b6f00] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.ass' (2 styles, 1 events) +Stream mapping: + Stream #0:0 -> #0:0 (h264 (native) -> h264 (libx264)) +Press [q] to stop, [?] for help +[Parsed_ass_0 @ 0x155544001d80] libass API version: 0x1703000 +[Parsed_ass_0 @ 0x155544001d80] libass source: commit: 0.17.3-12-g5c00ca266dd49dd7cf93a79f09a661ab1ff66439 +[Parsed_ass_0 @ 0x155544001d80] Shaper: FriBidi 1.0.11 (SIMPLE) HarfBuzz-ng 3.1.1 (COMPLEX) +[Parsed_ass_0 @ 0x155544001d80] Using font provider fontconfig +[Parsed_ass_0 @ 0x155544001d80] Added subtitle file: '/lustre/fs11/portfolios/nvr/projects/nvr_elm_llm/users/shyang/lingbot/experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.ass' (2 styles, 1 events) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/dejavu/DejaVuSans.ttf, 0, DejaVuSans +[Parsed_ass_0 @ 0x155544001d80] Glyph 0x7B2C not found, selecting one more font for (Arial Unicode MS, 400, 0) +[Parsed_ass_0 @ 0x155544001d80] fontselect: (Arial Unicode MS, 400, 0) -> /usr/share/fonts/truetype/droid/DroidSansFallbackFull.ttf, 0, DroidSansFallback +[libx264 @ 0x6887dc0] using cpu capabilities: MMX2 SSE2Fast SSSE3 SSE4.2 AVX FMA3 BMI2 AVX2 +[libx264 @ 0x6887dc0] profile High, level 3.0, 4:2:0, 8-bit +[libx264 @ 0x6887dc0] 264 - core 164 r3191 4613ac3 - H.264/MPEG-4 AVC codec - Copyleft 2003-2024 - http://www.videolan.org/x264.html - options: cabac=1 ref=3 deblock=1:0:0 analyse=0x3:0x113 me=hex subme=7 psy=1 psy_rd=1.00:0.00 mixed_ref=1 me_range=16 chroma_me=1 trellis=1 8x8dct=1 cqm=0 deadzone=21,11 fast_pskip=1 chroma_qp_offset=-2 threads=4 lookahead_threads=1 sliced_threads=0 nr=0 decimate=1 interlaced=0 bluray_compat=0 constrained_intra=0 bframes=3 b_pyramid=2 b_adapt=1 b_bias=0 direct=1 weightb=1 open_gop=0 weightp=2 keyint=250 keyint_min=16 scenecut=40 intra_refresh=0 rc_lookahead=40 rc=crf mbtree=1 crf=18.0 qcomp=0.60 qpmin=0 qpmax=69 qpstep=4 ip_ratio=1.40 aq=1:1.00 +Output #0, mp4, to 'experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.mp4': + Metadata: + major_brand : isom + minor_version : 512 + compatible_brands: isomiso2avc1mp41 + encoder : Lavf61.1.100 + Stream #0:0(und): Video: h264 (avc1 / 0x31637661), yuv420p(progressive), 832x464, q=2-31, 16 fps, 16384 tbn (default) + Metadata: + handler_name : VideoHandler + vendor_id : [0][0][0][0] + encoder : Lavc61.3.100 libx264 + Side data: + cpb: bitrate max/min/avg: 0/0/0 buffer size: 0 vbv_delay: N/A +frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 0 fps=0.0 q=0.0 size= 0KiB time=N/A bitrate=N/A speed=N/A frame= 1 fps=0.7 q=22.0 size= 0KiB time=-00:00:00.06 bitrate=N/A speed=N/A frame= 2 fps=1.0 q=22.0 size= 0KiB time=00:00:00.00 bitrate=N/A speed= 0x frame= 11 fps=4.1 q=22.0 size= 256KiB time=00:00:00.56 bitrate=3729.0kbits/s speed=0.21x frame= 21 fps=6.4 q=22.0 size= 768KiB time=00:00:01.18 bitrate=5298.4kbits/s speed=0.363x frame= 29 fps=7.7 q=22.0 size= 1280KiB time=00:00:01.68 bitrate=6214.0kbits/s speed=0.447x frame= 37 fps=8.7 q=22.0 size= 1536KiB time=00:00:02.18 bitrate=5752.4kbits/s speed=0.512x frame= 46 fps=9.6 q=22.0 size= 1792KiB time=00:00:02.75 bitrate=5338.3kbits/s speed=0.576x frame= 55 fps= 10 q=22.0 size= 2304KiB time=00:00:03.31 bitrate=5698.0kbits/s speed=0.628x frame= 66 fps= 11 q=22.0 size= 2560KiB time=00:00:04.00 bitrate=5243.0kbits/s speed=0.693x frame= 78 fps= 12 q=22.0 size= 2816KiB time=00:00:04.75 bitrate=4856.6kbits/s speed=0.757x frame= 89 fps= 13 q=22.0 size= 3328KiB time=00:00:05.43 bitrate=5014.0kbits/s speed=0.803x frame= 103 fps= 14 q=22.0 size= 3584KiB time=00:00:06.31 bitrate=4651.2kbits/s speed=0.868x frame= 117 fps= 15 q=22.0 size= 4096KiB time=00:00:07.18 bitrate=4668.5kbits/s speed=0.924x frame= 130 fps= 16 q=22.0 size= 4608KiB time=00:00:08.00 bitrate=4718.6kbits/s speed=0.967x [out#0/mp4 @ 0x6889d00] video:5031KiB audio:0KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: 0.051048% +frame= 141 fps= 16 q=-1.0 Lsize= 5034KiB time=00:00:08.68 bitrate=4746.7kbits/s speed=0.988x +[libx264 @ 0x6887dc0] frame I:1 Avg QP:20.20 size: 92835 +[libx264 @ 0x6887dc0] frame P:35 Avg QP:19.21 size: 52149 +[libx264 @ 0x6887dc0] frame B:105 Avg QP:21.83 size: 30793 +[libx264 @ 0x6887dc0] consecutive B-frames: 0.7% 0.0% 0.0% 99.3% +[libx264 @ 0x6887dc0] mb I I16..4: 0.6% 47.9% 51.5% +[libx264 @ 0x6887dc0] mb P I16..4: 1.3% 27.0% 19.2% P16..4: 17.2% 17.9% 10.4% 0.0% 0.0% skip: 7.0% +[libx264 @ 0x6887dc0] mb B I16..4: 0.1% 5.3% 6.3% B16..8: 30.3% 19.0% 8.3% direct: 9.3% skip:21.2% L0:27.7% L1:47.5% BI:24.8% +[libx264 @ 0x6887dc0] 8x8 transform intra:51.8% inter:40.6% +[libx264 @ 0x6887dc0] coded y,uvDC,uvAC intra: 95.4% 96.2% 71.4% inter: 48.4% 36.0% 2.5% +[libx264 @ 0x6887dc0] i16 v,h,dc,p: 9% 23% 8% 59% +[libx264 @ 0x6887dc0] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 14% 16% 15% 6% 10% 8% 12% 8% 12% +[libx264 @ 0x6887dc0] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 14% 15% 14% 8% 11% 8% 11% 7% 11% +[libx264 @ 0x6887dc0] i8c dc,h,v,p: 53% 23% 17% 7% +[libx264 @ 0x6887dc0] Weighted P-Frames: Y:0.0% UV:0.0% +[libx264 @ 0x6887dc0] ref P L0: 41.7% 19.2% 26.4% 12.8% +[libx264 @ 0x6887dc0] ref B L0: 88.6% 8.5% 2.8% +[libx264 @ 0x6887dc0] ref B L1: 97.4% 2.6% +[libx264 @ 0x6887dc0] kb/s:4676.37 +wrote experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_01/fast_chunk_ablation_example05_chunk01_seed6200_subtitled.mp4 +wrote experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_02/fast_chunk_ablation_example05_chunk02_seed6200_subtitled.mp4 +wrote experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_03/fast_chunk_ablation_example05_chunk03_seed6200_subtitled.mp4 +wrote experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_04/fast_chunk_ablation_example05_chunk04_seed6200_subtitled.mp4 +wrote experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_06/fast_chunk_ablation_example05_chunk06_seed6200_subtitled.mp4 +wrote experiments/fast_chunk_ablation_20260528_pose_ui_corrected/outputs/chunk_size_12/fast_chunk_ablation_example05_chunk12_seed6200_subtitled.mp4 +validated upload payload: 6 raw videos, 6 subtitled videos +It seems you are trying to upload a large folder at once. This might take some time and then fail if the folder is too large. For such cases, it is recommended to upload in smaller batches or to use `HfApi().upload_large_folder(...)`/`hf upload-large-folder` instead. For more details, check out https://huggingface.co/docs/huggingface_hub/main/en/guides/upload#upload-a-large-folder. +Start hashing 63 files. +Finished hashing 63 files. + Processing Files (0 / 0) : | | 0.00B / 0.00B + New Data Upload : | | 0.00B / 0.00B  + + ..._corrected/intrinsics.npy: 100%|██████████| 15.5kB / 15.5kB  + + ..._corrected/intrinsics.npy: 100%|██████████| 15.5kB / 15.5kB  Processing Files (1 / 18) : 0%| | 15.5kB / 114MB, ???B/s + + + ...le05_chunk02_seed6200.mp4: 1%| | 67.1kB / 13.0MB  + + + + ...le05_chunk03_seed6200.mp4: 2%|▏ | 263kB / 14.6MB  + + ..._corrected/intrinsics.npy: 100%|██████████| 15.5kB / 15.5kB  + + + ...le05_chunk02_seed6200.mp4: 1%| | 67.1kB / 13.0MB  + + + + ...le05_chunk03_seed6200.mp4: 2%|▏ | 263kB / 14.6MB  Processing Files (1 / 18) : 0%| | 346kB / 114MB, 33.7kB/s + + + + + ...se_ui_corrected/poses.npy: 100%|██████████| 61.6kB / 61.6kB  + + + + + + ...le05_chunk06_seed6200.mp4: 0%| | 35.9kB / 12.9MB  + + + + + + + ...le05_chunk01_seed6200.mp4: 1%| | 135kB / 12.8MB  + + + + + + + + ...le05_chunk04_seed6200.mp4: 2%|▏ | 303kB / 15.5MB  + + ..._corrected/intrinsics.npy: 100%|██████████| 15.5kB / 15.5kB  + + + ...le05_chunk02_seed6200.mp4: 1%| | 67.1kB / 13.0MB  + + + + ...le05_chunk03_seed6200.mp4: 2%|▏ | 263kB / 14.6MB  + + + + + ...se_ui_corrected/poses.npy: 100%|██████████| 61.6kB / 61.6kB  + + + + + + ...le05_chunk06_seed6200.mp4: 0%| | 35.9kB / 12.9MB  + + + + + + + ...le05_chunk01_seed6200.mp4: 1%| | 135kB / 12.8MB  + + + + + + + + ...le05_chunk04_seed6200.mp4: 2%|▏ | 303kB / 15.5MB  Processing Files (2 / 18) : 1%| | 881kB / 114MB, 85.0kB/s + + + + + + + + + ...le05_chunk12_seed6200.mp4: 0%| | 8.65kB / 13.6MB  + + ..._corrected/intrinsics.npy: 100%|██████████| 15.5kB / 15.5kB  + + + ...le05_chunk02_seed6200.mp4: 1%| | 67.1kB / 13.0MB  + + + + ...le05_chunk03_seed6200.mp4: 2%|▏ | 263kB / 14.6MB  + + + + + ...se_ui_corrected/poses.npy: 100%|██████████| 61.6kB / 61.6kB  + + + + + + ...le05_chunk06_seed6200.mp4: 0%| | 35.9kB / 12.9MB  + + + + + + + ...le05_chunk01_seed6200.mp4: 1%| | 135kB / 12.8MB  + + + + + + + + ...le05_chunk04_seed6200.mp4: 2%|▏ | 303kB / 15.5MB  + + + + + + + + + ...le05_chunk12_seed6200.mp4: 0%| | 8.65kB / 13.6MB  Processing Files (2 / 18) : 1%| | 890kB / 114MB, 83.8kB/s + + + + + + + + + + ...02_seed6200_subtitled.mp4: 1%| | 37.7kB / 4.80MB  + + + + + + + + + + + ...12_seed6200_subtitled.mp4: 1%| | 40.5kB / 5.15MB  + + ..._corrected/intrinsics.npy: 100%|██████████| 15.5kB / 15.5kB  + + + ...le05_chunk02_seed6200.mp4: 1%|▏ | 168kB / 13.0MB  + + + + ...le05_chunk03_seed6200.mp4: 3%|▎ | 376kB / 14.6MB  + + + + + ...se_ui_corrected/poses.npy: 100%|██████████| 61.6kB / 61.6kB  + + + + + + ...le05_chunk06_seed6200.mp4: 1%|▏ | 181kB / 12.9MB  + + + + + + + ...le05_chunk01_seed6200.mp4: 2%|▏ | 234kB / 12.8MB  + + + + + + + + ...le05_chunk04_seed6200.mp4: 3%|▎ | 476kB / 15.5MB  + + + + + + + + + ...le05_chunk12_seed6200.mp4: 1%| | 162kB / 13.6MB  + + + + + + + + + + ...02_seed6200_subtitled.mp4: 1%| | 37.7kB / 4.80MB  + + + + + + + + + + + [+ 8 files] : 1%| | 227kB / 26.8MB  Processing Files (2 / 18) : 2%|▏ | 1.94MB / 114MB, 182kB/s + New Data Upload : 1%| | 1.05MB / 113MB, 100kB/s  + + ...le05_chunk02_seed6200.mp4: 2%|▏ | 269kB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 3%|▎ | 488kB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 3%|▎ | 326kB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 3%|▎ | 334kB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 4%|▍ | 648kB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 2%|▏ | 315kB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 2%|▏ | 75.3kB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 2%|▏ | 81.0kB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 2%|▏ | 243B / 15.5kB  + + + + + + + + + + + [+ 6 files] : 2%|▏ | 373kB / 21.7MB  Processing Files (2 / 18) : 3%|▎ | 2.99MB / 114MB, 278kB/s + New Data Upload : 2%|▏ | 2.10MB / 113MB, 198kB/s  + + ...le05_chunk02_seed6200.mp4: 8%|▊ | 978kB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 9%|▊ | 1.28MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 9%|▉ | 1.20MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 8%|▊ | 1.03MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 11%|█ | 1.68MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 9%|▉ | 1.24MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 7%|▋ | 339kB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 7%|▋ | 364kB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 7%|▋ | 1.10kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 8%|▊ | 1.62MB / 21.7MB  Processing Files (2 / 18) : 9%|▊ | 9.81MB / 114MB, 915kB/s + New Data Upload : 8%|▊ | 8.92MB / 113MB, 837kB/s  + + ...le05_chunk02_seed6200.mp4: 11%|█▏ | 1.48MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 13%|█▎ | 1.84MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 16%|█▌ | 2.07MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 12%|█▏ | 1.53MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 17%|█▋ | 2.71MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 16%|█▌ | 2.15MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 11%|█ | 527kB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 11%|█ | 567kB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 11%|█ | 1.71kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 12%|█▏ | 2.61MB / 21.7MB  Processing Files (2 / 18) : 14%|█▎ | 15.6MB / 114MB, 1.43MB/s + New Data Upload : 13%|█▎ | 14.7MB / 113MB, 1.36MB/s  + + ...le05_chunk02_seed6200.mp4: 17%|█▋ | 2.19MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 18%|█▊ | 2.63MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 23%|██▎ | 2.94MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 17%|█▋ | 2.23MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 24%|██▍ | 3.74MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 23%|██▎ | 3.07MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 16%|█▋ | 791kB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 16%|█▋ | 850kB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 16%|█▋ | 2.56kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 18%|█▊ | 3.86MB / 21.7MB  Processing Files (2 / 18) : 20%|█▉ | 22.4MB / 114MB, 2.04MB/s + New Data Upload : 19%|█▉ | 21.5MB / 113MB, 1.96MB/s  + + ...le05_chunk02_seed6200.mp4: 26%|██▋ | 3.41MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 27%|██▋ | 3.98MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 36%|███▋ | 4.68MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 27%|██▋ | 3.43MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 37%|███▋ | 5.81MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 36%|███▌ | 4.91MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 26%|██▌ | 1.24MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 26%|██▌ | 1.34MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 26%|██▌ | 4.02kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 28%|██▊ | 6.10MB / 21.7MB  Processing Files (2 / 18) : 31%|███ | 35.0MB / 114MB, 3.16MB/s + New Data Upload : 30%|███ | 34.1MB / 113MB, 3.09MB/s  + + ...le05_chunk02_seed6200.mp4: 36%|███▋ | 4.72MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 37%|███▋ | 5.44MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 50%|████▉ | 6.42MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 37%|███▋ | 4.73MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 51%|█████ | 7.87MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 50%|████▉ | 6.75MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 36%|███▌ | 1.73MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 36%|███▌ | 1.86MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 36%|███▌ | 5.60kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 39%|███▉ | 8.47MB / 21.7MB  Processing Files (2 / 18) : 42%|████▏ | 48.1MB / 114MB, 4.29MB/s + New Data Upload : 42%|████▏ | 47.2MB / 113MB, 4.22MB/s  + + ...le05_chunk02_seed6200.mp4: 48%|████▊ | 6.24MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 49%|████▉ | 7.13MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 62%|██████▏ | 8.01MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 48%|████▊ | 6.23MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 63%|██████▎ | 9.77MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 62%|██████▏ | 8.44MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 48%|████▊ | 2.30MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 48%|████▊ | 2.47MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 48%|████▊ | 7.43kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 51%|█████ | 11.1MB / 21.7MB  Processing Files (2 / 18) : 54%|█████▍ | 61.7MB / 114MB, 5.45MB/s + New Data Upload : 54%|█████▎ | 60.8MB / 113MB, 5.38MB/s  + + ...le05_chunk02_seed6200.mp4: 59%|█████▉ | 7.66MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 60%|█████▉ | 8.71MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 79%|███████▉ | 10.2MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 59%|█████▉ | 7.62MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 79%|███████▉ | 12.3MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 79%|███████▉ | 10.7MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 59%|█████▉ | 2.83MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 59%|█████▉ | 3.04MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 59%|█████▉ | 9.13kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 63%|██████▎ | 13.7MB / 21.7MB  Processing Files (2 / 18) : 67%|██████▋ | 76.9MB / 114MB, 6.71MB/s + New Data Upload : 67%|██████▋ | 76.0MB / 113MB, 6.64MB/s  + + ...le05_chunk02_seed6200.mp4: 69%|██████▊ | 8.87MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 69%|██████▉ | 10.1MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 96%|█████████▌| 12.4MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 69%|██████▊ | 8.82MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 96%|█████████▌| 14.9MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 96%|█████████▌| 13.0MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 68%|██████▊ | 3.28MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 68%|██████▊ | 3.52MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 68%|██████▊ | 10.6kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 74%|███████▍ | 16.1MB / 21.7MB  Processing Files (2 / 18) : 80%|███████▉ | 91.1MB / 114MB, 7.84MB/s + New Data Upload : 80%|███████▉ | 90.2MB / 113MB, 7.77MB/s  + + ...le05_chunk02_seed6200.mp4: 80%|████████ | 10.4MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 80%|████████ | 11.8MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 99%|█████████▉| 12.8MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 80%|████████ | 10.3MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 99%|█████████▉| 15.4MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 99%|█████████▉| 13.5MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 80%|████████ | 3.84MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 80%|████████ | 4.13MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 80%|████████ | 12.4kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 84%|████████▍ | 18.3MB / 21.7MB  Processing Files (2 / 18) : 88%|████████▊ | 101MB / 114MB, 8.51MB/s + New Data Upload : 88%|████████▊ | 99.6MB / 113MB, 8.45MB/s  + + ...le05_chunk02_seed6200.mp4: 92%|█████████▏| 11.9MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 92%|█████████▏| 13.4MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 99%|█████████▉| 12.8MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 92%|█████████▏| 11.8MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 99%|█████████▉| 15.4MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 99%|█████████▉| 13.5MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 92%|█████████▏| 4.41MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 92%|█████████▏| 4.74MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 92%|█████████▏| 14.2kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 94%|█████████▎| 20.3MB / 21.7MB  Processing Files (2 / 18) : 95%|█████████▌| 108MB / 114MB, 9.03MB/s + New Data Upload : 95%|█████████▍| 108MB / 113MB, 8.96MB/s  + + ...le05_chunk02_seed6200.mp4: 100%|█████████▉| 12.9MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 100%|█████████▉| 14.6MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 99%|█████████▉| 12.8MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 100%|█████████▉| 12.8MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 99%|█████████▉| 15.4MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 99%|█████████▉| 13.5MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 100%|█████████▉| 4.78MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 100%|█████████▉| 5.14MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 100%|█████████▉| 15.5kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 100%|█████████▉| 21.6MB / 21.7MB  Processing Files (2 / 18) : 100%|█████████▉| 114MB / 114MB, 9.29MB/s + New Data Upload : 100%|█████████▉| 113MB / 113MB, 9.23MB/s  + + ...le05_chunk02_seed6200.mp4: 100%|█████████▉| 12.9MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 100%|█████████▉| 14.6MB / 14.6MB  + + + + ...le05_chunk06_seed6200.mp4: 100%|██████████| 12.9MB / 12.9MB  + + + + + ...le05_chunk01_seed6200.mp4: 100%|█████████▉| 12.8MB / 12.8MB  + + + + + + ...le05_chunk04_seed6200.mp4: 100%|██████████| 15.5MB / 15.5MB  + + + + + + + ...le05_chunk12_seed6200.mp4: 100%|██████████| 13.6MB / 13.6MB  + + + + + + + + ...02_seed6200_subtitled.mp4: 100%|█████████▉| 4.78MB / 4.80MB  + + + + + + + + + ...12_seed6200_subtitled.mp4: 100%|█████████▉| 5.14MB / 5.15MB  + + + + + + + + + + ...corrected/wasd_action.npy: 100%|█████████▉| 15.5kB / 15.5kB  + + + + + + + + + + + [+ 6 files] : 100%|█████████▉| 21.6MB / 21.7MB  Processing Files (6 / 18) : 100%|█████████▉| 114MB / 114MB, 9.11MB/s + New Data Upload : 100%|█████████▉| 113MB / 113MB, 9.05MB/s  + + ...le05_chunk02_seed6200.mp4: 100%|██████████| 13.0MB / 13.0MB  + + + ...le05_chunk03_seed6200.mp4: 100%|██████████| 14.6MB / 14.6MB  + + + + ...le05_chunk01_seed6200.mp4: 100%|██████████| 12.8MB / 12.8MB  + + + + + ...02_seed6200_subtitled.mp4: 100%|██████████| 4.80MB / 4.80MB  + + + + + + ...12_seed6200_subtitled.mp4: 100%|██████████| 5.15MB / 5.15MB  + + + + + + + ...corrected/wasd_action.npy: 100%|██████████| 15.5kB / 15.5kB  + + + + + + + + ...03_seed6200_subtitled.mp4: 100%|██████████| 5.80MB / 5.80MB  + + + + + + + + + ...corrected/ijkl_action.npy: 100%|██████████| 15.5kB / 15.5kB  + + + + + + + + + + ...04_seed6200_subtitled.mp4: 100%|██████████| 6.20MB / 6.20MB  + + + + + + + + + + + [+ 2 files] : 100%|██████████| 4.88MB / 4.88MB  Processing Files (18 / 18) : 100%|██████████| 114MB / 114MB, 8.54MB/s + New Data Upload : 100%|██████████| 113MB / 113MB, 8.48MB/s  Processing Files (18 / 18) : 100%|██████████| 114MB / 114MB, 8.54MB/s + New Data Upload : 100%|██████████| 113MB / 113MB, 8.48MB/s + ...le05_chunk02_seed6200.mp4: 100%|██████████| 13.0MB / 13.0MB + ...le05_chunk03_seed6200.mp4: 100%|██████████| 14.6MB / 14.6MB + ...le05_chunk01_seed6200.mp4: 100%|██████████| 12.8MB / 12.8MB + ...02_seed6200_subtitled.mp4: 100%|██████████| 4.80MB / 4.80MB + ...12_seed6200_subtitled.mp4: 100%|██████████| 5.15MB / 5.15MB + ...corrected/wasd_action.npy: 100%|██████████| 15.5kB / 15.5kB + ...03_seed6200_subtitled.mp4: 100%|██████████| 5.80MB / 5.80MB + ...corrected/ijkl_action.npy: 100%|██████████| 15.5kB / 15.5kB + ...04_seed6200_subtitled.mp4: 100%|██████████| 6.20MB / 6.20MB + [+ 2 files] : 100%|██████████| 4.88MB / 4.88MB diff --git a/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/upload_results.sh b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/upload_results.sh new file mode 100644 index 0000000000000000000000000000000000000000..94c82f8e6227ed1f414e996a1452fb9b6855df84 --- /dev/null +++ b/fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/20260529_005115/upload_results.sh @@ -0,0 +1,164 @@ +#!/bin/bash +set -euo pipefail + +REPO_ID="${REPO_ID:-Andyson/lingbot}" +REPO_TYPE="${REPO_TYPE:-dataset}" +LOCAL_ROOT="${LOCAL_ROOT:-experiments/fast_chunk_ablation_20260528_pose_ui_corrected}" +REMOTE_ROOT="${REMOTE_ROOT:-fast-inference/chunk-size-ablation/2026-05-28_same_prompt_pose_ui_corrected/$(date +%Y%m%d_%H%M%S)}" +EXPECTED_ROWS="${EXPECTED_ROWS:-6}" +SUBTITLE_WORKERS="${SUBTITLE_WORKERS:-1}" + +python - <<'PY' +import csv +import json +import os +from pathlib import Path + +local_root = Path(os.environ.get("LOCAL_ROOT", "experiments/fast_chunk_ablation_20260528_pose_ui_corrected")) +expected = int(os.environ.get("EXPECTED_ROWS", "6")) +manifest = json.loads((local_root / "manifest.json").read_text(encoding="utf-8")) +rows = [] +bad = [] +for case in manifest["cases"]: + metrics_file = local_root / "metrics" / f"{case['case_id']}.json" + if not metrics_file.exists(): + bad.append(f"{case['case_id']}: missing metrics") + continue + data = json.loads(metrics_file.read_text(encoding="utf-8")) + row = dict(data.get("summary", {})) + gpu = data.get("gpu", {}) + row["peak_single_gpu_memory_used_mib"] = gpu.get("peak_single_gpu_memory_used_mib") + row["peak_total_memory_used_mib"] = gpu.get("peak_total_memory_used_mib") + row["mean_gpu_utilization_percent"] = gpu.get("mean_gpu_utilization_percent") + row["mean_nonzero_gpu_count"] = gpu.get("mean_nonzero_gpu_count") + rows.append(row) + if ( + data.get("status") != "ok" + or row.get("frame_num") != manifest["frame_num"] + or row.get("seed") != manifest["seed"] + or row.get("output_exists") is not True + ): + bad.append(f"{case['case_id']}: {data.get('status')}") +if len(rows) != expected: + bad.append(f"expected {expected} rows, found {len(rows)}") +if bad: + raise SystemExit("Unfinished chunk ablation rows: " + "; ".join(bad)) + +csv_path = local_root / "run_manifest.csv" +json_path = local_root / "run_manifest.json" +fieldnames = [ + "case_id", + "chunk_size", + "chunk_frame_span", + "chunk_seconds", + "frame_num", + "fps", + "effective_duration_seconds", + "seed", + "status", + "returncode", + "wall_time_seconds", + "peak_single_gpu_memory_used_mib", + "peak_total_memory_used_mib", + "mean_gpu_utilization_percent", + "mean_nonzero_gpu_count", + "output_file", + "output_size_bytes", + "nproc_per_node", +] +with csv_path.open("w", encoding="utf-8", newline="") as handle: + writer = csv.DictWriter(handle, fieldnames=fieldnames) + writer.writeheader() + for row in sorted(rows, key=lambda item: item["chunk_size"]): + writer.writerow({key: row.get(key) for key in fieldnames}) +json_path.write_text(json.dumps(sorted(rows, key=lambda item: item["chunk_size"]), indent=2, ensure_ascii=False) + "\n", encoding="utf-8") + +report = [ + "# LingBot Fast Chunk Size Ablation", + "", + f"- Description: {manifest.get('description', '')}", + f"- Prompt: {manifest['prompt_en']}", + f"- Image: `{manifest['image']}`", + f"- Camera/action path: `{manifest['action_path']}`", + f"- Visualization controls: {manifest.get('visualization_control_source', 'not specified')}", + f"- Seed: `{manifest['seed']}`", + f"- Frame num / FPS: `{manifest['frame_num']}` / `{manifest['fps']}`", + f"- Effective duration: `{manifest['effective_duration_seconds']}` seconds", + "", + "| chunk_size | chunk seconds | status | wall time s | peak single GPU MiB | mean GPU util % | output |", + "| ---: | ---: | --- | ---: | ---: | ---: | --- |", +] +for row in sorted(rows, key=lambda item: item["chunk_size"]): + report.append( + "| {chunk_size} | {chunk_seconds} | {status} | {wall_time_seconds:.1f} | {peak_single_gpu_memory_used_mib} | {mean_gpu_utilization_percent} | `{output_file}` |".format(**row) + ) +(local_root / "report.md").write_text("\n".join(report) + "\n", encoding="utf-8") +print(f"Wrote {len(rows)} chunk ablation rows") +PY + +PROMPT_EN="$(python - <<'PY' +import json +import os +from pathlib import Path +local_root = Path(os.environ.get("LOCAL_ROOT", "experiments/fast_chunk_ablation_20260528_pose_ui_corrected")) +data = json.loads((local_root / "manifest.json").read_text(encoding="utf-8")) +print(data["prompt_en"]) +PY +)" +PROMPT_ZH="$(python - <<'PY' +import json +import os +from pathlib import Path +local_root = Path(os.environ.get("LOCAL_ROOT", "experiments/fast_chunk_ablation_20260528_pose_ui_corrected")) +data = json.loads((local_root / "manifest.json").read_text(encoding="utf-8")) +print(data["prompt_zh"]) +PY +)" + +python .codex/skills/lingbot-bilingual-video-subtitles/scripts/add_bilingual_subtitles.py \ + --root "${LOCAL_ROOT}/outputs" \ + --prompt "${PROMPT_EN}" \ + --prompt-zh "${PROMPT_ZH}" \ + --frame-num 141 \ + --fps 16 \ + --workers "${SUBTITLE_WORKERS}" \ + --force + +python - <<'PY' +import os +from pathlib import Path + +local_root = Path(os.environ.get("LOCAL_ROOT", "experiments/fast_chunk_ablation_20260528_pose_ui_corrected")) +expected = int(os.environ.get("EXPECTED_ROWS", "6")) +raw = sorted( + path for path in (local_root / "outputs").glob("*/*.mp4") + if not path.stem.endswith("_subtitled") +) +subtitled = sorted((local_root / "outputs").glob("*/*_subtitled.mp4")) +ass_files = sorted((local_root / "outputs").glob("*/*_subtitled.ass")) +srt_files = sorted((local_root / "outputs").glob("*/*_subtitled.srt")) +for label, files in { + "raw MP4s": raw, + "subtitled MP4s": subtitled, + "ASS subtitle files": ass_files, + "SRT subtitle files": srt_files, +}.items(): + if len(files) != expected: + raise SystemExit(f"Expected {expected} {label}, found {len(files)}") +for path in subtitled: + if path.stat().st_size == 0: + raise SystemExit(f"Empty subtitled video: {path}") +for path in ass_files + srt_files: + text = path.read_text(encoding="utf-8") + if not any(ord(ch) > 127 for ch in text): + raise SystemExit(f"Subtitle file lacks Chinese text: {path}") +print(f"validated upload payload: {len(raw)} raw videos, {len(subtitled)} subtitled videos") +PY + +hf auth whoami >/dev/null +hf repo create "${REPO_ID}" --repo-type "${REPO_TYPE}" --exist-ok >/dev/null +hf upload "${REPO_ID}" "${LOCAL_ROOT}" "${REMOTE_ROOT}" \ + --repo-type "${REPO_TYPE}" \ + --commit-message "Upload LingBot fast chunk-size ablation" + +echo "Uploaded ${LOCAL_ROOT} to ${REPO_ID}/${REMOTE_ROOT}"