Upload SCOPE FastVideo LoRA comparisons
Browse files- .gitattributes +15 -0
- SCOPE_fastvideo/clips/example_0_scope_30step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_0_scope_3step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_0_scope_lora_3step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_0_scope_lora_vsa_3step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_1_scope_30step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_1_scope_3step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_1_scope_lora_3step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_1_scope_lora_vsa_3step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_2_scope_30step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_2_scope_3step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_2_scope_lora_3step.mp4 +3 -0
- SCOPE_fastvideo/clips/example_2_scope_lora_vsa_3step.mp4 +3 -0
- SCOPE_fastvideo/comparisons/example_0_comparison.mp4 +3 -0
- SCOPE_fastvideo/comparisons/example_1_comparison.mp4 +3 -0
- SCOPE_fastvideo/comparisons/example_2_comparison.mp4 +3 -0
- SCOPE_fastvideo/environment.json +13 -0
- SCOPE_fastvideo/logs/example_0_scope_30step.log +31 -0
- SCOPE_fastvideo/logs/example_0_scope_3step.log +31 -0
- SCOPE_fastvideo/logs/example_0_scope_lora_3step.log +31 -0
- SCOPE_fastvideo/logs/example_0_scope_lora_vsa_3step.log +33 -0
- SCOPE_fastvideo/logs/example_1_scope_30step.log +31 -0
- SCOPE_fastvideo/logs/example_1_scope_3step.log +31 -0
- SCOPE_fastvideo/logs/example_1_scope_lora_3step.log +31 -0
- SCOPE_fastvideo/logs/example_1_scope_lora_vsa_3step.log +33 -0
- SCOPE_fastvideo/logs/example_2_scope_30step.log +31 -0
- SCOPE_fastvideo/logs/example_2_scope_3step.log +31 -0
- SCOPE_fastvideo/logs/example_2_scope_lora_3step.log +31 -0
- SCOPE_fastvideo/logs/example_2_scope_lora_vsa_3step.log +33 -0
- SCOPE_fastvideo/manifest.json +254 -0
.gitattributes
CHANGED
|
@@ -37,3 +37,18 @@ NAVA/idx0_画面呈现电影质感的写实风格,整体色调以暖-av-0.mp4
|
|
| 37 |
NAVA/idx1_这是一段具有史诗级写实电影质感的视频,整-av-0.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 38 |
NAVA/idx2_画面呈现奇幻写实电影质感,整体色调以幽蓝-av-0.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 39 |
NAVA/idx3_画面呈现史诗级蒸汽朋克写实电影风格,整体-av-0.mp4 filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
NAVA/idx1_这是一段具有史诗级写实电影质感的视频,整-av-0.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 38 |
NAVA/idx2_画面呈现奇幻写实电影质感,整体色调以幽蓝-av-0.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 39 |
NAVA/idx3_画面呈现史诗级蒸汽朋克写实电影风格,整体-av-0.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 40 |
+
SCOPE_fastvideo/clips/example_0_scope_30step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 41 |
+
SCOPE_fastvideo/clips/example_0_scope_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 42 |
+
SCOPE_fastvideo/clips/example_0_scope_lora_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 43 |
+
SCOPE_fastvideo/clips/example_0_scope_lora_vsa_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 44 |
+
SCOPE_fastvideo/clips/example_1_scope_30step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 45 |
+
SCOPE_fastvideo/clips/example_1_scope_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 46 |
+
SCOPE_fastvideo/clips/example_1_scope_lora_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 47 |
+
SCOPE_fastvideo/clips/example_1_scope_lora_vsa_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 48 |
+
SCOPE_fastvideo/clips/example_2_scope_30step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 49 |
+
SCOPE_fastvideo/clips/example_2_scope_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 50 |
+
SCOPE_fastvideo/clips/example_2_scope_lora_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 51 |
+
SCOPE_fastvideo/clips/example_2_scope_lora_vsa_3step.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 52 |
+
SCOPE_fastvideo/comparisons/example_0_comparison.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 53 |
+
SCOPE_fastvideo/comparisons/example_1_comparison.mp4 filter=lfs diff=lfs merge=lfs -text
|
| 54 |
+
SCOPE_fastvideo/comparisons/example_2_comparison.mp4 filter=lfs diff=lfs merge=lfs -text
|
SCOPE_fastvideo/clips/example_0_scope_30step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1426997ac4a4f593295a6c0e1b260c6008a3224bf1497cc10633be0f654eb42a
|
| 3 |
+
size 2140754
|
SCOPE_fastvideo/clips/example_0_scope_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7c28358d92c62447e6b5df95a3c3a93bda45ea7e447e26a3f8ec0418f8fab973
|
| 3 |
+
size 3109044
|
SCOPE_fastvideo/clips/example_0_scope_lora_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95a2882d7e07f700a83c149b1692293f7bcd8510db3db2062527ddb891c9d24b
|
| 3 |
+
size 2804542
|
SCOPE_fastvideo/clips/example_0_scope_lora_vsa_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95a2882d7e07f700a83c149b1692293f7bcd8510db3db2062527ddb891c9d24b
|
| 3 |
+
size 2804542
|
SCOPE_fastvideo/clips/example_1_scope_30step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:02d662c49cf70e2dd731c3da88e4df0ce13a52b9eb611bbf5537a63359e6499e
|
| 3 |
+
size 1022290
|
SCOPE_fastvideo/clips/example_1_scope_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:da6fc7f0f7faf59dbf13ef2a9aeab1be72ebfaa0070e2c64f6949a774e158037
|
| 3 |
+
size 1478917
|
SCOPE_fastvideo/clips/example_1_scope_lora_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cd9fa811994742d3ed7d7efe0b04d006dfaf4abec848162130463a6a04664cb2
|
| 3 |
+
size 1196829
|
SCOPE_fastvideo/clips/example_1_scope_lora_vsa_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cd9fa811994742d3ed7d7efe0b04d006dfaf4abec848162130463a6a04664cb2
|
| 3 |
+
size 1196829
|
SCOPE_fastvideo/clips/example_2_scope_30step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d87239ae553ec384af6f1112a2985668c33001ec026e7b927103b687554a59da
|
| 3 |
+
size 836780
|
SCOPE_fastvideo/clips/example_2_scope_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:41de2feb13d639cef3b0e70f3520898d955865ec7ce008508f3499a9fa4b4981
|
| 3 |
+
size 780348
|
SCOPE_fastvideo/clips/example_2_scope_lora_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cab82db53f47ce815998103bfab72f23b2f69b2d99a81c9efe6998dd3dcf3431
|
| 3 |
+
size 869663
|
SCOPE_fastvideo/clips/example_2_scope_lora_vsa_3step.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cab82db53f47ce815998103bfab72f23b2f69b2d99a81c9efe6998dd3dcf3431
|
| 3 |
+
size 869663
|
SCOPE_fastvideo/comparisons/example_0_comparison.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b0b231018f8a039322af29ad331fb2155c4f1081660e6afef13bd3ed6706cb6c
|
| 3 |
+
size 25240204
|
SCOPE_fastvideo/comparisons/example_1_comparison.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:45639cfd68c70e12e0372d57d68c128f67c71b53bde8628dcc0cd8d976a72aa8
|
| 3 |
+
size 11494531
|
SCOPE_fastvideo/comparisons/example_2_comparison.mp4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c5b97e3cf92b1963a207ab0e429dab52f98c3bd59b1cd17595e50461b7c2767e
|
| 3 |
+
size 6476827
|
SCOPE_fastvideo/environment.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"python": "3.12.9 | packaged by Anaconda, Inc. | (main, Feb 6 2025, 18:56:27) [GCC 11.2.0]",
|
| 3 |
+
"executable": "/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python",
|
| 4 |
+
"cwd": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA",
|
| 5 |
+
"time": "2026-06-08 04:40:04",
|
| 6 |
+
"torch": {
|
| 7 |
+
"version": "2.5.0+cu124",
|
| 8 |
+
"cuda": "12.4",
|
| 9 |
+
"cuda_available": true,
|
| 10 |
+
"bf16_supported": true,
|
| 11 |
+
"device_name": "NVIDIA H100 80GB HBM3"
|
| 12 |
+
}
|
| 13 |
+
}
|
SCOPE_fastvideo/logs/example_0_scope_30step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/30 [00:00<?, ?it/s]
|
| 1 |
3%|▎ | 1/30 [00:01<00:40, 1.39s/it]
|
| 2 |
7%|▋ | 2/30 [00:02<00:30, 1.08s/it]
|
| 3 |
10%|█ | 3/30 [00:03<00:26, 1.01it/s]
|
| 4 |
13%|█▎ | 4/30 [00:03<00:24, 1.06it/s]
|
| 5 |
17%|█▋ | 5/30 [00:04<00:22, 1.09it/s]
|
| 6 |
20%|██ | 6/30 [00:05<00:21, 1.11it/s]
|
| 7 |
23%|██▎ | 7/30 [00:06<00:20, 1.12it/s]
|
| 8 |
27%|██▋ | 8/30 [00:07<00:19, 1.13it/s]
|
| 9 |
30%|███ | 9/30 [00:08<00:18, 1.14it/s]
|
| 10 |
33%|███▎ | 10/30 [00:09<00:17, 1.14it/s]
|
| 11 |
37%|███▋ | 11/30 [00:10<00:16, 1.14it/s]
|
| 12 |
40%|████ | 12/30 [00:10<00:15, 1.14it/s]
|
| 13 |
43%|████▎ | 13/30 [00:11<00:14, 1.14it/s]
|
| 14 |
47%|████▋ | 14/30 [00:12<00:13, 1.14it/s]
|
| 15 |
50%|█████ | 15/30 [00:13<00:13, 1.14it/s]
|
| 16 |
53%|█████▎ | 16/30 [00:14<00:12, 1.14it/s]
|
| 17 |
57%|█████▋ | 17/30 [00:15<00:11, 1.14it/s]
|
| 18 |
60%|██████ | 18/30 [00:16<00:10, 1.14it/s]
|
| 19 |
63%|██████▎ | 19/30 [00:17<00:09, 1.14it/s]
|
| 20 |
67%|██████▋ | 20/30 [00:17<00:08, 1.14it/s]
|
| 21 |
70%|███████ | 21/30 [00:18<00:07, 1.14it/s]
|
| 22 |
73%|███████▎ | 22/30 [00:19<00:06, 1.14it/s]
|
| 23 |
77%|███████▋ | 23/30 [00:20<00:06, 1.14it/s]
|
| 24 |
80%|████████ | 24/30 [00:21<00:05, 1.14it/s]
|
| 25 |
83%|████████▎ | 25/30 [00:22<00:04, 1.14it/s]
|
| 26 |
87%|████████▋ | 26/30 [00:23<00:03, 1.14it/s]
|
| 27 |
90%|█████████ | 27/30 [00:24<00:02, 1.14it/s]
|
| 28 |
93%|█████████▎| 28/30 [00:24<00:01, 1.14it/s]
|
| 29 |
97%|█████████▋| 29/30 [00:25<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_0/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_0/action.parquet --prompt 'In a whimsical, toy-inspired garden, the first-person view reveals a tactical weapon aimed forward along a sunlit path. The vibrant environment includes oversized flowers, colorful building blocks, and towering structures. In the foreground, a carved wooden sign reads ROSE GARDEN, accompanied by giant puzzle pieces scattered across the ground. The surroundings are filled with fairy lights hanging from trees, creating a warm, golden glow. The garden looks like a miniature diorama, set in a charming and playful atmosphere.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_0/scope_30step --height 480 --width 832 --max_frames 81 --num_inference_steps 30 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/30 [00:00<?, ?it/s]
|
| 31 |
3%|▎ | 1/30 [00:01<00:40, 1.39s/it]
|
| 32 |
7%|▋ | 2/30 [00:02<00:30, 1.08s/it]
|
| 33 |
10%|█ | 3/30 [00:03<00:26, 1.01it/s]
|
| 34 |
13%|█▎ | 4/30 [00:03<00:24, 1.06it/s]
|
| 35 |
17%|█▋ | 5/30 [00:04<00:22, 1.09it/s]
|
| 36 |
20%|██ | 6/30 [00:05<00:21, 1.11it/s]
|
| 37 |
23%|██▎ | 7/30 [00:06<00:20, 1.12it/s]
|
| 38 |
27%|██▋ | 8/30 [00:07<00:19, 1.13it/s]
|
| 39 |
30%|███ | 9/30 [00:08<00:18, 1.14it/s]
|
| 40 |
33%|███▎ | 10/30 [00:09<00:17, 1.14it/s]
|
| 41 |
37%|███▋ | 11/30 [00:10<00:16, 1.14it/s]
|
| 42 |
40%|████ | 12/30 [00:10<00:15, 1.14it/s]
|
| 43 |
43%|████▎ | 13/30 [00:11<00:14, 1.14it/s]
|
| 44 |
47%|████▋ | 14/30 [00:12<00:13, 1.14it/s]
|
| 45 |
50%|█████ | 15/30 [00:13<00:13, 1.14it/s]
|
| 46 |
53%|█████▎ | 16/30 [00:14<00:12, 1.14it/s]
|
| 47 |
57%|█████▋ | 17/30 [00:15<00:11, 1.14it/s]
|
| 48 |
60%|██████ | 18/30 [00:16<00:10, 1.14it/s]
|
| 49 |
63%|██████▎ | 19/30 [00:17<00:09, 1.14it/s]
|
| 50 |
67%|██████▋ | 20/30 [00:17<00:08, 1.14it/s]
|
| 51 |
70%|███████ | 21/30 [00:18<00:07, 1.14it/s]
|
| 52 |
73%|███████▎ | 22/30 [00:19<00:06, 1.14it/s]
|
| 53 |
77%|███████▋ | 23/30 [00:20<00:06, 1.14it/s]
|
| 54 |
80%|████████ | 24/30 [00:21<00:05, 1.14it/s]
|
| 55 |
83%|████████▎ | 25/30 [00:22<00:04, 1.14it/s]
|
| 56 |
87%|████████▋ | 26/30 [00:23<00:03, 1.14it/s]
|
| 57 |
90%|█████████ | 27/30 [00:24<00:02, 1.14it/s]
|
| 58 |
93%|█████████▎| 28/30 [00:24<00:01, 1.14it/s]
|
| 59 |
97%|█████████▋| 29/30 [00:25<00:00, 1.14it/s]
|
| 60 |
+
|
| 61 |
+
|
SCOPE_fastvideo/logs/example_0_scope_3step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.12it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_0/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_0/action.parquet --prompt 'In a whimsical, toy-inspired garden, the first-person view reveals a tactical weapon aimed forward along a sunlit path. The vibrant environment includes oversized flowers, colorful building blocks, and towering structures. In the foreground, a carved wooden sign reads ROSE GARDEN, accompanied by giant puzzle pieces scattered across the ground. The surroundings are filled with fairy lights hanging from trees, creating a warm, golden glow. The garden looks like a miniature diorama, set in a charming and playful atmosphere.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_0/scope_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 31 |
33%|███▎ | 1/3 [00:00<00:01, 1.12it/s]
|
| 32 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
| 33 |
+
|
| 34 |
+
|
SCOPE_fastvideo/logs/example_0_scope_lora_3step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_0/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_0/action.parquet --prompt 'In a whimsical, toy-inspired garden, the first-person view reveals a tactical weapon aimed forward along a sunlit path. The vibrant environment includes oversized flowers, colorful building blocks, and towering structures. In the foreground, a carved wooden sign reads ROSE GARDEN, accompanied by giant puzzle pieces scattered across the ground. The surroundings are filled with fairy lights hanging from trees, creating a warm, golden glow. The garden looks like a miniature diorama, set in a charming and playful atmosphere.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_0/scope_lora_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode --fastvideo_lora_path_or_repo /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/fastvideo_dmd_lora/adapter_r32.safetensors --lora_strength 1.0
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 31 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 32 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
| 33 |
+
|
| 34 |
+
|
SCOPE_fastvideo/logs/example_0_scope_lora_vsa_3step.log
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/diffsynth/models/scope_dit.py:99: RuntimeWarning: SCOPE VSA kernel disabled for this runtime; falling back to full attention. device=NVIDIA H100 80GB HBM3, torch=2.5.0+cu124, triton=3.1.0, fastvideo-kernel=0.2.6. Set SCOPE_FORCE_VSA_KERNEL=1 to bypass this guard.
|
|
|
|
|
|
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_0/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_0/action.parquet --prompt 'In a whimsical, toy-inspired garden, the first-person view reveals a tactical weapon aimed forward along a sunlit path. The vibrant environment includes oversized flowers, colorful building blocks, and towering structures. In the foreground, a carved wooden sign reads ROSE GARDEN, accompanied by giant puzzle pieces scattered across the ground. The surroundings are filled with fairy lights hanging from trees, creating a warm, golden glow. The garden looks like a miniature diorama, set in a charming and playful atmosphere.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_0/scope_lora_vsa_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode --fastvideo_lora_path_or_repo /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/fastvideo_dmd_lora/adapter_r32.safetensors --lora_strength 1.0 --use_vsa --vsa_sparsity 0.9
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/diffsynth/models/scope_dit.py:99: RuntimeWarning: SCOPE VSA kernel disabled for this runtime; falling back to full attention. device=NVIDIA H100 80GB HBM3, torch=2.5.0+cu124, triton=3.1.0, fastvideo-kernel=0.2.6. Set SCOPE_FORCE_VSA_KERNEL=1 to bypass this guard.
|
| 31 |
+
warnings.warn(
|
| 32 |
+
|
| 33 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 34 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
| 35 |
+
|
| 36 |
+
|
SCOPE_fastvideo/logs/example_1_scope_30step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/30 [00:00<?, ?it/s]
|
| 1 |
3%|▎ | 1/30 [00:00<00:25, 1.13it/s]
|
| 2 |
7%|▋ | 2/30 [00:01<00:24, 1.14it/s]
|
| 3 |
10%|█ | 3/30 [00:02<00:23, 1.15it/s]
|
| 4 |
13%|█▎ | 4/30 [00:03<00:22, 1.15it/s]
|
| 5 |
17%|█▋ | 5/30 [00:04<00:21, 1.15it/s]
|
| 6 |
20%|██ | 6/30 [00:05<00:20, 1.15it/s]
|
| 7 |
23%|██▎ | 7/30 [00:06<00:20, 1.15it/s]
|
| 8 |
27%|██▋ | 8/30 [00:06<00:19, 1.15it/s]
|
| 9 |
30%|███ | 9/30 [00:07<00:18, 1.15it/s]
|
| 10 |
33%|███▎ | 10/30 [00:08<00:17, 1.15it/s]
|
| 11 |
37%|███▋ | 11/30 [00:09<00:16, 1.15it/s]
|
| 12 |
40%|████ | 12/30 [00:10<00:15, 1.15it/s]
|
| 13 |
43%|████▎ | 13/30 [00:11<00:14, 1.15it/s]
|
| 14 |
47%|████▋ | 14/30 [00:12<00:13, 1.15it/s]
|
| 15 |
50%|█████ | 15/30 [00:13<00:13, 1.15it/s]
|
| 16 |
53%|█████▎ | 16/30 [00:13<00:12, 1.14it/s]
|
| 17 |
57%|█████▋ | 17/30 [00:14<00:11, 1.14it/s]
|
| 18 |
60%|██████ | 18/30 [00:15<00:10, 1.14it/s]
|
| 19 |
63%|██████▎ | 19/30 [00:16<00:09, 1.14it/s]
|
| 20 |
67%|██████▋ | 20/30 [00:17<00:08, 1.14it/s]
|
| 21 |
70%|███████ | 21/30 [00:18<00:07, 1.14it/s]
|
| 22 |
73%|███████▎ | 22/30 [00:19<00:06, 1.14it/s]
|
| 23 |
77%|███████▋ | 23/30 [00:20<00:06, 1.14it/s]
|
| 24 |
80%|████████ | 24/30 [00:20<00:05, 1.14it/s]
|
| 25 |
83%|████████▎ | 25/30 [00:21<00:04, 1.14it/s]
|
| 26 |
87%|████████▋ | 26/30 [00:22<00:03, 1.14it/s]
|
| 27 |
90%|��████████ | 27/30 [00:23<00:02, 1.14it/s]
|
| 28 |
93%|█████████▎| 28/30 [00:24<00:01, 1.14it/s]
|
| 29 |
97%|█████████▋| 29/30 [00:25<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_1/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_1/action.parquet --prompt 'A breathtaking view of a vibrant fantasy world spread across a vast, open valley, seen through the lens of an FPS perspective. The scene merges intricate architecture, lush terrain, and majestic cliffs, with the forefront focusing on a finely detailed, sleek assault rifle held firmly by a character in dark attire. The weapon, with its matte black finish and tactical design, gleams softly in the light. It is equipped with a reflex sight and a foregrip, ready for action. All of this is set under a vivid blue sky filled with puffy clouds and bathed in soft light.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_1/scope_30step --height 480 --width 832 --max_frames 81 --num_inference_steps 30 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/30 [00:00<?, ?it/s]
|
| 31 |
3%|▎ | 1/30 [00:00<00:25, 1.13it/s]
|
| 32 |
7%|▋ | 2/30 [00:01<00:24, 1.14it/s]
|
| 33 |
10%|█ | 3/30 [00:02<00:23, 1.15it/s]
|
| 34 |
13%|█▎ | 4/30 [00:03<00:22, 1.15it/s]
|
| 35 |
17%|█▋ | 5/30 [00:04<00:21, 1.15it/s]
|
| 36 |
20%|██ | 6/30 [00:05<00:20, 1.15it/s]
|
| 37 |
23%|██▎ | 7/30 [00:06<00:20, 1.15it/s]
|
| 38 |
27%|██▋ | 8/30 [00:06<00:19, 1.15it/s]
|
| 39 |
30%|███ | 9/30 [00:07<00:18, 1.15it/s]
|
| 40 |
33%|███▎ | 10/30 [00:08<00:17, 1.15it/s]
|
| 41 |
37%|███▋ | 11/30 [00:09<00:16, 1.15it/s]
|
| 42 |
40%|████ | 12/30 [00:10<00:15, 1.15it/s]
|
| 43 |
43%|████▎ | 13/30 [00:11<00:14, 1.15it/s]
|
| 44 |
47%|████▋ | 14/30 [00:12<00:13, 1.15it/s]
|
| 45 |
50%|█████ | 15/30 [00:13<00:13, 1.15it/s]
|
| 46 |
53%|█████▎ | 16/30 [00:13<00:12, 1.14it/s]
|
| 47 |
57%|█████▋ | 17/30 [00:14<00:11, 1.14it/s]
|
| 48 |
60%|██████ | 18/30 [00:15<00:10, 1.14it/s]
|
| 49 |
63%|██████▎ | 19/30 [00:16<00:09, 1.14it/s]
|
| 50 |
67%|██████▋ | 20/30 [00:17<00:08, 1.14it/s]
|
| 51 |
70%|███████ | 21/30 [00:18<00:07, 1.14it/s]
|
| 52 |
73%|███████▎ | 22/30 [00:19<00:06, 1.14it/s]
|
| 53 |
77%|███████▋ | 23/30 [00:20<00:06, 1.14it/s]
|
| 54 |
80%|████████ | 24/30 [00:20<00:05, 1.14it/s]
|
| 55 |
83%|████████▎ | 25/30 [00:21<00:04, 1.14it/s]
|
| 56 |
87%|████████▋ | 26/30 [00:22<00:03, 1.14it/s]
|
| 57 |
90%|��████████ | 27/30 [00:23<00:02, 1.14it/s]
|
| 58 |
93%|█████████▎| 28/30 [00:24<00:01, 1.14it/s]
|
| 59 |
97%|█████████▋| 29/30 [00:25<00:00, 1.14it/s]
|
| 60 |
+
|
| 61 |
+
|
SCOPE_fastvideo/logs/example_1_scope_3step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_1/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_1/action.parquet --prompt 'A breathtaking view of a vibrant fantasy world spread across a vast, open valley, seen through the lens of an FPS perspective. The scene merges intricate architecture, lush terrain, and majestic cliffs, with the forefront focusing on a finely detailed, sleek assault rifle held firmly by a character in dark attire. The weapon, with its matte black finish and tactical design, gleams softly in the light. It is equipped with a reflex sight and a foregrip, ready for action. All of this is set under a vivid blue sky filled with puffy clouds and bathed in soft light.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_1/scope_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 31 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 32 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
| 33 |
+
|
| 34 |
+
|
SCOPE_fastvideo/logs/example_1_scope_lora_3step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_1/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_1/action.parquet --prompt 'A breathtaking view of a vibrant fantasy world spread across a vast, open valley, seen through the lens of an FPS perspective. The scene merges intricate architecture, lush terrain, and majestic cliffs, with the forefront focusing on a finely detailed, sleek assault rifle held firmly by a character in dark attire. The weapon, with its matte black finish and tactical design, gleams softly in the light. It is equipped with a reflex sight and a foregrip, ready for action. All of this is set under a vivid blue sky filled with puffy clouds and bathed in soft light.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_1/scope_lora_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode --fastvideo_lora_path_or_repo /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/fastvideo_dmd_lora/adapter_r32.safetensors --lora_strength 1.0
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 31 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 32 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
| 33 |
+
|
| 34 |
+
|
SCOPE_fastvideo/logs/example_1_scope_lora_vsa_3step.log
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/diffsynth/models/scope_dit.py:99: RuntimeWarning: SCOPE VSA kernel disabled for this runtime; falling back to full attention. device=NVIDIA H100 80GB HBM3, torch=2.5.0+cu124, triton=3.1.0, fastvideo-kernel=0.2.6. Set SCOPE_FORCE_VSA_KERNEL=1 to bypass this guard.
|
|
|
|
|
|
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_1/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_1/action.parquet --prompt 'A breathtaking view of a vibrant fantasy world spread across a vast, open valley, seen through the lens of an FPS perspective. The scene merges intricate architecture, lush terrain, and majestic cliffs, with the forefront focusing on a finely detailed, sleek assault rifle held firmly by a character in dark attire. The weapon, with its matte black finish and tactical design, gleams softly in the light. It is equipped with a reflex sight and a foregrip, ready for action. All of this is set under a vivid blue sky filled with puffy clouds and bathed in soft light.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_1/scope_lora_vsa_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode --fastvideo_lora_path_or_repo /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/fastvideo_dmd_lora/adapter_r32.safetensors --lora_strength 1.0 --use_vsa --vsa_sparsity 0.9
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/diffsynth/models/scope_dit.py:99: RuntimeWarning: SCOPE VSA kernel disabled for this runtime; falling back to full attention. device=NVIDIA H100 80GB HBM3, torch=2.5.0+cu124, triton=3.1.0, fastvideo-kernel=0.2.6. Set SCOPE_FORCE_VSA_KERNEL=1 to bypass this guard.
|
| 31 |
+
warnings.warn(
|
| 32 |
+
|
| 33 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 34 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
| 35 |
+
|
| 36 |
+
|
SCOPE_fastvideo/logs/example_2_scope_30step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/30 [00:00<?, ?it/s]
|
| 1 |
3%|▎ | 1/30 [00:00<00:25, 1.13it/s]
|
| 2 |
7%|▋ | 2/30 [00:01<00:24, 1.14it/s]
|
| 3 |
10%|█ | 3/30 [00:02<00:23, 1.15it/s]
|
| 4 |
13%|█▎ | 4/30 [00:03<00:22, 1.15it/s]
|
| 5 |
17%|█▋ | 5/30 [00:04<00:21, 1.15it/s]
|
| 6 |
20%|██ | 6/30 [00:05<00:20, 1.15it/s]
|
| 7 |
23%|██▎ | 7/30 [00:06<00:20, 1.15it/s]
|
| 8 |
27%|██▋ | 8/30 [00:06<00:19, 1.15it/s]
|
| 9 |
30%|███ | 9/30 [00:07<00:18, 1.15it/s]
|
| 10 |
33%|███▎ | 10/30 [00:08<00:17, 1.15it/s]
|
| 11 |
37%|███▋ | 11/30 [00:09<00:16, 1.15it/s]
|
| 12 |
40%|████ | 12/30 [00:10<00:15, 1.15it/s]
|
| 13 |
43%|████▎ | 13/30 [00:11<00:14, 1.14it/s]
|
| 14 |
47%|████▋ | 14/30 [00:12<00:13, 1.14it/s]
|
| 15 |
50%|█████ | 15/30 [00:13<00:13, 1.14it/s]
|
| 16 |
53%|█████▎ | 16/30 [00:13<00:12, 1.15it/s]
|
| 17 |
57%|█████▋ | 17/30 [00:14<00:11, 1.15it/s]
|
| 18 |
60%|██████ | 18/30 [00:15<00:10, 1.14it/s]
|
| 19 |
63%|██████▎ | 19/30 [00:16<00:09, 1.14it/s]
|
| 20 |
67%|██████▋ | 20/30 [00:17<00:08, 1.14it/s]
|
| 21 |
70%|███████ | 21/30 [00:18<00:07, 1.14it/s]
|
| 22 |
73%|███████▎ | 22/30 [00:19<00:06, 1.14it/s]
|
| 23 |
77%|███████▋ | 23/30 [00:20<00:06, 1.14it/s]
|
| 24 |
80%|████████ | 24/30 [00:20<00:05, 1.14it/s]
|
| 25 |
83%|████████▎ | 25/30 [00:21<00:04, 1.14it/s]
|
| 26 |
87%|████████▋ | 26/30 [00:22<00:03, 1.14it/s]
|
| 27 |
90%|█████████ | 27/30 [00:23<00:02, 1.14it/s]
|
| 28 |
93%|█████████▎| 28/30 [00:24<00:01, 1.14it/s]
|
| 29 |
97%|█████████▋| 29/30 [00:25<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_2/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_2/action.parquet --prompt 'In a mist-shrouded, rain-soaked mountain valley, a tactical rifle is held firmly in the player'"'"'s hand, aiming down a stone-paved path towards a monumental gate. The scene is filled with ancient stone statues, lanterns, and a distant Asian-style temple framed by towering cliffs, all shrouded in fog and lit by the glow of torches, creating an atmosphere of mystery and looming danger.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_2/scope_30step --height 480 --width 832 --max_frames 81 --num_inference_steps 30 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/30 [00:00<?, ?it/s]
|
| 31 |
3%|▎ | 1/30 [00:00<00:25, 1.13it/s]
|
| 32 |
7%|▋ | 2/30 [00:01<00:24, 1.14it/s]
|
| 33 |
10%|█ | 3/30 [00:02<00:23, 1.15it/s]
|
| 34 |
13%|█▎ | 4/30 [00:03<00:22, 1.15it/s]
|
| 35 |
17%|█▋ | 5/30 [00:04<00:21, 1.15it/s]
|
| 36 |
20%|██ | 6/30 [00:05<00:20, 1.15it/s]
|
| 37 |
23%|██▎ | 7/30 [00:06<00:20, 1.15it/s]
|
| 38 |
27%|██▋ | 8/30 [00:06<00:19, 1.15it/s]
|
| 39 |
30%|███ | 9/30 [00:07<00:18, 1.15it/s]
|
| 40 |
33%|███▎ | 10/30 [00:08<00:17, 1.15it/s]
|
| 41 |
37%|███▋ | 11/30 [00:09<00:16, 1.15it/s]
|
| 42 |
40%|████ | 12/30 [00:10<00:15, 1.15it/s]
|
| 43 |
43%|████▎ | 13/30 [00:11<00:14, 1.14it/s]
|
| 44 |
47%|████▋ | 14/30 [00:12<00:13, 1.14it/s]
|
| 45 |
50%|█████ | 15/30 [00:13<00:13, 1.14it/s]
|
| 46 |
53%|█████▎ | 16/30 [00:13<00:12, 1.15it/s]
|
| 47 |
57%|█████▋ | 17/30 [00:14<00:11, 1.15it/s]
|
| 48 |
60%|██████ | 18/30 [00:15<00:10, 1.14it/s]
|
| 49 |
63%|██████▎ | 19/30 [00:16<00:09, 1.14it/s]
|
| 50 |
67%|██████▋ | 20/30 [00:17<00:08, 1.14it/s]
|
| 51 |
70%|███████ | 21/30 [00:18<00:07, 1.14it/s]
|
| 52 |
73%|███████▎ | 22/30 [00:19<00:06, 1.14it/s]
|
| 53 |
77%|███████▋ | 23/30 [00:20<00:06, 1.14it/s]
|
| 54 |
80%|████████ | 24/30 [00:20<00:05, 1.14it/s]
|
| 55 |
83%|████████▎ | 25/30 [00:21<00:04, 1.14it/s]
|
| 56 |
87%|████████▋ | 26/30 [00:22<00:03, 1.14it/s]
|
| 57 |
90%|█████████ | 27/30 [00:23<00:02, 1.14it/s]
|
| 58 |
93%|█████████▎| 28/30 [00:24<00:01, 1.14it/s]
|
| 59 |
97%|█████████▋| 29/30 [00:25<00:00, 1.14it/s]
|
| 60 |
+
|
| 61 |
+
|
SCOPE_fastvideo/logs/example_2_scope_3step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.02it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.10it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_2/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_2/action.parquet --prompt 'In a mist-shrouded, rain-soaked mountain valley, a tactical rifle is held firmly in the player'"'"'s hand, aiming down a stone-paved path towards a monumental gate. The scene is filled with ancient stone statues, lanterns, and a distant Asian-style temple framed by towering cliffs, all shrouded in fog and lit by the glow of torches, creating an atmosphere of mystery and looming danger.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_2/scope_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 31 |
33%|███▎ | 1/3 [00:00<00:01, 1.02it/s]
|
| 32 |
67%|██████▋ | 2/3 [00:01<00:00, 1.10it/s]
|
| 33 |
+
|
| 34 |
+
|
SCOPE_fastvideo/logs/example_2_scope_lora_3step.log
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_2/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_2/action.parquet --prompt 'In a mist-shrouded, rain-soaked mountain valley, a tactical rifle is held firmly in the player'"'"'s hand, aiming down a stone-paved path towards a monumental gate. The scene is filled with ancient stone statues, lanterns, and a distant Asian-style temple framed by towering cliffs, all shrouded in fog and lit by the glow of torches, creating an atmosphere of mystery and looming danger.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_2/scope_lora_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode --fastvideo_lora_path_or_repo /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/fastvideo_dmd_lora/adapter_r32.safetensors --lora_strength 1.0
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]
|
| 31 |
33%|███▎ | 1/3 [00:00<00:01, 1.13it/s]
|
| 32 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
| 33 |
+
|
| 34 |
+
|
SCOPE_fastvideo/logs/example_2_scope_lora_vsa_3step.log
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 0 |
0%| | 0/3 [00:00<?, ?it/s]/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/diffsynth/models/scope_dit.py:99: RuntimeWarning: SCOPE VSA kernel disabled for this runtime; falling back to full attention. device=NVIDIA H100 80GB HBM3, torch=2.5.0+cu124, triton=3.1.0, fastvideo-kernel=0.2.6. Set SCOPE_FORCE_VSA_KERNEL=1 to bypass this guard.
|
|
|
|
|
|
|
| 1 |
33%|███▎ | 1/3 [00:00<00:01, 1.12it/s]
|
| 2 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
|
|
|
|
|
|
|
|
| 1 |
+
$ /lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/bin/python /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/inference.py --model_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE --input_image /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_2/image.png --action_path /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples/example_2/action.parquet --prompt 'In a mist-shrouded, rain-soaked mountain valley, a tactical rifle is held firmly in the player'"'"'s hand, aiming down a stone-paved path towards a monumental gate. The scene is filled with ancient stone statues, lanterns, and a distant Asian-style temple framed by towering cliffs, all shrouded in fog and lit by the glow of torches, creating an atmosphere of mystery and looming danger.' --output_dir /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/runs/example_2/scope_lora_vsa_3step --height 480 --width 832 --max_frames 81 --num_inference_steps 3 --seed 0 --fps 20 --dmd_denoising_steps 1000,757,522 --comparison_mode --fastvideo_lora_path_or_repo /lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/fastvideo_dmd_lora/adapter_r32.safetensors --lora_strength 1.0 --use_vsa --vsa_sparsity 0.9
|
| 2 |
+
|
| 3 |
+
/lustre/fsw/portfolios/nvr/users/shyang/miniconda3/envs/diffusion-pipe/lib/python3.12/site-packages/torch/cuda/__init__.py:61: FutureWarning: The pynvml package is deprecated. Please install nvidia-ml-py instead. If you did not install pynvml directly, please report this to the maintainers of the package that installed pynvml for you.
|
| 4 |
+
import pynvml # type: ignore[import]
|
| 5 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth"
|
| 6 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 7 |
+
Loaded model: {
|
| 8 |
+
"model_name": "wan_video_text_encoder",
|
| 9 |
+
"model_class": "diffsynth.models.wan_video_text_encoder.WanTextEncoder",
|
| 10 |
+
"extra_kwargs": null
|
| 11 |
+
}
|
| 12 |
+
Loading models from: "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth"
|
| 13 |
+
Detected non-safetensors files, which may cause slower loading. It's recommended to convert it to a safetensors file.
|
| 14 |
+
Loaded model: {
|
| 15 |
+
"model_name": "wan_video_vae",
|
| 16 |
+
"model_class": "diffsynth.models.wan_video_vae.WanVideoVAE38",
|
| 17 |
+
"extra_kwargs": null
|
| 18 |
+
}
|
| 19 |
+
Using wan_video_text_encoder from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/models_t5_umt5-xxl-enc-bf16.pth".
|
| 20 |
+
No wan_video_dit models available. This is not an error.
|
| 21 |
+
Using wan_video_vae from "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE/Wan2.2_VAE.pth".
|
| 22 |
+
No wan_video_image_encoder models available. This is not an error.
|
| 23 |
+
No wan_video_motion_controller models available. This is not an error.
|
| 24 |
+
No wan_video_vace models available. This is not an error.
|
| 25 |
+
No wan_video_vap models available. This is not an error.
|
| 26 |
+
No wans2v_audio_encoder models available. This is not an error.
|
| 27 |
+
No wan_video_animate_adapter models available. This is not an error.
|
| 28 |
+
|
| 29 |
+
|
| 30 |
0%| | 0/3 [00:00<?, ?it/s]/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/diffsynth/models/scope_dit.py:99: RuntimeWarning: SCOPE VSA kernel disabled for this runtime; falling back to full attention. device=NVIDIA H100 80GB HBM3, torch=2.5.0+cu124, triton=3.1.0, fastvideo-kernel=0.2.6. Set SCOPE_FORCE_VSA_KERNEL=1 to bypass this guard.
|
| 31 |
+
warnings.warn(
|
| 32 |
+
|
| 33 |
33%|███▎ | 1/3 [00:00<00:01, 1.12it/s]
|
| 34 |
67%|██████▋ | 2/3 [00:01<00:00, 1.14it/s]
|
| 35 |
+
|
| 36 |
+
|
SCOPE_fastvideo/manifest.json
ADDED
|
@@ -0,0 +1,254 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"cases": [
|
| 3 |
+
{
|
| 4 |
+
"case": "example_0",
|
| 5 |
+
"prompt": "In a whimsical, toy-inspired garden, the first-person view reveals a tactical weapon aimed forward along a sunlit path. The vibrant environment includes oversized flowers, colorful building blocks, and towering structures. In the foreground, a carved wooden sign reads ROSE GARDEN, accompanied by giant puzzle pieces scattered across the ground. The surroundings are filled with fairy lights hanging from trees, creating a warm, golden glow. The garden looks like a miniature diorama, set in a charming and playful atmosphere.",
|
| 6 |
+
"variants": [
|
| 7 |
+
{
|
| 8 |
+
"name": "scope_30step",
|
| 9 |
+
"label": "SCOPE 30-step",
|
| 10 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_0_scope_30step.mp4",
|
| 11 |
+
"seconds": 514.189,
|
| 12 |
+
"probe": {
|
| 13 |
+
"width": 832,
|
| 14 |
+
"height": 480,
|
| 15 |
+
"avg_frame_rate": "20.000000/1",
|
| 16 |
+
"nb_read_frames": "81",
|
| 17 |
+
"duration": 4.05,
|
| 18 |
+
"size_bytes": 2140754
|
| 19 |
+
},
|
| 20 |
+
"skipped_existing": false
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"name": "scope_3step",
|
| 24 |
+
"label": "SCOPE 3-step",
|
| 25 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_0_scope_3step.mp4",
|
| 26 |
+
"seconds": 81.669,
|
| 27 |
+
"probe": {
|
| 28 |
+
"width": 832,
|
| 29 |
+
"height": 480,
|
| 30 |
+
"avg_frame_rate": "20.000000/1",
|
| 31 |
+
"nb_read_frames": "81",
|
| 32 |
+
"duration": 4.05,
|
| 33 |
+
"size_bytes": 3109044
|
| 34 |
+
},
|
| 35 |
+
"skipped_existing": false
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"name": "scope_lora_3step",
|
| 39 |
+
"label": "SCOPE + DMD LoRA",
|
| 40 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_0_scope_lora_3step.mp4",
|
| 41 |
+
"seconds": 86.319,
|
| 42 |
+
"probe": {
|
| 43 |
+
"width": 832,
|
| 44 |
+
"height": 480,
|
| 45 |
+
"avg_frame_rate": "20.000000/1",
|
| 46 |
+
"nb_read_frames": "81",
|
| 47 |
+
"duration": 4.05,
|
| 48 |
+
"size_bytes": 2804542
|
| 49 |
+
},
|
| 50 |
+
"skipped_existing": false
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"name": "scope_lora_vsa_3step",
|
| 54 |
+
"label": "SCOPE + DMD LoRA + VSA",
|
| 55 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_0_scope_lora_vsa_3step.mp4",
|
| 56 |
+
"seconds": 83.593,
|
| 57 |
+
"probe": {
|
| 58 |
+
"width": 832,
|
| 59 |
+
"height": 480,
|
| 60 |
+
"avg_frame_rate": "20.000000/1",
|
| 61 |
+
"nb_read_frames": "81",
|
| 62 |
+
"duration": 4.05,
|
| 63 |
+
"size_bytes": 2804542
|
| 64 |
+
},
|
| 65 |
+
"skipped_existing": false
|
| 66 |
+
}
|
| 67 |
+
],
|
| 68 |
+
"comparison_path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/comparisons/example_0_comparison.mp4",
|
| 69 |
+
"comparison_probe": {
|
| 70 |
+
"width": 1664,
|
| 71 |
+
"height": 960,
|
| 72 |
+
"avg_frame_rate": "20.000000/1",
|
| 73 |
+
"nb_read_frames": "81",
|
| 74 |
+
"duration": 4.05,
|
| 75 |
+
"size_bytes": 25240204
|
| 76 |
+
}
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"case": "example_1",
|
| 80 |
+
"prompt": "A breathtaking view of a vibrant fantasy world spread across a vast, open valley, seen through the lens of an FPS perspective. The scene merges intricate architecture, lush terrain, and majestic cliffs, with the forefront focusing on a finely detailed, sleek assault rifle held firmly by a character in dark attire. The weapon, with its matte black finish and tactical design, gleams softly in the light. It is equipped with a reflex sight and a foregrip, ready for action. All of this is set under a vivid blue sky filled with puffy clouds and bathed in soft light.",
|
| 81 |
+
"variants": [
|
| 82 |
+
{
|
| 83 |
+
"name": "scope_30step",
|
| 84 |
+
"label": "SCOPE 30-step",
|
| 85 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_1_scope_30step.mp4",
|
| 86 |
+
"seconds": 102.503,
|
| 87 |
+
"probe": {
|
| 88 |
+
"width": 832,
|
| 89 |
+
"height": 480,
|
| 90 |
+
"avg_frame_rate": "20.000000/1",
|
| 91 |
+
"nb_read_frames": "81",
|
| 92 |
+
"duration": 4.05,
|
| 93 |
+
"size_bytes": 1022290
|
| 94 |
+
},
|
| 95 |
+
"skipped_existing": false
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"name": "scope_3step",
|
| 99 |
+
"label": "SCOPE 3-step",
|
| 100 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_1_scope_3step.mp4",
|
| 101 |
+
"seconds": 78.51,
|
| 102 |
+
"probe": {
|
| 103 |
+
"width": 832,
|
| 104 |
+
"height": 480,
|
| 105 |
+
"avg_frame_rate": "20.000000/1",
|
| 106 |
+
"nb_read_frames": "81",
|
| 107 |
+
"duration": 4.05,
|
| 108 |
+
"size_bytes": 1478917
|
| 109 |
+
},
|
| 110 |
+
"skipped_existing": false
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"name": "scope_lora_3step",
|
| 114 |
+
"label": "SCOPE + DMD LoRA",
|
| 115 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_1_scope_lora_3step.mp4",
|
| 116 |
+
"seconds": 83.359,
|
| 117 |
+
"probe": {
|
| 118 |
+
"width": 832,
|
| 119 |
+
"height": 480,
|
| 120 |
+
"avg_frame_rate": "20.000000/1",
|
| 121 |
+
"nb_read_frames": "81",
|
| 122 |
+
"duration": 4.05,
|
| 123 |
+
"size_bytes": 1196829
|
| 124 |
+
},
|
| 125 |
+
"skipped_existing": false
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"name": "scope_lora_vsa_3step",
|
| 129 |
+
"label": "SCOPE + DMD LoRA + VSA",
|
| 130 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_1_scope_lora_vsa_3step.mp4",
|
| 131 |
+
"seconds": 83.723,
|
| 132 |
+
"probe": {
|
| 133 |
+
"width": 832,
|
| 134 |
+
"height": 480,
|
| 135 |
+
"avg_frame_rate": "20.000000/1",
|
| 136 |
+
"nb_read_frames": "81",
|
| 137 |
+
"duration": 4.05,
|
| 138 |
+
"size_bytes": 1196829
|
| 139 |
+
},
|
| 140 |
+
"skipped_existing": false
|
| 141 |
+
}
|
| 142 |
+
],
|
| 143 |
+
"comparison_path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/comparisons/example_1_comparison.mp4",
|
| 144 |
+
"comparison_probe": {
|
| 145 |
+
"width": 1664,
|
| 146 |
+
"height": 960,
|
| 147 |
+
"avg_frame_rate": "20.000000/1",
|
| 148 |
+
"nb_read_frames": "81",
|
| 149 |
+
"duration": 4.05,
|
| 150 |
+
"size_bytes": 11494531
|
| 151 |
+
}
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"case": "example_2",
|
| 155 |
+
"prompt": "In a mist-shrouded, rain-soaked mountain valley, a tactical rifle is held firmly in the player's hand, aiming down a stone-paved path towards a monumental gate. The scene is filled with ancient stone statues, lanterns, and a distant Asian-style temple framed by towering cliffs, all shrouded in fog and lit by the glow of torches, creating an atmosphere of mystery and looming danger.",
|
| 156 |
+
"variants": [
|
| 157 |
+
{
|
| 158 |
+
"name": "scope_30step",
|
| 159 |
+
"label": "SCOPE 30-step",
|
| 160 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_2_scope_30step.mp4",
|
| 161 |
+
"seconds": 103.597,
|
| 162 |
+
"probe": {
|
| 163 |
+
"width": 832,
|
| 164 |
+
"height": 480,
|
| 165 |
+
"avg_frame_rate": "20.000000/1",
|
| 166 |
+
"nb_read_frames": "81",
|
| 167 |
+
"duration": 4.05,
|
| 168 |
+
"size_bytes": 836780
|
| 169 |
+
},
|
| 170 |
+
"skipped_existing": false
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"name": "scope_3step",
|
| 174 |
+
"label": "SCOPE 3-step",
|
| 175 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_2_scope_3step.mp4",
|
| 176 |
+
"seconds": 190.249,
|
| 177 |
+
"probe": {
|
| 178 |
+
"width": 832,
|
| 179 |
+
"height": 480,
|
| 180 |
+
"avg_frame_rate": "20.000000/1",
|
| 181 |
+
"nb_read_frames": "81",
|
| 182 |
+
"duration": 4.05,
|
| 183 |
+
"size_bytes": 780348
|
| 184 |
+
},
|
| 185 |
+
"skipped_existing": false
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"name": "scope_lora_3step",
|
| 189 |
+
"label": "SCOPE + DMD LoRA",
|
| 190 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_2_scope_lora_3step.mp4",
|
| 191 |
+
"seconds": 95.115,
|
| 192 |
+
"probe": {
|
| 193 |
+
"width": 832,
|
| 194 |
+
"height": 480,
|
| 195 |
+
"avg_frame_rate": "20.000000/1",
|
| 196 |
+
"nb_read_frames": "81",
|
| 197 |
+
"duration": 4.05,
|
| 198 |
+
"size_bytes": 869663
|
| 199 |
+
},
|
| 200 |
+
"skipped_existing": false
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"name": "scope_lora_vsa_3step",
|
| 204 |
+
"label": "SCOPE + DMD LoRA + VSA",
|
| 205 |
+
"path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/clips/example_2_scope_lora_vsa_3step.mp4",
|
| 206 |
+
"seconds": 84.021,
|
| 207 |
+
"probe": {
|
| 208 |
+
"width": 832,
|
| 209 |
+
"height": 480,
|
| 210 |
+
"avg_frame_rate": "20.000000/1",
|
| 211 |
+
"nb_read_frames": "81",
|
| 212 |
+
"duration": 4.05,
|
| 213 |
+
"size_bytes": 869663
|
| 214 |
+
},
|
| 215 |
+
"skipped_existing": false
|
| 216 |
+
}
|
| 217 |
+
],
|
| 218 |
+
"comparison_path": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full/comparisons/example_2_comparison.mp4",
|
| 219 |
+
"comparison_probe": {
|
| 220 |
+
"width": 1664,
|
| 221 |
+
"height": 960,
|
| 222 |
+
"avg_frame_rate": "20.000000/1",
|
| 223 |
+
"nb_read_frames": "81",
|
| 224 |
+
"duration": 4.05,
|
| 225 |
+
"size_bytes": 6476827
|
| 226 |
+
}
|
| 227 |
+
}
|
| 228 |
+
],
|
| 229 |
+
"settings": {
|
| 230 |
+
"scope_dir": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE",
|
| 231 |
+
"examples_dir": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/SCOPE/examples",
|
| 232 |
+
"scope_model_dir": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/SCOPE",
|
| 233 |
+
"fastvideo_lora_path_or_repo": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/models/fastvideo_dmd_lora/adapter_r32.safetensors",
|
| 234 |
+
"output_dir": "/lustre/fs1/portfolios/nvr/projects/nvr_elm_llm/users/shyang/FastVideo_LoRA/outputs/scope_fastvideo/full",
|
| 235 |
+
"case_ids": "0,1,2",
|
| 236 |
+
"smoke": false,
|
| 237 |
+
"skip_lora_variants": false,
|
| 238 |
+
"height": 480,
|
| 239 |
+
"width": 832,
|
| 240 |
+
"max_frames": 81,
|
| 241 |
+
"baseline_steps": 30,
|
| 242 |
+
"fast_steps": 3,
|
| 243 |
+
"seed": 0,
|
| 244 |
+
"fps": 20,
|
| 245 |
+
"lora_strength": 1.0,
|
| 246 |
+
"vsa_sparsity": 0.9,
|
| 247 |
+
"dmd_denoising_steps": "1000,757,522",
|
| 248 |
+
"upload": true,
|
| 249 |
+
"hf_repo_id": "Andyson/lora",
|
| 250 |
+
"hf_repo_type": "model",
|
| 251 |
+
"hf_path_in_repo": "SCOPE_fastvideo"
|
| 252 |
+
},
|
| 253 |
+
"expected_frames": 81
|
| 254 |
+
}
|