Upload 8 files
Browse files- paper-v6/image_dit_30k.pt +3 -0
- paper-v6/latent_f4t4d8_stats.json +28 -0
- paper-v6/mini_wan_40m_decode_30k.pt +3 -0
- paper-v6/pixel_dit_factorised_image_30k.pt +3 -0
- paper-v6/pixel_dit_factorised_random_30k.pt +3 -0
- paper-v6/pixel_dit_local_mixer_image_30k.pt +3 -0
- paper-v6/release_manifest.json +59 -0
- paper-v6/vae_f4t4d8_10k.pt +3 -0
paper-v6/image_dit_30k.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d3865d04e2f7f0660d13c56050af89f625f1d034d437b6a9c2c07b5ae4b084e9
|
| 3 |
+
size 159746547
|
paper-v6/latent_f4t4d8_stats.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"channels": 8,
|
| 3 |
+
"mean": [
|
| 4 |
+
0.4728064926616968,
|
| 5 |
+
0.253224215791297,
|
| 6 |
+
0.8443441166443078,
|
| 7 |
+
-1.157613306456275,
|
| 8 |
+
0.03358597843731022,
|
| 9 |
+
0.6394486125262936,
|
| 10 |
+
0.5051163805852871,
|
| 11 |
+
-0.08217054168672425
|
| 12 |
+
],
|
| 13 |
+
"std": [
|
| 14 |
+
0.31159355199510336,
|
| 15 |
+
0.39531605812855236,
|
| 16 |
+
0.685314486987186,
|
| 17 |
+
0.6057140898225262,
|
| 18 |
+
0.27991039660269335,
|
| 19 |
+
0.32068461999693293,
|
| 20 |
+
0.3639785196127743,
|
| 21 |
+
0.36329357178525307
|
| 22 |
+
],
|
| 23 |
+
"spatial_compression": 4,
|
| 24 |
+
"temporal_compression": 4,
|
| 25 |
+
"posterior": "mean",
|
| 26 |
+
"codec_file": "vae_f4t4d8_10k.pt",
|
| 27 |
+
"codec_sha256": "9d0d609616bf66cb084d461bc83830d26ff23f08883a1308a4bd15da4b7abf47"
|
| 28 |
+
}
|
paper-v6/mini_wan_40m_decode_30k.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:86902b9227d5f6e211b7d5f04ea8bade481304be0c8290718e575d0f43dbcc3a
|
| 3 |
+
size 157791253
|
paper-v6/pixel_dit_factorised_image_30k.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2b3faba9d1262b71d668e1f13678ed33e53c281a6d89cec754a26e5288db91da
|
| 3 |
+
size 159845727
|
paper-v6/pixel_dit_factorised_random_30k.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:112e275dd1243b7ec9df14dce4e2d9a3f823f6474db84864019d62362f4b2821
|
| 3 |
+
size 159845663
|
paper-v6/pixel_dit_local_mixer_image_30k.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d9b53e25f98ab6aa06b051cc9fa20d7816eae611d90c08b4c79af9d8ee500de9
|
| 3 |
+
size 167474643
|
paper-v6/release_manifest.json
ADDED
|
@@ -0,0 +1,59 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"release": "paper-v6",
|
| 3 |
+
"paper_title": "Dancing Stick Figures: An Introductory Dataset for Training Video Generation Models",
|
| 4 |
+
"dataset": "sprited/dancing-stick-figures",
|
| 5 |
+
"code": "https://github.com/sprited-ai/dancing-stick-figures",
|
| 6 |
+
"checkpoints": {
|
| 7 |
+
"image_dit_30k.pt": {
|
| 8 |
+
"source": "gin:/data/dancing-stick-figure-paper/results/paper1_v02c_t2i30k_s0/ckpt_030000.pt",
|
| 9 |
+
"sha256": "d3865d04e2f7f0660d13c56050af89f625f1d034d437b6a9c2c07b5ae4b084e9",
|
| 10 |
+
"step": 30000,
|
| 11 |
+
"parameters": 39900000
|
| 12 |
+
},
|
| 13 |
+
"pixel_dit_factorised_random_30k.pt": {
|
| 14 |
+
"source": "gin:/data/dancing-stick-figure-paper/results/paper1_v04_t2v64_factorised_random30k_rgba_b8a2_s0/ckpt_030000.pt",
|
| 15 |
+
"sha256": "112e275dd1243b7ec9df14dce4e2d9a3f823f6474db84864019d62362f4b2821",
|
| 16 |
+
"step": 30000,
|
| 17 |
+
"parameters": 39900000
|
| 18 |
+
},
|
| 19 |
+
"pixel_dit_factorised_image_30k.pt": {
|
| 20 |
+
"source": "gin:/data/dancing-stick-figure-paper/results/paper1_v04_t2v64_factorised_image30k_rgba_b8a2_s0/ckpt_030000.pt",
|
| 21 |
+
"sha256": "2b3faba9d1262b71d668e1f13678ed33e53c281a6d89cec754a26e5288db91da",
|
| 22 |
+
"step": 30000,
|
| 23 |
+
"parameters": 39900000,
|
| 24 |
+
"initialization_sha256": "d3865d04e2f7f0660d13c56050af89f625f1d034d437b6a9c2c07b5ae4b084e9"
|
| 25 |
+
},
|
| 26 |
+
"pixel_dit_local_mixer_image_30k.pt": {
|
| 27 |
+
"source": "gin:/data/dancing-stick-figure-paper/results/paper1_v03c_t2v64_local3d_image30k_rgba_s0/ckpt_030000.pt",
|
| 28 |
+
"sha256": "d9b53e25f98ab6aa06b051cc9fa20d7816eae611d90c08b4c79af9d8ee500de9",
|
| 29 |
+
"step": 30000,
|
| 30 |
+
"parameters": 41800000,
|
| 31 |
+
"initialization_sha256": "d3865d04e2f7f0660d13c56050af89f625f1d034d437b6a9c2c07b5ae4b084e9",
|
| 32 |
+
"note": "The run resumed from its verified 10k full-state checkpoint; the final EMA-only checkpoint records that resume path rather than the original image checkpoint."
|
| 33 |
+
},
|
| 34 |
+
"mini_wan_40m_decode_30k.pt": {
|
| 35 |
+
"source": "gin:/data/dancing-stick-figure-paper/results/paper1_v02c_wanmini40_dec_s0/ckpt_030000.pt",
|
| 36 |
+
"sha256": "86902b9227d5f6e211b7d5f04ea8bade481304be0c8290718e575d0f43dbcc3a",
|
| 37 |
+
"step": 30000,
|
| 38 |
+
"parameters": 39400000,
|
| 39 |
+
"decode_loss_weight": 1.0,
|
| 40 |
+
"codec": "vae_f4t4d8_10k.pt"
|
| 41 |
+
},
|
| 42 |
+
"vae_f4t4d8_10k.pt": {
|
| 43 |
+
"source": "gin:/data/dancing-stick-figure-paper/results/vae_f4t4d8_aug_10k/ckpt_010000.pt",
|
| 44 |
+
"sha256": "9d0d609616bf66cb084d461bc83830d26ff23f08883a1308a4bd15da4b7abf47",
|
| 45 |
+
"step": 10000,
|
| 46 |
+
"temporal_compression": 4,
|
| 47 |
+
"spatial_compression": 4,
|
| 48 |
+
"latent_channels": 8
|
| 49 |
+
}
|
| 50 |
+
},
|
| 51 |
+
"evaluation": {
|
| 52 |
+
"frames": 64,
|
| 53 |
+
"reference_stride": 1,
|
| 54 |
+
"held_out_source_motions": 128,
|
| 55 |
+
"sampling_seeds": [0, 1, 2],
|
| 56 |
+
"euler_steps": 50,
|
| 57 |
+
"cfg": 3
|
| 58 |
+
}
|
| 59 |
+
}
|
paper-v6/vae_f4t4d8_10k.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9d0d609616bf66cb084d461bc83830d26ff23f08883a1308a4bd15da4b7abf47
|
| 3 |
+
size 115806359
|