Hugging Face's logo

jefequien
/
S2PD

Image-to-Video
Diffusers
Safetensors
s2pd
video-generation
diffusion
evaluation
Model card Files
xet
 Community
1
jefequien commited on
Commit
ea1766e
·
0 Parent(s):

S2PD release files

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +35 -0
  2. README.md +86 -0
  3. checkpoints/BC/dit-b-pixel/2048/config.json +87 -0
  4. checkpoints/BC/dit-b-pixel/2048/model_ema.safetensors +3 -0
  5. checkpoints/BC/dit-b-pixel/SHA256SUMS +24 -0
  6. checkpoints/BC/dit-b-pixel/chess/config.json +87 -0
  7. checkpoints/BC/dit-b-pixel/chess/model_ema.safetensors +3 -0
  8. checkpoints/BC/dit-b-pixel/colliding_balls/config.json +87 -0
  9. checkpoints/BC/dit-b-pixel/colliding_balls/model_ema.safetensors +3 -0
  10. checkpoints/BC/dit-b-pixel/colliding_balls_3d/config.json +87 -0
  11. checkpoints/BC/dit-b-pixel/colliding_balls_3d/model_ema.safetensors +3 -0
  12. checkpoints/BC/dit-b-pixel/conway/config.json +87 -0
  13. checkpoints/BC/dit-b-pixel/conway/model_ema.safetensors +3 -0
  14. checkpoints/BC/dit-b-pixel/double_pendulum/config.json +87 -0
  15. checkpoints/BC/dit-b-pixel/double_pendulum/model_ema.safetensors +3 -0
  16. checkpoints/BC/dit-b-pixel/fifteen_puzzle/config.json +87 -0
  17. checkpoints/BC/dit-b-pixel/fifteen_puzzle/model_ema.safetensors +3 -0
  18. checkpoints/BC/dit-b-pixel/manifest.json +384 -0
  19. checkpoints/BC/dit-b-pixel/pong/config.json +87 -0
  20. checkpoints/BC/dit-b-pixel/pong/model_ema.safetensors +3 -0
  21. checkpoints/BC/dit-b-pixel/rubiks_3d/config.json +87 -0
  22. checkpoints/BC/dit-b-pixel/rubiks_3d/model_ema.safetensors +3 -0
  23. checkpoints/BC/dit-b-pixel/snake/config.json +87 -0
  24. checkpoints/BC/dit-b-pixel/snake/model_ema.safetensors +3 -0
  25. checkpoints/BC/dit-b-pixel/tetris/config.json +87 -0
  26. checkpoints/BC/dit-b-pixel/tetris/model_ema.safetensors +3 -0
  27. checkpoints/BC/dit-b-pixel/three_body/config.json +87 -0
  28. checkpoints/BC/dit-b-pixel/three_body/model_ema.safetensors +3 -0
  29. checkpoints/BC/wan-5b-lora/SHA256SUMS +8 -0
  30. checkpoints/BC/wan-5b-lora/kubric_movi_a/config.json +87 -0
  31. checkpoints/BC/wan-5b-lora/kubric_movi_a/model_ema.safetensors +3 -0
  32. checkpoints/BC/wan-5b-lora/kubric_movi_c/config.json +87 -0
  33. checkpoints/BC/wan-5b-lora/kubric_movi_c/model_ema.safetensors +3 -0
  34. checkpoints/BC/wan-5b-lora/manifest.json +144 -0
  35. checkpoints/BC/wan-5b-lora/mpm_worlds/config.json +87 -0
  36. checkpoints/BC/wan-5b-lora/mpm_worlds/model_ema.safetensors +3 -0
  37. checkpoints/BC/wan-5b-lora/rubiks_real/config.json +87 -0
  38. checkpoints/BC/wan-5b-lora/rubiks_real/model_ema.safetensors +3 -0
  39. checkpoints/BD/dit-b-pixel/2048/config.json +87 -0
  40. checkpoints/BD/dit-b-pixel/2048/model_ema.safetensors +3 -0
  41. checkpoints/BD/dit-b-pixel/SHA256SUMS +24 -0
  42. checkpoints/BD/dit-b-pixel/chess/config.json +87 -0
  43. checkpoints/BD/dit-b-pixel/chess/model_ema.safetensors +3 -0
  44. checkpoints/BD/dit-b-pixel/colliding_balls/config.json +87 -0
  45. checkpoints/BD/dit-b-pixel/colliding_balls/model_ema.safetensors +3 -0
  46. checkpoints/BD/dit-b-pixel/colliding_balls_3d/config.json +87 -0
  47. checkpoints/BD/dit-b-pixel/colliding_balls_3d/model_ema.safetensors +3 -0
  48. checkpoints/BD/dit-b-pixel/conway/config.json +87 -0
  49. checkpoints/BD/dit-b-pixel/conway/model_ema.safetensors +3 -0
  50. checkpoints/BD/dit-b-pixel/double_pendulum/config.json +87 -0
.gitattributes ADDED
@@ -0,0 +1,35 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ *.7z filter=lfs diff=lfs merge=lfs -text
2
+ *.arrow filter=lfs diff=lfs merge=lfs -text
3
+ *.bin filter=lfs diff=lfs merge=lfs -text
4
+ *.bz2 filter=lfs diff=lfs merge=lfs -text
5
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
6
+ *.ftz filter=lfs diff=lfs merge=lfs -text
7
+ *.gz filter=lfs diff=lfs merge=lfs -text
8
+ *.h5 filter=lfs diff=lfs merge=lfs -text
9
+ *.joblib filter=lfs diff=lfs merge=lfs -text
10
+ *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
+ *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
+ *.model filter=lfs diff=lfs merge=lfs -text
13
+ *.msgpack filter=lfs diff=lfs merge=lfs -text
14
+ *.npy filter=lfs diff=lfs merge=lfs -text
15
+ *.npz filter=lfs diff=lfs merge=lfs -text
16
+ *.onnx filter=lfs diff=lfs merge=lfs -text
17
+ *.ot filter=lfs diff=lfs merge=lfs -text
18
+ *.parquet filter=lfs diff=lfs merge=lfs -text
19
+ *.pb filter=lfs diff=lfs merge=lfs -text
20
+ *.pickle filter=lfs diff=lfs merge=lfs -text
21
+ *.pkl filter=lfs diff=lfs merge=lfs -text
22
+ *.pt filter=lfs diff=lfs merge=lfs -text
23
+ *.pth filter=lfs diff=lfs merge=lfs -text
24
+ *.rar filter=lfs diff=lfs merge=lfs -text
25
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
26
+ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
+ *.tar.* filter=lfs diff=lfs merge=lfs -text
28
+ *.tar filter=lfs diff=lfs merge=lfs -text
29
+ *.tflite filter=lfs diff=lfs merge=lfs -text
30
+ *.tgz filter=lfs diff=lfs merge=lfs -text
31
+ *.wasm filter=lfs diff=lfs merge=lfs -text
32
+ *.xz filter=lfs diff=lfs merge=lfs -text
33
+ *.zip filter=lfs diff=lfs merge=lfs -text
34
+ *.zst filter=lfs diff=lfs merge=lfs -text
35
+ *tfevents* filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,86 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: s2pd
3
+ tags:
4
+ - video-generation
5
+ - diffusion
6
+ - safetensors
7
+ - evaluation
8
+ ---
9
+
10
+ # Serial-to-Parallel Diffusion
11
+
12
+ Pretrained checkpoints for four methods: Bidirectional, causal Diffusion Forcing
13
+ Transformer (cDF), block-causal diffusion (BC), and S2PD.
14
+
15
+ Each cell links to the checkpoint.
16
+
17
+ ## DiT-B
18
+
19
+ Pixel-space models trained from scratch.
20
+
21
+ | Dataset | Bidirectional | cDF | BC | S2PD (Ours) |
22
+ | --- | --- | --- | --- | --- |
23
+ | Conway | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/conway) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/conway) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/conway) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/conway) |
24
+ | Chess | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/chess) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/chess) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/chess) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/chess) |
25
+ | 2048 | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/2048) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/2048) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/2048) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/2048) |
26
+ | Fifteen Puzzle | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/fifteen_puzzle) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/fifteen_puzzle) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/fifteen_puzzle) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/fifteen_puzzle) |
27
+ | Tetris | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/tetris) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/tetris) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/tetris) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/tetris) |
28
+ | Snake | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/snake) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/snake) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/snake) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/snake) |
29
+ | Rubik's Cube 3D | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/rubiks_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/rubiks_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/rubiks_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/rubiks_3d) |
30
+ | Double Pendulum | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/double_pendulum) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/double_pendulum) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/double_pendulum) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/double_pendulum) |
31
+ | Three Body | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/three_body) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/three_body) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/three_body) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/three_body) |
32
+ | Colliding Balls 2D | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/colliding_balls) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/colliding_balls) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/colliding_balls) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/colliding_balls) |
33
+ | Colliding Balls 3D | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/colliding_balls_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/colliding_balls_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/colliding_balls_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/colliding_balls_3d) |
34
+ | Pong | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/dit-b-pixel/pong) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/dit-b-pixel/pong) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/dit-b-pixel/pong) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/dit-b-pixel/pong) |
35
+
36
+ ## Wan-5B-LoRA
37
+
38
+ LoRA adapters for Wan-5B. These require the pretrained backbone; it is not included.
39
+
40
+ | Dataset | Bidirectional | cDF | BC | S2PD (Ours) |
41
+ | --- | --- | --- | --- | --- |
42
+ | Kubric MOVi-A | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/wan-5b-lora/kubric_movi_a) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/wan-5b-lora/kubric_movi_a) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/wan-5b-lora/kubric_movi_a) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/wan-5b-lora/kubric_movi_a) |
43
+ | Kubric MOVi-C | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/wan-5b-lora/kubric_movi_c) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/wan-5b-lora/kubric_movi_c) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/wan-5b-lora/kubric_movi_c) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/wan-5b-lora/kubric_movi_c) |
44
+ | MPMWorlds | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/wan-5b-lora/mpm_worlds) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/wan-5b-lora/mpm_worlds) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/wan-5b-lora/mpm_worlds) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/wan-5b-lora/mpm_worlds) |
45
+ | Rubik's Cube Real | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BD/wan-5b-lora/rubiks_real) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/CDF/wan-5b-lora/rubiks_real) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/BC/wan-5b-lora/rubiks_real) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/S2PD/wan-5b-lora/rubiks_real) |
46
+
47
+ <details>
48
+ <summary><strong>Token-order ablations (DiT-B)</strong></summary>
49
+
50
+ Separate S2PD runs using 8-token/full-video training blocks, weighted 80/20.
51
+
52
+ | Dataset | Raster | Zigzag | Hilbert |
53
+ | --- | --- | --- | --- |
54
+ | Conway | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/conway) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/conway) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/conway) |
55
+ | Chess | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/chess) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/chess) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/chess) |
56
+ | 2048 | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/2048) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/2048) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/2048) |
57
+ | Fifteen Puzzle | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/fifteen_puzzle) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/fifteen_puzzle) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/fifteen_puzzle) |
58
+ | Tetris | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/tetris) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/tetris) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/tetris) |
59
+ | Snake | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/snake) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/snake) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/snake) |
60
+ | Rubik's Cube 3D | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/rubiks_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/rubiks_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/rubiks_3d) |
61
+ | Double Pendulum | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/double_pendulum) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/double_pendulum) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/double_pendulum) |
62
+ | Three Body | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/three_body) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/three_body) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/three_body) |
63
+ | Colliding Balls 2D | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/colliding_balls) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/colliding_balls) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/colliding_balls) |
64
+ | Colliding Balls 3D | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/colliding_balls_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/colliding_balls_3d) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/colliding_balls_3d) |
65
+ | Pong | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/raster/dit-b-pixel/pong) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/zigzag/dit-b-pixel/pong) | [Ckpt](https://huggingface.co/jefequien/S2PD/tree/main/checkpoints/ablations/token-order/hilbert/dit-b-pixel/pong) |
66
+
67
+ </details>
68
+
69
+ ## Download
70
+
71
+ For example, download the S2PD Conway checkpoint:
72
+
73
+ ```bash
74
+ hf download jefequien/S2PD --include 'checkpoints/S2PD/dit-b-pixel/conway/*' --local-dir .
75
+ ```
76
+
77
+ Each folder contains EMA weights (`model_ema.safetensors`) and a resolved
78
+ `config.json` for use with the S2PD code. These are inference checkpoints, not
79
+ training-resume checkpoints. Family manifests retain checksums and provenance.
80
+
81
+ ## Datasets
82
+
83
+ - [Rubik’s Cube Real](https://huggingface.co/jefequien/S2PD/tree/main/data/rubiks_real): 5,000 training and 512 validation clips.
84
+ - [Chess](https://huggingface.co/jefequien/S2PD/tree/main/data/chess): 1 million training and 10,000 validation games from the Lichess open database (CC0-1.0).
85
+
86
+
checkpoints/BC/dit-b-pixel/2048/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "2048",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6168215",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_2048_slurm_6168215",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_2048_slurm_6168215"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/2048/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8f546655b8e1ffcbcc0b32b9e88b72e09ac23b070cfac24263e047a130acca92
3
+ size 530702616
checkpoints/BC/dit-b-pixel/SHA256SUMS ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 444dab2d82375c45c1d0eefb4ffe557a1cb5a3049b5e9388ee2fbf625174c36f conway/config.json
2
+ cd81a573a38ebe19955b1c33195c3ecb00d8e176954bd80228fec3421bc0f4b1 chess/config.json
3
+ 3e71cf05d246f9fcff4848d61e67c4aa23adf9d6b417d9eb28e03bdb233add20 fifteen_puzzle/config.json
4
+ 31eb2abf5dcb7be97ee9bfd10cfea49b647c8de88628d53d6b6127e1fe10673f 2048/config.json
5
+ ddec09342c22f1e4b6c5acf095ed7d6ae241b428d97a5b49b358bd94f0075514 tetris/config.json
6
+ 73946e518ee155cd241105287fa6c49a4f1faab2d0733f19903beb742dd535f1 snake/config.json
7
+ f09c082ab1e558f09dbb0a1cec168265930c6b0867870f3678b4416e1f787f11 double_pendulum/config.json
8
+ a5b227df1e4b1d84bedabcdf47a24c0eca70904ea27b35c0e6dac14cde15b982 three_body/config.json
9
+ 813363cc1941ea24b7347e800c867e5267bb786cf89f00e60ca1aeaea7335514 pong/config.json
10
+ b8bf7633280c4490bc0104ad44fb501fb83e30bd94893eeaf16271f187d33c7a colliding_balls/config.json
11
+ 93ab29c26cda5a3453fa51f8585840f870f021b27deb3607e52b81e8d5735ff3 colliding_balls_3d/config.json
12
+ c73a1b867cdffdef9aeef04ed5210c56d4d106fe2473b069ea6fae00adc24c76 rubiks_3d/config.json
13
+ 91a2da56c23e9008d3789ee8649d6b94a883b15bb5919f6b75d62c7d05c17e35 conway/model_ema.safetensors
14
+ ddf210436697ad7b41711384496eecc048da09107ef5e354ea183be2c280bc5d chess/model_ema.safetensors
15
+ 75609febc339ab14dd466ace04231544cd79b7628f53079270cc93f13ea0954d fifteen_puzzle/model_ema.safetensors
16
+ 8f546655b8e1ffcbcc0b32b9e88b72e09ac23b070cfac24263e047a130acca92 2048/model_ema.safetensors
17
+ 4eef634cc5a8b599b5c323e2c6c976c2cb7ee58c73860af2f2c3e12419190732 tetris/model_ema.safetensors
18
+ beb3ebcd0cf3ae5bb47c036f0311a203b0bb4516096aeecfe2b615d46137b34a snake/model_ema.safetensors
19
+ 9aa9e9cb5571133f0aa19ef0725c794b062c1e6956721fbdd951aa2a5c57bcf6 double_pendulum/model_ema.safetensors
20
+ 1019f8b41e80a544fdeb0906a684d2206b9e53f83f794a09ae8d5a1579d53725 three_body/model_ema.safetensors
21
+ cec52749b00aea7e24cb83b7bf1668ba0790c65a3f91242e01ec132ee83ab691 pong/model_ema.safetensors
22
+ 52bf7a95a6904f52c084292a4454d54b2d20ec727c04ed5bf81a575c01a6069d colliding_balls/model_ema.safetensors
23
+ 226c8689edf3adf79f04bc15933c642c9062bd80b7b26b25df0ff86c0abd6127 colliding_balls_3d/model_ema.safetensors
24
+ 72e36eabb0a94eee348fe0f5488baca0e600cb9d621e34c0ae0d3520b91624a2 rubiks_3d/model_ema.safetensors
checkpoints/BC/dit-b-pixel/chess/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "chess",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6168205",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_chess_slurm_6168205",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_chess_slurm_6168205"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/chess/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ddf210436697ad7b41711384496eecc048da09107ef5e354ea183be2c280bc5d
3
+ size 530702616
checkpoints/BC/dit-b-pixel/colliding_balls/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "colliding_balls",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6130789",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_colliding_balls_slurm_6130789",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_colliding_balls_slurm_6130789"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/colliding_balls/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52bf7a95a6904f52c084292a4454d54b2d20ec727c04ed5bf81a575c01a6069d
3
+ size 530702680
checkpoints/BC/dit-b-pixel/colliding_balls_3d/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "colliding_balls_3d",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6130791",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_colliding_balls_3d_slurm_6130791",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_colliding_balls_3d_slurm_6130791"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/colliding_balls_3d/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:226c8689edf3adf79f04bc15933c642c9062bd80b7b26b25df0ff86c0abd6127
3
+ size 530702680
checkpoints/BC/dit-b-pixel/conway/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "conway",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6130767",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_conway_slurm_6130767",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_conway_slurm_6130767"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/conway/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91a2da56c23e9008d3789ee8649d6b94a883b15bb5919f6b75d62c7d05c17e35
3
+ size 530702672
checkpoints/BC/dit-b-pixel/double_pendulum/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "double_pendulum",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6130781",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_double_pendulum_slurm_6130781",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_double_pendulum_slurm_6130781"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/double_pendulum/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9aa9e9cb5571133f0aa19ef0725c794b062c1e6956721fbdd951aa2a5c57bcf6
3
+ size 530702680
checkpoints/BC/dit-b-pixel/fifteen_puzzle/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "fifteen_puzzle",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6168210",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_fifteen_puzzle_slurm_6168210",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_fifteen_puzzle_slurm_6168210"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/fifteen_puzzle/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:75609febc339ab14dd466ace04231544cd79b7628f53079270cc93f13ea0954d
3
+ size 530702624
checkpoints/BC/dit-b-pixel/manifest.json ADDED
@@ -0,0 +1,384 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "archive_code": {
3
+ "commit": "ebf1597b1246aedd27ede2bdaa23837628ee234a",
4
+ "repository": "https://github.com/jefequien/S2PD"
5
+ },
6
+ "archive_format_version": 1,
7
+ "diffusion_mode": "BC",
8
+ "family": "dit-b-pixel",
9
+ "intended_use": [
10
+ "evaluation",
11
+ "paper plots and visualizations",
12
+ "checkpoint transfer"
13
+ ],
14
+ "model": "DiT-B",
15
+ "repository": "jefequien/S2PD",
16
+ "representation": "pixel",
17
+ "storage_dtype": "float32",
18
+ "total_weight_file_bytes": 6368431760,
19
+ "training_resume_supported": false,
20
+ "variant_count": 12,
21
+ "variants": [
22
+ {
23
+ "config": {
24
+ "file": "checkpoints/BC/dit-b-pixel/conway/config.json",
25
+ "file_bytes": 2051,
26
+ "sha256": "444dab2d82375c45c1d0eefb4ffe557a1cb5a3049b5e9388ee2fbf625174c36f"
27
+ },
28
+ "dataset": "conway",
29
+ "directory": "checkpoints/BC/dit-b-pixel/conway",
30
+ "display_name": "Conway's Game of Life",
31
+ "training": {
32
+ "epoch_index": 99,
33
+ "git_commit": "d614ef7ae14bd96b14111329a8a7a93cb79b2f65",
34
+ "global_step": 100000,
35
+ "slurm_job_id": "6130767",
36
+ "source_checkpoint_bytes": 2123118179,
37
+ "source_run": "dit-b-pixel-bc-raster_conway_slurm_6130767"
38
+ },
39
+ "weights": {
40
+ "dtypes": [
41
+ "float32"
42
+ ],
43
+ "field": "model_ema",
44
+ "file": "checkpoints/BC/dit-b-pixel/conway/model_ema.safetensors",
45
+ "file_bytes": 530702672,
46
+ "format": "s2pd-ema-v1",
47
+ "sha256": "91a2da56c23e9008d3789ee8649d6b94a883b15bb5919f6b75d62c7d05c17e35",
48
+ "tensor_bytes": 530672640,
49
+ "tensor_count": 292
50
+ }
51
+ },
52
+ {
53
+ "config": {
54
+ "file": "checkpoints/BC/dit-b-pixel/chess/config.json",
55
+ "file_bytes": 2048,
56
+ "sha256": "cd81a573a38ebe19955b1c33195c3ecb00d8e176954bd80228fec3421bc0f4b1"
57
+ },
58
+ "dataset": "chess",
59
+ "directory": "checkpoints/BC/dit-b-pixel/chess",
60
+ "display_name": "Chess",
61
+ "training": {
62
+ "epoch_index": 99,
63
+ "git_commit": "ebf1597b1246aedd27ede2bdaa23837628ee234a",
64
+ "global_step": 100000,
65
+ "slurm_job_id": "6168205",
66
+ "source_checkpoint_bytes": 2123118179,
67
+ "source_run": "dit-b-pixel-bc-raster_chess_slurm_6168205"
68
+ },
69
+ "weights": {
70
+ "dtypes": [
71
+ "float32"
72
+ ],
73
+ "field": "model_ema",
74
+ "file": "checkpoints/BC/dit-b-pixel/chess/model_ema.safetensors",
75
+ "file_bytes": 530702616,
76
+ "format": "s2pd-ema-v1",
77
+ "sha256": "ddf210436697ad7b41711384496eecc048da09107ef5e354ea183be2c280bc5d",
78
+ "tensor_bytes": 530672640,
79
+ "tensor_count": 292
80
+ }
81
+ },
82
+ {
83
+ "config": {
84
+ "file": "checkpoints/BC/dit-b-pixel/fifteen_puzzle/config.json",
85
+ "file_bytes": 2075,
86
+ "sha256": "3e71cf05d246f9fcff4848d61e67c4aa23adf9d6b417d9eb28e03bdb233add20"
87
+ },
88
+ "dataset": "fifteen_puzzle",
89
+ "directory": "checkpoints/BC/dit-b-pixel/fifteen_puzzle",
90
+ "display_name": "Fifteen Puzzle",
91
+ "training": {
92
+ "epoch_index": 99,
93
+ "git_commit": "ebf1597b1246aedd27ede2bdaa23837628ee234a",
94
+ "global_step": 100000,
95
+ "slurm_job_id": "6168210",
96
+ "source_checkpoint_bytes": 2123118179,
97
+ "source_run": "dit-b-pixel-bc-raster_fifteen_puzzle_slurm_6168210"
98
+ },
99
+ "weights": {
100
+ "dtypes": [
101
+ "float32"
102
+ ],
103
+ "field": "model_ema",
104
+ "file": "checkpoints/BC/dit-b-pixel/fifteen_puzzle/model_ema.safetensors",
105
+ "file_bytes": 530702624,
106
+ "format": "s2pd-ema-v1",
107
+ "sha256": "75609febc339ab14dd466ace04231544cd79b7628f53079270cc93f13ea0954d",
108
+ "tensor_bytes": 530672640,
109
+ "tensor_count": 292
110
+ }
111
+ },
112
+ {
113
+ "config": {
114
+ "file": "checkpoints/BC/dit-b-pixel/2048/config.json",
115
+ "file_bytes": 2045,
116
+ "sha256": "31eb2abf5dcb7be97ee9bfd10cfea49b647c8de88628d53d6b6127e1fe10673f"
117
+ },
118
+ "dataset": "2048",
119
+ "directory": "checkpoints/BC/dit-b-pixel/2048",
120
+ "display_name": "2048",
121
+ "training": {
122
+ "epoch_index": 99,
123
+ "git_commit": "ebf1597b1246aedd27ede2bdaa23837628ee234a",
124
+ "global_step": 100000,
125
+ "slurm_job_id": "6168215",
126
+ "source_checkpoint_bytes": 2123118179,
127
+ "source_run": "dit-b-pixel-bc-raster_2048_slurm_6168215"
128
+ },
129
+ "weights": {
130
+ "dtypes": [
131
+ "float32"
132
+ ],
133
+ "field": "model_ema",
134
+ "file": "checkpoints/BC/dit-b-pixel/2048/model_ema.safetensors",
135
+ "file_bytes": 530702616,
136
+ "format": "s2pd-ema-v1",
137
+ "sha256": "8f546655b8e1ffcbcc0b32b9e88b72e09ac23b070cfac24263e047a130acca92",
138
+ "tensor_bytes": 530672640,
139
+ "tensor_count": 292
140
+ }
141
+ },
142
+ {
143
+ "config": {
144
+ "file": "checkpoints/BC/dit-b-pixel/tetris/config.json",
145
+ "file_bytes": 2051,
146
+ "sha256": "ddec09342c22f1e4b6c5acf095ed7d6ae241b428d97a5b49b358bd94f0075514"
147
+ },
148
+ "dataset": "tetris",
149
+ "directory": "checkpoints/BC/dit-b-pixel/tetris",
150
+ "display_name": "Tetris",
151
+ "training": {
152
+ "epoch_index": 99,
153
+ "git_commit": "ebf1597b1246aedd27ede2bdaa23837628ee234a",
154
+ "global_step": 100000,
155
+ "slurm_job_id": "6168223",
156
+ "source_checkpoint_bytes": 2123118179,
157
+ "source_run": "dit-b-pixel-bc-raster_tetris_slurm_6168223"
158
+ },
159
+ "weights": {
160
+ "dtypes": [
161
+ "float32"
162
+ ],
163
+ "field": "model_ema",
164
+ "file": "checkpoints/BC/dit-b-pixel/tetris/model_ema.safetensors",
165
+ "file_bytes": 530702616,
166
+ "format": "s2pd-ema-v1",
167
+ "sha256": "4eef634cc5a8b599b5c323e2c6c976c2cb7ee58c73860af2f2c3e12419190732",
168
+ "tensor_bytes": 530672640,
169
+ "tensor_count": 292
170
+ }
171
+ },
172
+ {
173
+ "config": {
174
+ "file": "checkpoints/BC/dit-b-pixel/snake/config.json",
175
+ "file_bytes": 2048,
176
+ "sha256": "73946e518ee155cd241105287fa6c49a4f1faab2d0733f19903beb742dd535f1"
177
+ },
178
+ "dataset": "snake",
179
+ "directory": "checkpoints/BC/dit-b-pixel/snake",
180
+ "display_name": "Snake",
181
+ "training": {
182
+ "epoch_index": 99,
183
+ "git_commit": "ebf1597b1246aedd27ede2bdaa23837628ee234a",
184
+ "global_step": 100000,
185
+ "slurm_job_id": "6168228",
186
+ "source_checkpoint_bytes": 2123118179,
187
+ "source_run": "dit-b-pixel-bc-raster_snake_slurm_6168228"
188
+ },
189
+ "weights": {
190
+ "dtypes": [
191
+ "float32"
192
+ ],
193
+ "field": "model_ema",
194
+ "file": "checkpoints/BC/dit-b-pixel/snake/model_ema.safetensors",
195
+ "file_bytes": 530702616,
196
+ "format": "s2pd-ema-v1",
197
+ "sha256": "beb3ebcd0cf3ae5bb47c036f0311a203b0bb4516096aeecfe2b615d46137b34a",
198
+ "tensor_bytes": 530672640,
199
+ "tensor_count": 292
200
+ }
201
+ },
202
+ {
203
+ "config": {
204
+ "file": "checkpoints/BC/dit-b-pixel/double_pendulum/config.json",
205
+ "file_bytes": 2078,
206
+ "sha256": "f09c082ab1e558f09dbb0a1cec168265930c6b0867870f3678b4416e1f787f11"
207
+ },
208
+ "dataset": "double_pendulum",
209
+ "directory": "checkpoints/BC/dit-b-pixel/double_pendulum",
210
+ "display_name": "Double Pendulum",
211
+ "training": {
212
+ "epoch_index": 99,
213
+ "git_commit": "d614ef7ae14bd96b14111329a8a7a93cb79b2f65",
214
+ "global_step": 100000,
215
+ "slurm_job_id": "6130781",
216
+ "source_checkpoint_bytes": 2123118179,
217
+ "source_run": "dit-b-pixel-bc-raster_double_pendulum_slurm_6130781"
218
+ },
219
+ "weights": {
220
+ "dtypes": [
221
+ "float32"
222
+ ],
223
+ "field": "model_ema",
224
+ "file": "checkpoints/BC/dit-b-pixel/double_pendulum/model_ema.safetensors",
225
+ "file_bytes": 530702680,
226
+ "format": "s2pd-ema-v1",
227
+ "sha256": "9aa9e9cb5571133f0aa19ef0725c794b062c1e6956721fbdd951aa2a5c57bcf6",
228
+ "tensor_bytes": 530672640,
229
+ "tensor_count": 292
230
+ }
231
+ },
232
+ {
233
+ "config": {
234
+ "file": "checkpoints/BC/dit-b-pixel/three_body/config.json",
235
+ "file_bytes": 2063,
236
+ "sha256": "a5b227df1e4b1d84bedabcdf47a24c0eca70904ea27b35c0e6dac14cde15b982"
237
+ },
238
+ "dataset": "three_body",
239
+ "directory": "checkpoints/BC/dit-b-pixel/three_body",
240
+ "display_name": "Three Body",
241
+ "training": {
242
+ "epoch_index": 99,
243
+ "git_commit": "d614ef7ae14bd96b14111329a8a7a93cb79b2f65",
244
+ "global_step": 100000,
245
+ "slurm_job_id": "6130783",
246
+ "source_checkpoint_bytes": 2123118179,
247
+ "source_run": "dit-b-pixel-bc-raster_three_body_slurm_6130783"
248
+ },
249
+ "weights": {
250
+ "dtypes": [
251
+ "float32"
252
+ ],
253
+ "field": "model_ema",
254
+ "file": "checkpoints/BC/dit-b-pixel/three_body/model_ema.safetensors",
255
+ "file_bytes": 530702672,
256
+ "format": "s2pd-ema-v1",
257
+ "sha256": "1019f8b41e80a544fdeb0906a684d2206b9e53f83f794a09ae8d5a1579d53725",
258
+ "tensor_bytes": 530672640,
259
+ "tensor_count": 292
260
+ }
261
+ },
262
+ {
263
+ "config": {
264
+ "file": "checkpoints/BC/dit-b-pixel/pong/config.json",
265
+ "file_bytes": 2045,
266
+ "sha256": "813363cc1941ea24b7347e800c867e5267bb786cf89f00e60ca1aeaea7335514"
267
+ },
268
+ "dataset": "pong",
269
+ "directory": "checkpoints/BC/dit-b-pixel/pong",
270
+ "display_name": "Pong",
271
+ "training": {
272
+ "epoch_index": 99,
273
+ "git_commit": "d614ef7ae14bd96b14111329a8a7a93cb79b2f65",
274
+ "global_step": 100000,
275
+ "slurm_job_id": "6130785",
276
+ "source_checkpoint_bytes": 2123118179,
277
+ "source_run": "dit-b-pixel-bc-raster_pong_slurm_6130785"
278
+ },
279
+ "weights": {
280
+ "dtypes": [
281
+ "float32"
282
+ ],
283
+ "field": "model_ema",
284
+ "file": "checkpoints/BC/dit-b-pixel/pong/model_ema.safetensors",
285
+ "file_bytes": 530702664,
286
+ "format": "s2pd-ema-v1",
287
+ "sha256": "cec52749b00aea7e24cb83b7bf1668ba0790c65a3f91242e01ec132ee83ab691",
288
+ "tensor_bytes": 530672640,
289
+ "tensor_count": 292
290
+ }
291
+ },
292
+ {
293
+ "config": {
294
+ "file": "checkpoints/BC/dit-b-pixel/colliding_balls/config.json",
295
+ "file_bytes": 2078,
296
+ "sha256": "b8bf7633280c4490bc0104ad44fb501fb83e30bd94893eeaf16271f187d33c7a"
297
+ },
298
+ "dataset": "colliding_balls",
299
+ "directory": "checkpoints/BC/dit-b-pixel/colliding_balls",
300
+ "display_name": "Colliding Balls",
301
+ "training": {
302
+ "epoch_index": 99,
303
+ "git_commit": "d614ef7ae14bd96b14111329a8a7a93cb79b2f65",
304
+ "global_step": 100000,
305
+ "slurm_job_id": "6130789",
306
+ "source_checkpoint_bytes": 2123118179,
307
+ "source_run": "dit-b-pixel-bc-raster_colliding_balls_slurm_6130789"
308
+ },
309
+ "weights": {
310
+ "dtypes": [
311
+ "float32"
312
+ ],
313
+ "field": "model_ema",
314
+ "file": "checkpoints/BC/dit-b-pixel/colliding_balls/model_ema.safetensors",
315
+ "file_bytes": 530702680,
316
+ "format": "s2pd-ema-v1",
317
+ "sha256": "52bf7a95a6904f52c084292a4454d54b2d20ec727c04ed5bf81a575c01a6069d",
318
+ "tensor_bytes": 530672640,
319
+ "tensor_count": 292
320
+ }
321
+ },
322
+ {
323
+ "config": {
324
+ "file": "checkpoints/BC/dit-b-pixel/colliding_balls_3d/config.json",
325
+ "file_bytes": 2087,
326
+ "sha256": "93ab29c26cda5a3453fa51f8585840f870f021b27deb3607e52b81e8d5735ff3"
327
+ },
328
+ "dataset": "colliding_balls_3d",
329
+ "directory": "checkpoints/BC/dit-b-pixel/colliding_balls_3d",
330
+ "display_name": "Colliding Balls 3D",
331
+ "training": {
332
+ "epoch_index": 99,
333
+ "git_commit": "d614ef7ae14bd96b14111329a8a7a93cb79b2f65",
334
+ "global_step": 100000,
335
+ "slurm_job_id": "6130791",
336
+ "source_checkpoint_bytes": 2123118179,
337
+ "source_run": "dit-b-pixel-bc-raster_colliding_balls_3d_slurm_6130791"
338
+ },
339
+ "weights": {
340
+ "dtypes": [
341
+ "float32"
342
+ ],
343
+ "field": "model_ema",
344
+ "file": "checkpoints/BC/dit-b-pixel/colliding_balls_3d/model_ema.safetensors",
345
+ "file_bytes": 530702680,
346
+ "format": "s2pd-ema-v1",
347
+ "sha256": "226c8689edf3adf79f04bc15933c642c9062bd80b7b26b25df0ff86c0abd6127",
348
+ "tensor_bytes": 530672640,
349
+ "tensor_count": 292
350
+ }
351
+ },
352
+ {
353
+ "config": {
354
+ "file": "checkpoints/BC/dit-b-pixel/rubiks_3d/config.json",
355
+ "file_bytes": 2060,
356
+ "sha256": "c73a1b867cdffdef9aeef04ed5210c56d4d106fe2473b069ea6fae00adc24c76"
357
+ },
358
+ "dataset": "rubiks_3d",
359
+ "directory": "checkpoints/BC/dit-b-pixel/rubiks_3d",
360
+ "display_name": "Rubik's Cube 3D",
361
+ "training": {
362
+ "epoch_index": 99,
363
+ "git_commit": "ebf1597b1246aedd27ede2bdaa23837628ee234a",
364
+ "global_step": 100000,
365
+ "slurm_job_id": "6168254",
366
+ "source_checkpoint_bytes": 2123118179,
367
+ "source_run": "dit-b-pixel-bc-raster_rubiks_3d_slurm_6168254"
368
+ },
369
+ "weights": {
370
+ "dtypes": [
371
+ "float32"
372
+ ],
373
+ "field": "model_ema",
374
+ "file": "checkpoints/BC/dit-b-pixel/rubiks_3d/model_ema.safetensors",
375
+ "file_bytes": 530702624,
376
+ "format": "s2pd-ema-v1",
377
+ "sha256": "72e36eabb0a94eee348fe0f5488baca0e600cb9d621e34c0ae0d3520b91624a2",
378
+ "tensor_bytes": 530672640,
379
+ "tensor_count": 292
380
+ }
381
+ }
382
+ ],
383
+ "weights": "model_ema"
384
+ }
checkpoints/BC/dit-b-pixel/pong/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "pong",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6130785",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_pong_slurm_6130785",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_pong_slurm_6130785"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/pong/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cec52749b00aea7e24cb83b7bf1668ba0790c65a3f91242e01ec132ee83ab691
3
+ size 530702664
checkpoints/BC/dit-b-pixel/rubiks_3d/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "rubiks_3d",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6168254",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_rubiks_3d_slurm_6168254",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_rubiks_3d_slurm_6168254"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/rubiks_3d/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72e36eabb0a94eee348fe0f5488baca0e600cb9d621e34c0ae0d3520b91624a2
3
+ size 530702624
checkpoints/BC/dit-b-pixel/snake/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "snake",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6168228",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_snake_slurm_6168228",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_snake_slurm_6168228"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/snake/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:beb3ebcd0cf3ae5bb47c036f0311a203b0bb4516096aeecfe2b615d46137b34a
3
+ size 530702616
checkpoints/BC/dit-b-pixel/tetris/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "tetris",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6168223",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_tetris_slurm_6168223",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_tetris_slurm_6168223"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/tetris/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4eef634cc5a8b599b5c323e2c6c976c2cb7ee58c73860af2f2c3e12419190732
3
+ size 530702616
checkpoints/BC/dit-b-pixel/three_body/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "three_body",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6130783",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bc-raster_three_body_slurm_6130783",
85
+ "output_dir": "outputs/training/dit-b-pixel-bc-raster_three_body_slurm_6130783"
86
+ }
87
+ }
checkpoints/BC/dit-b-pixel/three_body/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1019f8b41e80a544fdeb0906a684d2206b9e53f83f794a09ae8d5a1579d53725
3
+ size 530702672
checkpoints/BC/wan-5b-lora/SHA256SUMS ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ 160d1b59711b38af416150290ce5aab30bed9fed7d8d55ed1ef0399e4e1dd384 mpm_worlds/config.json
2
+ f5941d9310db86c3afcf039ad7689810609a0af68466bd79e7b8baea7d1c387b rubiks_real/config.json
3
+ abeedb3eda438df3fa015ad5068a13cd9e5fc98604bd3ca300f4cedc4783a47e kubric_movi_c/config.json
4
+ 840b0da82b3942bfc9546c3228e9a7a09e517282af945cdb430a949652962cb0 kubric_movi_a/config.json
5
+ 658d55e2a49a5aa41c80491494719284e65f5c1d0d2740826af26cd3ede92e54 mpm_worlds/model_ema.safetensors
6
+ 4b154e1ad86dfe2ad8bee30046d534ba18a76b6140a76c3fb7511db40ac1ce97 rubiks_real/model_ema.safetensors
7
+ 2d325c3b2c4ecfe93e0c1997f4faa675750490bdf93072995feb1d32dd99e573 kubric_movi_c/model_ema.safetensors
8
+ fe6a359f937442f2a8ba244c8e3dcfaad7ff49f958e54802392d58bdfcaa4b91 kubric_movi_a/model_ema.safetensors
checkpoints/BC/wan-5b-lora/kubric_movi_a/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "kubric_movi_a",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "Wan-5B-LoRA",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "v",
19
+ "representation": "latent",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": "outputs/training/wan-5b-lora-bc-raster_kubric_movi_a_slurm_6181166/checkpoint-last.pth",
34
+ "run_name_suffix": "slurm_6202916",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.999,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 1,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.01
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 16
82
+ },
83
+ "run": {
84
+ "name": "wan-5b-lora-bc-raster_kubric_movi_a_slurm_6202916",
85
+ "output_dir": "outputs/training/wan-5b-lora-bc-raster_kubric_movi_a_slurm_6202916"
86
+ }
87
+ }
checkpoints/BC/wan-5b-lora/kubric_movi_a/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe6a359f937442f2a8ba244c8e3dcfaad7ff49f958e54802392d58bdfcaa4b91
3
+ size 322510808
checkpoints/BC/wan-5b-lora/kubric_movi_c/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "kubric_movi_c",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "Wan-5B-LoRA",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "v",
19
+ "representation": "latent",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": "outputs/training/wan-5b-lora-bc-raster_kubric_movi_c_slurm_6173030/checkpoint-last.pth",
34
+ "run_name_suffix": "slurm_6191624",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.999,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 1,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.01
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 16
82
+ },
83
+ "run": {
84
+ "name": "wan-5b-lora-bc-raster_kubric_movi_c_slurm_6191624",
85
+ "output_dir": "outputs/training/wan-5b-lora-bc-raster_kubric_movi_c_slurm_6191624"
86
+ }
87
+ }
checkpoints/BC/wan-5b-lora/kubric_movi_c/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2d325c3b2c4ecfe93e0c1997f4faa675750490bdf93072995feb1d32dd99e573
3
+ size 322510808
checkpoints/BC/wan-5b-lora/manifest.json ADDED
@@ -0,0 +1,144 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "archive_code": {
3
+ "commit": "6aefbf894f79ba790bbb880d805bffa7419852cc",
4
+ "repository": "https://github.com/jefequien/S2PD"
5
+ },
6
+ "archive_format_version": 1,
7
+ "diffusion_mode": "BC",
8
+ "family": "wan-5b-lora",
9
+ "intended_use": [
10
+ "evaluation",
11
+ "paper plots and visualizations",
12
+ "checkpoint transfer"
13
+ ],
14
+ "model": "Wan-5B-LoRA",
15
+ "repository": "jefequien/S2PD",
16
+ "representation": "latent",
17
+ "storage_dtype": "float32",
18
+ "total_weight_file_bytes": 1290043696,
19
+ "training_resume_supported": false,
20
+ "variant_count": 4,
21
+ "variants": [
22
+ {
23
+ "config": {
24
+ "file": "checkpoints/BC/wan-5b-lora/mpm_worlds/config.json",
25
+ "file_bytes": 2205,
26
+ "sha256": "160d1b59711b38af416150290ce5aab30bed9fed7d8d55ed1ef0399e4e1dd384"
27
+ },
28
+ "dataset": "mpm_worlds",
29
+ "directory": "checkpoints/BC/wan-5b-lora/mpm_worlds",
30
+ "display_name": "MPMWorlds",
31
+ "training": {
32
+ "epoch_index": 99,
33
+ "git_commit": "93aa1e5854ea0cd302c8b3e36126b5cb7035eeff",
34
+ "global_step": 100000,
35
+ "slurm_job_id": "6150298",
36
+ "source_checkpoint_bytes": 1290648287,
37
+ "source_run": "wan-5b-lora-bc-raster_mpm_worlds_slurm_6150298"
38
+ },
39
+ "weights": {
40
+ "dtypes": [
41
+ "float32"
42
+ ],
43
+ "field": "model_ema",
44
+ "file": "checkpoints/BC/wan-5b-lora/mpm_worlds/model_ema.safetensors",
45
+ "file_bytes": 322511040,
46
+ "format": "s2pd-ema-v1",
47
+ "sha256": "658d55e2a49a5aa41c80491494719284e65f5c1d0d2740826af26cd3ede92e54",
48
+ "tensor_bytes": 322437120,
49
+ "tensor_count": 600
50
+ }
51
+ },
52
+ {
53
+ "config": {
54
+ "file": "checkpoints/BC/wan-5b-lora/rubiks_real/config.json",
55
+ "file_bytes": 2209,
56
+ "sha256": "f5941d9310db86c3afcf039ad7689810609a0af68466bd79e7b8baea7d1c387b"
57
+ },
58
+ "dataset": "rubiks_real",
59
+ "directory": "checkpoints/BC/wan-5b-lora/rubiks_real",
60
+ "display_name": "Rubik's Cube Real",
61
+ "training": {
62
+ "epoch_index": 99,
63
+ "git_commit": "93aa1e5854ea0cd302c8b3e36126b5cb7035eeff",
64
+ "global_step": 100000,
65
+ "slurm_job_id": "6150300",
66
+ "source_checkpoint_bytes": 1290648287,
67
+ "source_run": "wan-5b-lora-bc-raster_rubiks_real_slurm_6150300"
68
+ },
69
+ "weights": {
70
+ "dtypes": [
71
+ "float32"
72
+ ],
73
+ "field": "model_ema",
74
+ "file": "checkpoints/BC/wan-5b-lora/rubiks_real/model_ema.safetensors",
75
+ "file_bytes": 322511040,
76
+ "format": "s2pd-ema-v1",
77
+ "sha256": "4b154e1ad86dfe2ad8bee30046d534ba18a76b6140a76c3fb7511db40ac1ce97",
78
+ "tensor_bytes": 322437120,
79
+ "tensor_count": 600
80
+ }
81
+ },
82
+ {
83
+ "config": {
84
+ "file": "checkpoints/BC/wan-5b-lora/kubric_movi_c/config.json",
85
+ "file_bytes": 2166,
86
+ "sha256": "abeedb3eda438df3fa015ad5068a13cd9e5fc98604bd3ca300f4cedc4783a47e"
87
+ },
88
+ "dataset": "kubric_movi_c",
89
+ "directory": "checkpoints/BC/wan-5b-lora/kubric_movi_c",
90
+ "display_name": "Kubric MOVi-C",
91
+ "training": {
92
+ "epoch_index": 99,
93
+ "git_commit": "720d858446e483b74d73c6f71e118d7f280fec37",
94
+ "global_step": 100000,
95
+ "slurm_job_id": "6191624",
96
+ "source_checkpoint_bytes": 1290648287,
97
+ "source_run": "wan-5b-lora-bc-raster_kubric_movi_c_slurm_6191624"
98
+ },
99
+ "weights": {
100
+ "dtypes": [
101
+ "float32"
102
+ ],
103
+ "field": "model_ema",
104
+ "file": "checkpoints/BC/wan-5b-lora/kubric_movi_c/model_ema.safetensors",
105
+ "file_bytes": 322510808,
106
+ "format": "s2pd-ema-v1",
107
+ "sha256": "2d325c3b2c4ecfe93e0c1997f4faa675750490bdf93072995feb1d32dd99e573",
108
+ "tensor_bytes": 322437120,
109
+ "tensor_count": 600
110
+ }
111
+ },
112
+ {
113
+ "config": {
114
+ "file": "checkpoints/BC/wan-5b-lora/kubric_movi_a/config.json",
115
+ "file_bytes": 2166,
116
+ "sha256": "840b0da82b3942bfc9546c3228e9a7a09e517282af945cdb430a949652962cb0"
117
+ },
118
+ "dataset": "kubric_movi_a",
119
+ "directory": "checkpoints/BC/wan-5b-lora/kubric_movi_a",
120
+ "display_name": "Kubric MOVi-A",
121
+ "training": {
122
+ "epoch_index": 99,
123
+ "git_commit": "6aefbf894f79ba790bbb880d805bffa7419852cc",
124
+ "global_step": 100000,
125
+ "slurm_job_id": "6202916",
126
+ "source_checkpoint_bytes": 1290648287,
127
+ "source_run": "wan-5b-lora-bc-raster_kubric_movi_a_slurm_6202916"
128
+ },
129
+ "weights": {
130
+ "dtypes": [
131
+ "float32"
132
+ ],
133
+ "field": "model_ema",
134
+ "file": "checkpoints/BC/wan-5b-lora/kubric_movi_a/model_ema.safetensors",
135
+ "file_bytes": 322510808,
136
+ "format": "s2pd-ema-v1",
137
+ "sha256": "fe6a359f937442f2a8ba244c8e3dcfaad7ff49f958e54802392d58bdfcaa4b91",
138
+ "tensor_bytes": 322437120,
139
+ "tensor_count": 600
140
+ }
141
+ }
142
+ ],
143
+ "weights": "model_ema"
144
+ }
checkpoints/BC/wan-5b-lora/mpm_worlds/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "mpm_worlds",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "Wan-5B-LoRA",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "v",
19
+ "representation": "latent",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": "/lus/lfs1aip2/projects/u6el/jhh57/Code/jit_sandbox/outputs/training/wan-5b-lora-bc-raster_mpm_worlds_slurm_6130896/checkpoint-last.pth",
34
+ "run_name_suffix": "slurm_6150298",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.999,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 2,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.01
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 16
82
+ },
83
+ "run": {
84
+ "name": "wan-5b-lora-bc-raster_mpm_worlds_slurm_6150298",
85
+ "output_dir": "outputs/training/wan-5b-lora-bc-raster_mpm_worlds_slurm_6150298"
86
+ }
87
+ }
checkpoints/BC/wan-5b-lora/mpm_worlds/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:658d55e2a49a5aa41c80491494719284e65f5c1d0d2740826af26cd3ede92e54
3
+ size 322511040
checkpoints/BC/wan-5b-lora/rubiks_real/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "rubiks_real",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "block_causal",
12
+ "model": "Wan-5B-LoRA",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "v",
19
+ "representation": "latent",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bc",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": "/lus/lfs1aip2/projects/u6el/jhh57/Code/jit_sandbox/outputs/training/wan-5b-lora-bc-raster_rubiks_real_slurm_6130899/checkpoint-last.pth",
34
+ "run_name_suffix": "slurm_6150300",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "chunk",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 0.6,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bc",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.999,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "chunk"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 32,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 2,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 32000,
70
+ "use_context_stream": true,
71
+ "weight_decay": 0.01
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 16
82
+ },
83
+ "run": {
84
+ "name": "wan-5b-lora-bc-raster_rubiks_real_slurm_6150300",
85
+ "output_dir": "outputs/training/wan-5b-lora-bc-raster_rubiks_real_slurm_6150300"
86
+ }
87
+ }
checkpoints/BC/wan-5b-lora/rubiks_real/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b154e1ad86dfe2ad8bee30046d534ba18a76b6140a76c3fb7511db40ac1ce97
3
+ size 322511040
checkpoints/BD/dit-b-pixel/2048/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "2048",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "bidirectional",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bd",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6168217",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "video",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 1.0,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bd",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "video"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 64,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 64000,
70
+ "use_context_stream": false,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bd-raster_2048_slurm_6168217",
85
+ "output_dir": "outputs/training/dit-b-pixel-bd-raster_2048_slurm_6168217"
86
+ }
87
+ }
checkpoints/BD/dit-b-pixel/2048/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6da9d39c93fe50cfb42ed031f1f7079b77587c7a40a1baef85e7d524669defe
3
+ size 530702616
checkpoints/BD/dit-b-pixel/SHA256SUMS ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ faa3bca72b6ac685ca4758b3ad06107b330ec5642c44eaa41b06df63abf4c4c6 conway/config.json
2
+ 2e2a332a0fa4dd1b1530c5b6250a46f66624f281246a5099ee9e1a5ada135f07 chess/config.json
3
+ 53ab845d5a4cdeb938a22e9b27721e89aeb5220055462f2834f904c9c71cd1f8 fifteen_puzzle/config.json
4
+ 14dd5f7c1794d0954f771d68c6ace4b5d89c5c2f8810e095db609585ff9457ac 2048/config.json
5
+ 2f14a942ece7f951339aaea65b46489a5d6b8e3a6e3f2d5573dc2c163583263f tetris/config.json
6
+ 2283c62abc15e55ec9d1bcec8003e26d7b02c077687edfccf133e3ddf1a1ceb8 snake/config.json
7
+ ebb6c21c7dd8505eac11a5696302bfa7bb33579e51091a01e4f828ead1bb4114 double_pendulum/config.json
8
+ 281c809c6d7691183136a4ed7e5757586a92e8206dc9a919504f4aeb4ead38f4 three_body/config.json
9
+ 9fbb0588470fdbd1c4990a9dbb61cc8437fca9475adcb0c10cf4f91760fb1670 pong/config.json
10
+ 0c2bb655967dc989e6315fd54142a6feeeed8fbff99fb76932ad6c8e4569bc98 colliding_balls/config.json
11
+ 9a86d183d504b60b78d1a3e41182f48cac904ff3545979fd90d1de4b881db1f1 colliding_balls_3d/config.json
12
+ dbf4e4d3ce3ba61f3b476e6d7818d7b1471ac20b3d5b4aed3e7b86b8e8cb8ce2 rubiks_3d/config.json
13
+ c0c57462c4f961734153b200109acc2e08d068625392a1d8ab75e5ba1a3c493d conway/model_ema.safetensors
14
+ b8084d5cddb8b80bf496c961c62f3e2fa39d966d7161e747f5034db3f422764e chess/model_ema.safetensors
15
+ f0669433d452704fde45220dd4c4f1798a457144a6846eb139b50b91d3c7297d fifteen_puzzle/model_ema.safetensors
16
+ d6da9d39c93fe50cfb42ed031f1f7079b77587c7a40a1baef85e7d524669defe 2048/model_ema.safetensors
17
+ 590a47a28f0b7650ce4fe73febd9f2a65523b649d750b1f148a01d1da19f29d2 tetris/model_ema.safetensors
18
+ 67d876085511bed15865f44ff6047dff0d9d6404d3b2fd12a47505b1ffedc21c snake/model_ema.safetensors
19
+ b933672ffc400412d9d266fa6fe477437d2a13cb8953b5e56c5a4edb076cab68 double_pendulum/model_ema.safetensors
20
+ d9a631b79257d4402f812ae513d18d5e56296cf4cfbf0b67db61d66506ac4ac0 three_body/model_ema.safetensors
21
+ c83b672f493353aac5f3585474db752a874695363a03cf04ebe617ac6632fbae pong/model_ema.safetensors
22
+ 06c16885949eb4d02d1afcd878c987b6c8ee2f137358671f63d1f4e9247e1771 colliding_balls/model_ema.safetensors
23
+ c2294018bdb90c44f4dbb2e570272ad71a1bde692bfb7c871d8e3a13ff615a57 colliding_balls_3d/model_ema.safetensors
24
+ f3a6cd3645a0519f3ce2119c3a5ab649110ce7f0a4c4b86f5e47db256432c483 rubiks_3d/model_ema.safetensors
checkpoints/BD/dit-b-pixel/chess/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "chess",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "bidirectional",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bd",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6168207",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "video",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 1.0,
44
+ "num_samples": 16,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bd",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "video"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 64,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 64000,
70
+ "use_context_stream": false,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bd-raster_chess_slurm_6168207",
85
+ "output_dir": "outputs/training/dit-b-pixel-bd-raster_chess_slurm_6168207"
86
+ }
87
+ }
checkpoints/BD/dit-b-pixel/chess/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b8084d5cddb8b80bf496c961c62f3e2fa39d966d7161e747f5034db3f422764e
3
+ size 530702616
checkpoints/BD/dit-b-pixel/colliding_balls/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "colliding_balls",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "bidirectional",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bd",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6056468",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "video",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 1.0,
44
+ "num_samples": 8,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bd",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "video"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 64,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 64000,
70
+ "use_context_stream": false,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bd-raster_colliding_balls_slurm_6056468",
85
+ "output_dir": "outputs/training/dit-b-pixel-bd-raster_colliding_balls_slurm_6056468"
86
+ }
87
+ }
checkpoints/BD/dit-b-pixel/colliding_balls/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06c16885949eb4d02d1afcd878c987b6c8ee2f137358671f63d1f4e9247e1771
3
+ size 530702720
checkpoints/BD/dit-b-pixel/colliding_balls_3d/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "colliding_balls_3d",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "bidirectional",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bd",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6056469",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "video",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 1.0,
44
+ "num_samples": 8,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bd",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "video"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 64,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 64000,
70
+ "use_context_stream": false,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bd-raster_colliding_balls_3d_slurm_6056469",
85
+ "output_dir": "outputs/training/dit-b-pixel-bd-raster_colliding_balls_3d_slurm_6056469"
86
+ }
87
+ }
checkpoints/BD/dit-b-pixel/colliding_balls_3d/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c2294018bdb90c44f4dbb2e570272ad71a1bde692bfb7c871d8e3a13ff615a57
3
+ size 530702728
checkpoints/BD/dit-b-pixel/conway/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "conway",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "bidirectional",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bd",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6056474",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "video",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 1.0,
44
+ "num_samples": 8,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bd",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "video"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 64,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 64000,
70
+ "use_context_stream": false,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bd-raster_conway_slurm_6056474",
85
+ "output_dir": "outputs/training/dit-b-pixel-bd-raster_conway_slurm_6056474"
86
+ }
87
+ }
checkpoints/BD/dit-b-pixel/conway/model_ema.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0c57462c4f961734153b200109acc2e08d068625392a1d8ab75e5ba1a3c493d
3
+ size 530702712
checkpoints/BD/dit-b-pixel/double_pendulum/config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "config": {
3
+ "data": {
4
+ "fps": 24.0,
5
+ "materializer_workers": 2,
6
+ "num_workers": 4,
7
+ "source": "double_pendulum",
8
+ "target_frames": 240
9
+ },
10
+ "model": {
11
+ "attention_mode": "bidirectional",
12
+ "model": "DiT-B",
13
+ "patch_size": [
14
+ 4,
15
+ 32,
16
+ 32
17
+ ],
18
+ "prediction_type": "x",
19
+ "representation": "pixel",
20
+ "token_order": "raster"
21
+ },
22
+ "recipe_name": "bd",
23
+ "runtime": {
24
+ "checkpoint_every_epochs": 1,
25
+ "compile": true,
26
+ "dtype": "bf16",
27
+ "enable_wandb": true,
28
+ "evaluate_every_epochs": 10,
29
+ "keep_n_ckpts": 2,
30
+ "metrics_every_steps": 20,
31
+ "output_root": "./outputs/training",
32
+ "progress_every_steps": 20,
33
+ "resume_from": null,
34
+ "run_name_suffix": "slurm_6056466",
35
+ "seed": 0,
36
+ "visualize_every_epochs": 10
37
+ },
38
+ "sampling": {
39
+ "block_size": "video",
40
+ "checkpoint": null,
41
+ "measure_timing": false,
42
+ "micro_batch_size": 1,
43
+ "noise_handoff": 1.0,
44
+ "num_samples": 8,
45
+ "output_dir": "",
46
+ "rollout_frames": 240,
47
+ "seed": 0,
48
+ "strategy": "bd",
49
+ "total_nfes": 10
50
+ },
51
+ "train": {
52
+ "beta1": 0.9,
53
+ "beta2": 0.95,
54
+ "block_size_weights": null,
55
+ "block_sizes": [
56
+ "video"
57
+ ],
58
+ "ema_decay": 0.9996,
59
+ "global_batch_size": 64,
60
+ "grad_clip": 1.0,
61
+ "lr": 0.0001,
62
+ "lr_schedule": "cosine",
63
+ "max_train_epochs": 100,
64
+ "micro_batch_size": 4,
65
+ "min_lr": 1e-05,
66
+ "monitor_samples": 16,
67
+ "noise_logit_mean": 0.8,
68
+ "noise_logit_std": 0.8,
69
+ "samples_per_epoch": 64000,
70
+ "use_context_stream": false,
71
+ "weight_decay": 0.0
72
+ }
73
+ },
74
+ "distributed": {
75
+ "backend": "nccl",
76
+ "device": "cuda:0",
77
+ "enabled": true,
78
+ "init_method": "env://",
79
+ "local_rank": 0,
80
+ "rank": 0,
81
+ "world_size": 4
82
+ },
83
+ "run": {
84
+ "name": "dit-b-pixel-bd-raster_double_pendulum_slurm_6056466",
85
+ "output_dir": "outputs/training/dit-b-pixel-bd-raster_double_pendulum_slurm_6056466"
86
+ }
87
+ }