Hugging Face's logo

TencentARC
/
Pixal3D-D

Safetensors
Model card Files
xet
 Community
Yang2001 commited on
Commit
2317299
·
verified ·
1 Parent(s): bbef0c5

Upload folder using huggingface_hub

Browse files
Files changed (21) hide show
  1. README.md +1 -1
  2. dense/conditioner/config.json +9 -0
  3. dense/dit/config.json +24 -0
  4. dense/dit/model.safetensors +3 -0
  5. dense/scale_init/config.json +24 -0
  6. dense/scale_init/model.safetensors +3 -0
  7. dense/scheduler/config.json +7 -0
  8. dense/vae/config.json +25 -0
  9. dense/vae/model.safetensors +3 -0
  10. sparse1024/conditioner/config.json +11 -0
  11. sparse1024/dit/config.json +26 -0
  12. sparse1024/dit/model.safetensors +3 -0
  13. sparse1024/scheduler/config.json +7 -0
  14. sparse1024/vae/config.json +20 -0
  15. sparse1024/vae/model.safetensors +3 -0
  16. sparse512/conditioner/config.json +11 -0
  17. sparse512/dit/config.json +26 -0
  18. sparse512/dit/model.safetensors +3 -0
  19. sparse512/scheduler/config.json +7 -0
  20. sparse512/vae/config.json +19 -0
  21. sparse512/vae/model.safetensors +3 -0
README.md CHANGED
@@ -1,3 +1,3 @@
1
  ---
2
- license: apache-2.0
3
  ---
 
1
  ---
2
+ license: mit
3
  ---
dense/conditioner/config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "conditioner",
3
+ "model_class": "pixal3d.models.conditional_encoders.dinov2_project_grid.DinoEncoderProj",
4
+ "config": {
5
+ "model": "facebookresearch/dinov2",
6
+ "version": "dinov2_vitl14_reg",
7
+ "size": 518
8
+ }
9
+ }
dense/dit/config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "dense_dit",
3
+ "model_class": "pixal3d.models.transformers.dense_dit.DenseDiTDenoiser",
4
+ "resolution": 16,
5
+ "in_channels": 8,
6
+ "out_channels": 8,
7
+ "model_channels": 1024,
8
+ "cond_channels": 1024,
9
+ "num_blocks": 24,
10
+ "num_heads": 16,
11
+ "mlp_ratio": 4,
12
+ "patch_size": 1,
13
+ "pe_mode": "ape",
14
+ "qk_rms_norm": true,
15
+ "use_checkpoint": true,
16
+ "use_fp16": false,
17
+ "latent_shape": [
18
+ 8,
19
+ 16,
20
+ 16,
21
+ 16
22
+ ],
23
+ "image_attn_mode": "proj"
24
+ }
dense/dit/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9a4a6c34d6caa6272d63554f4d109e405c541c28efe49799590bf170eff47a2
3
+ size 2239009440
dense/scale_init/config.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "dense_dit",
3
+ "model_class": "pixal3d.models.transformers.dense_dit.DenseDiTDenoiser",
4
+ "resolution": 16,
5
+ "in_channels": 8,
6
+ "out_channels": 8,
7
+ "model_channels": 1024,
8
+ "cond_channels": 1024,
9
+ "num_blocks": 24,
10
+ "num_heads": 16,
11
+ "mlp_ratio": 4,
12
+ "patch_size": 1,
13
+ "pe_mode": "ape",
14
+ "qk_rms_norm": true,
15
+ "use_checkpoint": true,
16
+ "use_fp16": false,
17
+ "latent_shape": [
18
+ 8,
19
+ 16,
20
+ 16,
21
+ 16
22
+ ],
23
+ "image_attn_mode": "proj"
24
+ }
dense/scale_init/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:44eb629776b719fed975ba3e3bc11ce152b74359fd33d54834a322e7d90b445a
3
+ size 2239009440
dense/scheduler/config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "scheduler_class": "diffusers.schedulers.FlowMatchEulerDiscreteScheduler",
3
+ "scheduler_config": {
4
+ "num_train_timesteps": 1000,
5
+ "shift": 6.0
6
+ }
7
+ }
dense/vae/config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "vae",
3
+ "model_class": "pixal3d.models.autoencoders.dense_vae.DenseShapeVAE",
4
+ "use_checkpoint": true,
5
+ "embed_dim": 8,
6
+ "in_channels": 1,
7
+ "out_channels": 1,
8
+ "model_channels_encoder": [
9
+ 32,
10
+ 128,
11
+ 512
12
+ ],
13
+ "num_res_blocks_encoder": 2,
14
+ "num_res_blocks_middle_encoder": 2,
15
+ "model_channels_decoder": [
16
+ 512,
17
+ 128,
18
+ 64
19
+ ],
20
+ "num_res_blocks_decoder": 4,
21
+ "num_res_blocks_middle_decoder": 4,
22
+ "use_fp16": true,
23
+ "latents_scale": 1.0,
24
+ "latents_shift": 0.0
25
+ }
dense/vae/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b021bed4b53558e059f6bcd0b09be974a88715cf59975900ca5bd072c7e198ca
3
+ size 386095442
sparse1024/conditioner/config.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "conditioner",
3
+ "model_class": "pixal3d.models.conditional_encoders.dinov2_project_grid.DinoEncoderProj",
4
+ "config": {
5
+ "model": "facebookresearch/dinov2",
6
+ "version": "dinov2_vitl14_reg",
7
+ "size": 518,
8
+ "grid_resolution": 128,
9
+ "use_upsample": true
10
+ }
11
+ }
sparse1024/dit/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "sparse_dit",
3
+ "model_class": "pixal3d.models.transformers.sparse_dit.SparseDiTDenoiser",
4
+ "resolution": 128,
5
+ "in_channels": 16,
6
+ "out_channels": 16,
7
+ "model_channels": 1024,
8
+ "cond_channels": 1024,
9
+ "num_blocks": 24,
10
+ "num_heads": 32,
11
+ "num_kv_heads": 2,
12
+ "compression_block_size": 4,
13
+ "selection_block_size": 8,
14
+ "topk": 8,
15
+ "compression_version": "v2",
16
+ "pe_mode": "ape",
17
+ "factor": 0.5,
18
+ "sparse_conditions": true,
19
+ "qk_rms_norm": true,
20
+ "use_shift": true,
21
+ "use_checkpoint": true,
22
+ "use_fp16": false,
23
+ "use_visual_condition": true,
24
+ "visual_condition_dim": 1024,
25
+ "image_attn_mode": "proj"
26
+ }
sparse1024/dit/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a6fdd929d99b69292fbd958da46a5269073e758cdeceb637e8b992b779c25ad9
3
+ size 2080445392
sparse1024/scheduler/config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "scheduler_class": "diffusers.schedulers.FlowMatchEulerDiscreteScheduler",
3
+ "scheduler_config": {
4
+ "num_train_timesteps": 1000,
5
+ "shift": 8.0
6
+ }
7
+ }
sparse1024/vae/config.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "vae",
3
+ "model_class": "pixal3d.models.autoencoders.ss_vae.SparseSDFVAE",
4
+ "use_checkpoint": true,
5
+ "embed_dim": 16,
6
+ "num_head_channels_encoder": 64,
7
+ "model_channels_encoder": 512,
8
+ "num_heads_encoder": 8,
9
+ "num_blocks_encoder": 4,
10
+ "num_head_channels_decoder": 64,
11
+ "model_channels_decoder": 512,
12
+ "num_heads_decoder": 8,
13
+ "num_blocks_decoder": 4,
14
+ "resolution": 128,
15
+ "out_channels": 1,
16
+ "use_fp16": true,
17
+ "latents_scale": 1.0,
18
+ "latents_shift": 0.0,
19
+ "chunk_size": 4
20
+ }
sparse1024/vae/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a107525c430b4a72bc8ff70092a8b33b9e60925f0dddda445f4176731bbecee2
3
+ size 151508788
sparse512/conditioner/config.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "conditioner",
3
+ "model_class": "pixal3d.models.conditional_encoders.dinov2_project_grid.DinoEncoderProj",
4
+ "config": {
5
+ "model": "facebookresearch/dinov2",
6
+ "version": "dinov2_vitl14_reg",
7
+ "size": 518,
8
+ "grid_resolution": 64,
9
+ "use_upsample": true
10
+ }
11
+ }
sparse512/dit/config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "sparse_dit",
3
+ "model_class": "pixal3d.models.transformers.sparse_dit.SparseDiTDenoiser",
4
+ "resolution": 64,
5
+ "in_channels": 16,
6
+ "out_channels": 16,
7
+ "model_channels": 1024,
8
+ "cond_channels": 1024,
9
+ "num_blocks": 24,
10
+ "num_heads": 32,
11
+ "num_kv_heads": 2,
12
+ "compression_block_size": 4,
13
+ "selection_block_size": 8,
14
+ "topk": 32,
15
+ "compression_version": "v2",
16
+ "pe_mode": "ape",
17
+ "factor": 1,
18
+ "sparse_conditions": true,
19
+ "qk_rms_norm": true,
20
+ "use_shift": true,
21
+ "use_checkpoint": true,
22
+ "use_fp16": false,
23
+ "use_visual_condition": true,
24
+ "visual_condition_dim": 1024,
25
+ "image_attn_mode": "proj"
26
+ }
sparse512/dit/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0290082d7b3d4acd788f51e78279ec98448d2dfd45fcde68dc173874f9ef13b3
3
+ size 2080445392
sparse512/scheduler/config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "scheduler_class": "diffusers.schedulers.FlowMatchEulerDiscreteScheduler",
3
+ "scheduler_config": {
4
+ "num_train_timesteps": 1000,
5
+ "shift": 6.0
6
+ }
7
+ }
sparse512/vae/config.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "vae",
3
+ "model_class": "pixal3d.models.autoencoders.ss_vae.SparseSDFVAE",
4
+ "use_checkpoint": true,
5
+ "embed_dim": 16,
6
+ "num_head_channels_encoder": 64,
7
+ "model_channels_encoder": 512,
8
+ "num_heads_encoder": 8,
9
+ "num_blocks_encoder": 4,
10
+ "num_head_channels_decoder": 64,
11
+ "model_channels_decoder": 512,
12
+ "num_heads_decoder": 8,
13
+ "num_blocks_decoder": 4,
14
+ "resolution": 64,
15
+ "out_channels": 1,
16
+ "use_fp16": true,
17
+ "latents_scale": 1.0,
18
+ "latents_shift": 0.0
19
+ }
sparse512/vae/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cfae966e1485e4ffbec80c9bbfa1eb0c04f8f8ad8f6e8f74aaff5193a88d6d85
3
+ size 151508788