ZeyuLing commited on
Commit
78fa5ce
·
verified ·
1 Parent(s): 83625e8

Add files using upload-large-folder tool

Browse files
README.md ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: motius
3
+ pipeline_tag: text-to-motion
4
+ license: mit
5
+ tags:
6
+ - human-motion
7
+ - two-person-text-to-motion
8
+ - intermask
9
+ - interhuman
10
+ ---
11
+
12
+ # Motius InterMask InterHuman
13
+
14
+ Inference-only SafeTensors artifact for
15
+ [InterMask](https://arxiv.org/abs/2410.10010), adapted from the
16
+ [official MIT-licensed repository](https://github.com/gohar-malik/InterMask).
17
+ The VQ-VAE and masked Transformer contain no optimizer or scheduler state.
18
+
19
+ ```python
20
+ from motius.pipelines.intermask import InterMaskPipeline
21
+
22
+ pipe = InterMaskPipeline.from_pretrained(
23
+ "ZeyuLing/motius-intermask-interhuman",
24
+ bundle_kwargs={"device": "cuda"},
25
+ )
26
+ motion = pipe("two people hug", motion_len=120, seed=42)
27
+ # (1, 120, 2, 262)
28
+ ```
29
+
30
+ See the [full Motius model card](https://github.com/ZeyuLing/Motius/blob/main/docs/model_zoo/intermask.md).
artifact_manifest.json ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "format_version": 1,
3
+ "source": {
4
+ "type": "processed official checkpoint"
5
+ },
6
+ "files": [
7
+ {
8
+ "path": "README.md",
9
+ "bytes": 818,
10
+ "sha256": "555ce5b3d94bb48148ccfe012cd0665dc33f4931559cf8add6c9641be1713ba9"
11
+ },
12
+ {
13
+ "path": "intermask_config.json",
14
+ "bytes": 224,
15
+ "sha256": "894c0302cbc72bd894a8afda366129a05c25fab53a2d49cf4279e017e9e718a7"
16
+ },
17
+ {
18
+ "path": "stats/global_mean.npy",
19
+ "bytes": 1176,
20
+ "sha256": "2f97b88889c773551cd597c32e88bfa070c2fb1721a992a80bfb7cf4c22e03b6"
21
+ },
22
+ {
23
+ "path": "stats/global_std.npy",
24
+ "bytes": 1176,
25
+ "sha256": "dd5988b979e9803786ab497f5945316c2d52b97c90b593d0427220fd49b03e4b"
26
+ },
27
+ {
28
+ "path": "transformer.safetensors",
29
+ "bytes": 709376528,
30
+ "sha256": "3566b03bdf200be3eecd2f20bf2ca2a3877d9b8d5fe6b1d2554127bdb013fcde"
31
+ },
32
+ {
33
+ "path": "transformer_opt.txt",
34
+ "bytes": 535,
35
+ "sha256": "eea88116566997576f28c31dc0bf5e1407645f9d9a56c45d638a4da22b739535"
36
+ },
37
+ {
38
+ "path": "vq_model.safetensors",
39
+ "bytes": 174228896,
40
+ "sha256": "a9ec2532db6e56a533b587a2a0b5f12d7b8e576700acfc7a38072871e3822f59"
41
+ },
42
+ {
43
+ "path": "vq_opt.txt",
44
+ "bytes": 726,
45
+ "sha256": "4cf108b19ea48e43f0e002618dd79b03a9444d17fc0849f54e4e01230f17406d"
46
+ }
47
+ ]
48
+ }
intermask_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "intermask",
3
+ "library_name": "motius",
4
+ "tasks": [
5
+ "two-person-text-to-motion"
6
+ ],
7
+ "dataset_name": "interhuman",
8
+ "representation": "interhuman_native_262",
9
+ "checkpoint_format": "safetensors"
10
+ }
stats/global_mean.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2f97b88889c773551cd597c32e88bfa070c2fb1721a992a80bfb7cf4c22e03b6
3
+ size 1176
stats/global_std.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dd5988b979e9803786ab497f5945316c2d52b97c90b593d0427220fd49b03e4b
3
+ size 1176
transformer.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3566b03bdf200be3eecd2f20bf2ca2a3877d9b8d5fe6b1d2554127bdb013fcde
3
+ size 709376528
transformer_opt.txt ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ------------ Options -------------
2
+ batch_size: 52
3
+ cache: True
4
+ checkpoints_dir: ./checkpoints
5
+ cond_drop_prob: 0.1
6
+ dataset_name: interhuman
7
+ do_eval: True
8
+ dropout: 0.2
9
+ eval_every_e: 10
10
+ ff_size: 1024
11
+ force_mask: False
12
+ gamma: 0.3333333333333333
13
+ gpu_id: 0
14
+ gumbel_sample: False
15
+ interaction_mask_prob: 0.2
16
+ is_continue: False
17
+ is_train: True
18
+ latent_dim: 384
19
+ lr: 0.0002
20
+ max_epoch: 500
21
+ motion_rep: global
22
+ n_heads: 6
23
+ n_layers: 6
24
+ name: trans_default
25
+ seed: 0
26
+ step_unroll: 1
27
+ test_batch_size: 96
28
+ vq_name: vq_default
29
+ -------------- End ----------------
vq_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a9ec2532db6e56a533b587a2a0b5f12d7b8e576700acfc7a38072871e3822f59
3
+ size 174228896
vq_opt.txt ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ------------ Options -------------
2
+ batch_size: 256
3
+ cache: True
4
+ checkpoints_dir: ./checkpoints
5
+ code_dim: 512
6
+ commit: 0.02
7
+ dataset_name: interhuman
8
+ depth: 2
9
+ dilation_growth_rate: 3
10
+ do_eval: True
11
+ down_t: 2
12
+ eval_every_e: 1
13
+ feat_bias: 5
14
+ gamma: 0.1
15
+ gpu_id: 0
16
+ is_continue: False
17
+ is_train: True
18
+ loss_bn: 5
19
+ loss_explicit: 1
20
+ loss_fc: 500
21
+ loss_geo: 0.01
22
+ loss_vel: 100
23
+ lr: 0.0002
24
+ max_epoch: 50
25
+ motion_rep: smpl
26
+ mu: 0.99
27
+ name: vq_default
28
+ nb_code: 1024
29
+ num_quantizers: 1
30
+ output_emb_width: 512
31
+ quantize_dropout_prob: 0.2
32
+ recons_loss: l1_smooth
33
+ save_every_e: 1
34
+ shared_codebook: False
35
+ stride_t: 2
36
+ test_batch_size: 96
37
+ vq_act: relu
38
+ vq_norm: None
39
+ weight_decay: 0.0
40
+ width: 512
41
+ window_size: 64
42
+ window_stride: 10
43
+ -------------- End ----------------