X-Zhang commited on
Commit
c3bd43a
·
verified ·
1 Parent(s): 06dff65

Publish clip-matched BEAT2 predictions and ground truth

Browse files
MANIFEST.json CHANGED
@@ -7,17 +7,17 @@
7
  "archive": "streamtalk_all_e0940_cfg3_npz.tar.gz",
8
  "archive_sha256": "295d933883dc40acb9df5f2d2ba02767ec62c9d83f423a78900932499fc8edb4",
9
  "archive_size_bytes": 552982864,
10
- "cfg_scale": 3.0,
11
  "checkpoint": "model.pt",
12
  "checkpoint_sha256": "3a0e3558d788fd3f11a5a87a8198b42725104b5b9b24f95a4e7c84c14395c9a5",
13
  "clip_batch_size": 16,
14
  "clip_count": 265,
15
  "format": "streamtalk-inference-scope-v1",
16
  "generation_seconds_by_rank": [
17
- 0.0,
18
- 0.0,
19
- 0.0,
20
- 0.0
21
  ],
22
  "generation_note": "The successful all-speaker outputs were schema/hash verified and reused during final packaging; the resumed packaging pass did not repeat generation timing.",
23
  "published_artifact_rescore": {
@@ -31,13 +31,23 @@
31
  "metric_batch_size": 16,
32
  "retrieval_query_frames": 8,
33
  "scope": "all",
34
- "seed": 0
 
 
 
 
 
 
 
 
 
 
35
  },
36
  "speaker2": {
37
  "archive": "streamtalk_speaker2_e0946_cfg3_npz.tar.gz",
38
  "archive_sha256": "859cb968c04804a797e3f8eb17aeba1f91bda29305f970564c9be481e41fa452",
39
  "archive_size_bytes": 29238804,
40
- "cfg_scale": 3.0,
41
  "checkpoint": "model.pt",
42
  "checkpoint_sha256": "07065d52e333cc48960e4e8fad5c0331f6b9e482f14aa1b78cb721904da5b075",
43
  "clip_batch_size": 1,
@@ -60,8 +70,19 @@
60
  "metric_batch_size": 16,
61
  "retrieval_query_frames": 8,
62
  "scope": "speaker2",
63
- "seed": 0
 
 
 
 
 
 
 
 
 
 
64
  }
65
  },
66
- "seed": 0
 
67
  }
 
7
  "archive": "streamtalk_all_e0940_cfg3_npz.tar.gz",
8
  "archive_sha256": "295d933883dc40acb9df5f2d2ba02767ec62c9d83f423a78900932499fc8edb4",
9
  "archive_size_bytes": 552982864,
10
+ "cfg_scale": 3,
11
  "checkpoint": "model.pt",
12
  "checkpoint_sha256": "3a0e3558d788fd3f11a5a87a8198b42725104b5b9b24f95a4e7c84c14395c9a5",
13
  "clip_batch_size": 16,
14
  "clip_count": 265,
15
  "format": "streamtalk-inference-scope-v1",
16
  "generation_seconds_by_rank": [
17
+ 0,
18
+ 0,
19
+ 0,
20
+ 0
21
  ],
22
  "generation_note": "The successful all-speaker outputs were schema/hash verified and reused during final packaging; the resumed packaging pass did not repeat generation timing.",
23
  "published_artifact_rescore": {
 
31
  "metric_batch_size": 16,
32
  "retrieval_query_frames": 8,
33
  "scope": "all",
34
+ "seed": 0,
35
+ "contains_ground_truth": false,
36
+ "paired_release": {
37
+ "archive": "StreamTalk_25speakers_predictions_and_gt.zip",
38
+ "archive_sha256": "e1e7304517e08dea0979d0577d7d86e03527b6f5e57932e384f0c517f3527bda",
39
+ "size_bytes": 1341193151,
40
+ "prediction_count": 265,
41
+ "ground_truth_count": 265,
42
+ "contains_ground_truth": true,
43
+ "ground_truth_source": "https://huggingface.co/datasets/haiyangliu0/BEAT2/tree/8689ecb43513ba31964fd60e0ca69be02d3b0872/beat_english_v2.0.0/smplxflame_30"
44
+ }
45
  },
46
  "speaker2": {
47
  "archive": "streamtalk_speaker2_e0946_cfg3_npz.tar.gz",
48
  "archive_sha256": "859cb968c04804a797e3f8eb17aeba1f91bda29305f970564c9be481e41fa452",
49
  "archive_size_bytes": 29238804,
50
+ "cfg_scale": 3,
51
  "checkpoint": "model.pt",
52
  "checkpoint_sha256": "07065d52e333cc48960e4e8fad5c0331f6b9e482f14aa1b78cb721904da5b075",
53
  "clip_batch_size": 1,
 
70
  "metric_batch_size": 16,
71
  "retrieval_query_frames": 8,
72
  "scope": "speaker2",
73
+ "seed": 0,
74
+ "contains_ground_truth": false,
75
+ "paired_release": {
76
+ "archive": "StreamTalk_speaker2_predictions_and_gt.zip",
77
+ "archive_sha256": "05aa172ef8ccc8c92c9aee6e921360e287972e17d8c07654ebcd8f8a64bf9f2a",
78
+ "size_bytes": 71001609,
79
+ "prediction_count": 15,
80
+ "ground_truth_count": 15,
81
+ "contains_ground_truth": true,
82
+ "ground_truth_source": "https://huggingface.co/datasets/haiyangliu0/BEAT2/tree/8689ecb43513ba31964fd60e0ca69be02d3b0872/beat_english_v2.0.0/smplxflame_30"
83
+ }
84
  }
85
  },
86
+ "seed": 0,
87
+ "contains_ground_truth": true
88
  }
README.md CHANGED
@@ -22,9 +22,16 @@ Ready-to-compare SMPL-X predictions from the official retrained StreamTalk CFG c
22
  - [Project page](https://xiangyuezhang.com/StreamTalk/) · [Code](https://github.com/Xiangyue-Zhang/StreamTalk)
23
  - [Model checkpoints](https://huggingface.co/X-Zhang/StreamTalk)
24
 
25
- No BEAT2 ground-truth motion, audio, training cache, or model checkpoint is included in these archives.
26
 
27
- ## Downloads
 
 
 
 
 
 
 
28
 
29
  | Scope | Clips | Checkpoint | CFG | Archive | SHA-256 |
30
  | :-- | --: | :-- | --: | :-- | :-- |
@@ -53,7 +60,7 @@ Both differences are below the selected engineering reporting tolerance of `1e-3
53
 
54
  ## NPZ schema
55
 
56
- Every archive contains one `res_<BEAT2-test-id>.npz` per official test ID plus a scope manifest. Load files with `allow_pickle=False`.
57
 
58
  | Key | Shape/type | Meaning |
59
  | :-- | :-- | :-- |
@@ -76,7 +83,7 @@ with np.load("res_2_scott_0_1_1.npz", allow_pickle=False) as motion:
76
 
77
  ## Re-score with StreamTalk
78
 
79
- The generated files do not include BEAT2 ground truth. After obtaining BEAT2 under its own terms, use the scorer in the [StreamTalk repository](https://github.com/Xiangyue-Zhang/StreamTalk):
80
 
81
  ```bash
82
  python tools/evaluate_generated_fgd.py \
@@ -97,4 +104,4 @@ python tools/evaluate_generated_fgd.py \
97
  - All-speaker checkpoint SHA-256: `3a0e3558d788fd3f11a5a87a8198b42725104b5b9b24f95a4e7c84c14395c9a5`.
98
  - Seed: 0. Speaker 2 generation uses clip batch 1; all-speaker generation uses clip batch 16; both use CFG 3 and an 8-frame retrieval query.
99
 
100
- The predictions are derived from experiments on BEAT2 and use the SMPL-X body model. Users remain responsible for complying with the licenses and terms of the [BEAT2 dataset](https://huggingface.co/datasets/H-Liu1997/BEAT2) and [SMPL-X](https://smpl-x.is.tue.mpg.de/). This dataset card does not redistribute BEAT2 ground truth, audio, or SMPL-X model files.
 
22
  - [Project page](https://xiangyuezhang.com/StreamTalk/) · [Code](https://github.com/Xiangyue-Zhang/StreamTalk)
23
  - [Model checkpoints](https://huggingface.co/X-Zhang/StreamTalk)
24
 
25
+ ## Compare with ground truth
26
 
27
+ | BEAT2 evaluation protocol | Predictions | Ground-truth clips | Download |
28
+ | --- | ---: | ---: | --- |
29
+ | Speaker 2 (Scott) | 15 | 15 | [Speaker 2 predictions and GT](https://huggingface.co/datasets/X-Zhang/StreamTalk-Inference-Data/resolve/main/StreamTalk_speaker2_predictions_and_gt.zip) |
30
+ | 25 English speakers | 265 | 265 | [25-speaker predictions and GT](https://huggingface.co/datasets/X-Zhang/StreamTalk-Inference-Data/resolve/main/StreamTalk_25speakers_predictions_and_gt.zip) |
31
+
32
+ Each paired ZIP contains the original prediction files under `predictions/res_<id>.npz` and the unmodified [BEAT2 reference motions](https://huggingface.co/datasets/haiyangliu0/BEAT2/tree/8689ecb43513ba31964fd60e0ca69be02d3b0872/beat_english_v2.0.0/smplxflame_30) under `ground_truth/<id>.npz`. The release scorer reads these directories directly. The file `pairs.csv` records both frame counts and SHA-256 hashes for every clip. These are the same predicted bytes used in the released re-score below. [ECCV proceedings](https://doi.org/10.1007/978-3-032-37595-7_32).
33
+
34
+ ## Original prediction-only downloads
35
 
36
  | Scope | Clips | Checkpoint | CFG | Archive | SHA-256 |
37
  | :-- | --: | :-- | --: | :-- | :-- |
 
60
 
61
  ## NPZ schema
62
 
63
+ Each original prediction-only archive contains one `res_<BEAT2-test-id>.npz` per official test ID plus a scope manifest. Load files with `allow_pickle=False`.
64
 
65
  | Key | Shape/type | Meaning |
66
  | :-- | :-- | :-- |
 
83
 
84
  ## Re-score with StreamTalk
85
 
86
+ Extract a paired ZIP, then pass its `predictions` and `ground_truth` directories to the scorer in the [StreamTalk repository](https://github.com/Xiangyue-Zhang/StreamTalk). For the original prediction-only TAR files, supply the corresponding reference directory from BEAT2.
87
 
88
  ```bash
89
  python tools/evaluate_generated_fgd.py \
 
104
  - All-speaker checkpoint SHA-256: `3a0e3558d788fd3f11a5a87a8198b42725104b5b9b24f95a4e7c84c14395c9a5`.
105
  - Seed: 0. Speaker 2 generation uses clip batch 1; all-speaker generation uses clip batch 16; both use CFG 3 and an 8-frame retrieval query.
106
 
107
+ The predictions are derived from experiments on BEAT2 and use the SMPL-X body model. Users remain responsible for complying with the licenses and terms of the [BEAT2 dataset](https://huggingface.co/datasets/H-Liu1997/BEAT2) and [SMPL-X](https://smpl-x.is.tue.mpg.de/). The paired ZIPs include the original BEAT2 motion references. The two original prediction-only TAR files each contain zero ground-truth clips.
SHA256SUMS CHANGED
@@ -1,8 +1,10 @@
1
- a6fc3be63a595ad18e9a9119875736dc16da0cd8ab6b24e768f1524d25689b80 README.md
2
- 2d8866bac3c97819bb35ed07025a4d5fc170b0c8a56a5132db1551da6d0c8a26 MANIFEST.json
3
  15d35f63d787752ab852e849df7369166567cc951429007674e5a658de3a6a9f speaker2_fgd.json
4
  9a859040981909ead510e5bd34e9d066b9d87aba7e8471583655a57bc10459b6 all_fgd.json
5
  d4031ae4e6180fd4eb366de3e0c5c4c849046c9cac79963d7293ab3a432321f3 speaker2/MANIFEST.json
6
  b42c459ec32a71a97a1fbbd86b7a09f80f00a9786e5783f9a8da19e3bf845878 all_speakers/MANIFEST.json
7
  859cb968c04804a797e3f8eb17aeba1f91bda29305f970564c9be481e41fa452 streamtalk_speaker2_e0946_cfg3_npz.tar.gz
8
  295d933883dc40acb9df5f2d2ba02767ec62c9d83f423a78900932499fc8edb4 streamtalk_all_e0940_cfg3_npz.tar.gz
 
 
 
1
+ 3ce40a84eb4cf45894b148caa7e425d4097f07983a57bf60e7dd54c031c9f83e README.md
2
+ 2ce9c5c4dcdd675a700cad3387598ea1747a36f89b82e3da71208351930ac700 MANIFEST.json
3
  15d35f63d787752ab852e849df7369166567cc951429007674e5a658de3a6a9f speaker2_fgd.json
4
  9a859040981909ead510e5bd34e9d066b9d87aba7e8471583655a57bc10459b6 all_fgd.json
5
  d4031ae4e6180fd4eb366de3e0c5c4c849046c9cac79963d7293ab3a432321f3 speaker2/MANIFEST.json
6
  b42c459ec32a71a97a1fbbd86b7a09f80f00a9786e5783f9a8da19e3bf845878 all_speakers/MANIFEST.json
7
  859cb968c04804a797e3f8eb17aeba1f91bda29305f970564c9be481e41fa452 streamtalk_speaker2_e0946_cfg3_npz.tar.gz
8
  295d933883dc40acb9df5f2d2ba02767ec62c9d83f423a78900932499fc8edb4 streamtalk_all_e0940_cfg3_npz.tar.gz
9
+ 05aa172ef8ccc8c92c9aee6e921360e287972e17d8c07654ebcd8f8a64bf9f2a StreamTalk_speaker2_predictions_and_gt.zip
10
+ e1e7304517e08dea0979d0577d7d86e03527b6f5e57932e384f0c517f3527bda StreamTalk_25speakers_predictions_and_gt.zip
StreamTalk_25speakers_predictions_and_gt.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e1e7304517e08dea0979d0577d7d86e03527b6f5e57932e384f0c517f3527bda
3
+ size 1341193151
StreamTalk_speaker2_predictions_and_gt.zip ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05aa172ef8ccc8c92c9aee6e921360e287972e17d8c07654ebcd8f8a64bf9f2a
3
+ size 71001609