anonstreammem commited on
Commit
bf02cb1
·
verified ·
1 Parent(s): e255a14

Reorganize: 3 top-level splits + standardized bucket names + single Croissant

Browse files

Move from 5 root-level parquets + 8 video dirs to a clean 3-split layout:
- text/{questions.parquet,questions.json}
- synthetic_video/{questions.parquet,questions.json,videos/L_<L>_frames/...}
- natural_video/{questions.parquet,questions.json,videos/nat_<L>_frames/...}
Single top-level Croissant 1.0 metadata.json covering all 3 RecordSets.
Renamed exact1fps_* buckets -> nat_<L>_frames. SoccerNet rows kept in parquet (provenance only).

This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +1 -0
  2. README.md +116 -86
  3. _split_sizes.json +0 -34
  4. metadata.json +101 -1507
  5. natural_video/questions.json +0 -0
  6. easyhuman_text.parquet → natural_video/questions.parquet +2 -2
  7. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_neg_01.mp4 +0 -0
  8. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_neg_04.mp4 +0 -0
  9. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_neg_05.mp4 +0 -0
  10. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_pos_04.mp4 +0 -0
  11. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_pos_05.mp4 +0 -0
  12. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/cut_nontomato_neg_04.mp4 +0 -0
  13. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/cut_nontomato_neg_05.mp4 +0 -0
  14. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_cupboard_neg_03.mp4 +0 -0
  15. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_drawer_neg_02.mp4 +0 -0
  16. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_drawer_neg_04.mp4 +0 -0
  17. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_neg_02.mp4 +0 -0
  18. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_neg_03.mp4 +0 -0
  19. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_neg_04.mp4 +0 -0
  20. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_pos_03.mp4 +0 -0
  21. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_pos_04.mp4 +0 -0
  22. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_pos_05.mp4 +0 -0
  23. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_neg_02.mp4 +0 -0
  24. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_neg_03.mp4 +0 -0
  25. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_neg_04.mp4 +0 -0
  26. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_pos_02.mp4 +0 -0
  27. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_pos_03.mp4 +0 -0
  28. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_pos_04.mp4 +0 -0
  29. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_pos_05.mp4 +0 -0
  30. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/rinse_X_neg_03.mp4 +0 -0
  31. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/rinse_X_neg_04.mp4 +0 -0
  32. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/rinse_X_pos_04.mp4 +0 -0
  33. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/rinse_X_pos_05.mp4 +0 -0
  34. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/turn_on_tap_neg_02.mp4 +0 -0
  35. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/turn_on_tap_neg_03.mp4 +0 -0
  36. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/turn_on_tap_neg_04.mp4 +0 -0
  37. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/turn_on_tap_pos_05.mp4 +0 -0
  38. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_neg_01.mp4 +0 -0
  39. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_neg_02.mp4 +0 -0
  40. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_neg_04.mp4 +0 -0
  41. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_neg_05.mp4 +0 -0
  42. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_pos_04.mp4 +0 -0
  43. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_pos_05.mp4 +0 -0
  44. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_neg_02.mp4 +0 -0
  45. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_neg_04.mp4 +0 -0
  46. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_neg_05.mp4 +0 -0
  47. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_pos_04.mp4 +0 -0
  48. {exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_pos_05.mp4 +0 -0
  49. {exact1fps_B3/videos → natural_video/videos/nat_128_frames}/clean_counter_neg_01.mp4 +0 -0
  50. {exact1fps_B3/videos → natural_video/videos/nat_128_frames}/clean_counter_neg_02.mp4 +0 -0
.gitattributes CHANGED
@@ -58,3 +58,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
 
 
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
61
+ text/questions.json filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -15,33 +15,21 @@ tags:
15
  - benchmark
16
  - video-llm
17
  configs:
18
- - config_name: easyhuman_text
19
- data_files: easyhuman_text.parquet
20
- - config_name: easyhuman_video
21
- data_files: easyhuman_video.parquet
22
- - config_name: substream_text
23
- data_files: substream_text.parquet
24
- - config_name: substream_video
25
- data_files: substream_video.parquet
26
  - config_name: natural_video
27
- data_files: natural_video.parquet
28
  dataset_info:
29
- - config_name: easyhuman_text
30
  splits:
31
  - name: train
32
- num_examples: 448
33
- - config_name: easyhuman_video
34
  splits:
35
  - name: train
36
- num_examples: 224
37
- - config_name: substream_text
38
- splits:
39
- - name: train
40
- num_examples: 7680
41
- - config_name: substream_video
42
- splits:
43
- - name: train
44
- num_examples: 6080
45
  - config_name: natural_video
46
  splits:
47
  - name: train
@@ -50,97 +38,139 @@ dataset_info:
50
 
51
  # Substream Recollection
52
 
53
- A controlled benchmark for substream membership recall in long-context VLMs and
54
  LLMs. Each row is a `(stream, probe, label)` tuple: the model sees a long input
55
  stream and a short probe, and must answer "yes" or "no" — did the probe occur
56
  inside the stream?
57
 
58
- The bundle has five splits:
59
-
60
- - **easyhuman_text** — text-mode EasyHuman: comma-separated symbol streams at
61
- L=256 and L=1024, with binary "did this short pattern appear?" probes.
62
- - **easyhuman_video** — video-mode EasyHuman at L=256: rendered 3-belt
63
- conveyor videos with the same probe families as `easyhuman_text` (X/Y
64
- block presence, variation absence, mistakes, invalid cycles).
65
- - **substream_text** synthetic text-mode Substream Recollection. All entropy
66
- bands (low / medium / uniform) at every L in {8, 16, 32, 64, 128, 256, 512,
67
- 1024, 2048, 4096}. Probes are sequential (subsequence-of) or spatial
68
- (lane-conditioned). The text stream is the comma-joined `S_tokens` from the
69
- source video.
70
- - **substream_video** — same `(video, question)` pairs as `substream_text`,
71
- but the input is the rendered video. Restricted to L ≤ 1024 because longer
72
- videos are impractical for current VLMs.
73
- - **natural_video** — natural-video benchmark drawn from EPIC-Kitchens-100 and
74
- SoccerNet, at L ∈ {8, 16, 64, 128, 512, 1024}. Each row asks whether a named
75
- event occurred in the clip. SoccerNet rows do **not** ship the underlying
76
- video files; provenance metadata is provided in the `source_provenance`
77
- column so the original videos can be obtained from
78
- https://www.soccer-net.org/data after signing the SoccerNet NDA.
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
79
 
80
  ## Schema
81
 
82
- Every parquet has the same columns:
 
 
 
83
 
84
  | column | type | notes |
85
  | --- | --- | --- |
86
- | `split` | str | one of the five split names |
87
- | `subset` | str | bucket id (e.g. `UNIFORM_EVAL_L0128_ELOW`, `len_1024`, `L=128`) |
 
88
  | `length_L` | int64 | normalized stream length |
89
  | `entropy_band` | str | `low` / `medium` / `uniform` / `natural` / `easyhuman` |
90
  | `modality` | str | `text` or `video` |
91
  | `question_variant` | str | `sequential` / `spatial` / `binary_natural` / `easyhuman_binary` |
92
- | `video_index` | int64 | per-bucket video id; null for text-only natural rows |
93
  | `question_index` | int64 | per-video question position |
94
- | `question_text` | str | prompt the model sees (no scaffolding) |
95
- | `candidate` | str (JSON) | probe metadata; null where N/A |
96
  | `answer` | str | ground-truth label (`yes` / `no`) |
97
- | `clip_path` | str | relative path to the probe clip (where applicable) |
98
- | `video_path` | str | relative path to the parent video (where applicable) |
99
  | `dataset_source` | str | `synthetic` / `easyhuman` / `epic-kitchens-100` / `soccernet` |
100
- | `input_sequence` | str | comma-joined input symbols (text mode payload) |
101
- | `source_class` | str | natural-only event-class label |
102
  | `source_dataset` | str | natural-only origin tag |
103
- | `video_id` | str | canonical video id (joins to eval state files) |
104
  | `question_id` | str | canonical question id |
105
- | `source_provenance` | str (JSON) | natural_video only; for SoccerNet rows, JSON describing the source match, half, time range, and license. Null elsewhere. |
106
-
107
- ## Files
108
-
109
- - `easyhuman_text.parquet` (448 rows)
110
- - `easyhuman_video.parquet` (224 rows)
111
- - `substream_text.parquet` (7,680 rows)
112
- - `substream_video.parquet` (6,080 rows)
113
- - `natural_video.parquet` (1,028 rows)
114
- - `metadata.json` — Croissant 1.0 dataset description
115
- - `_split_sizes.json` — machine-readable split-size sidecar
116
- - `videos/` and `clips/` — rendered synthetic substream media (substream_video)
117
- - `len_256/{videos,clips}/` — rendered EasyHuman media (easyhuman_video)
118
- - `exact1fps_short/`, `exact1fps_B{1..5}/` — EPIC-Kitchens-100 derived clips
119
- (natural_video). Each row's `video_path` column is a relative path under
120
- this layout. SoccerNet rows are excluded; see `source_provenance` column.
121
- - `LICENSES/` — full license text for each underlying source
122
-
123
- The 78 SoccerNet-derived rows are present in `natural_video.parquet` but the
124
- underlying video files are **not** redistributed here (NDA-gated source);
125
- see `LICENSES/SoccerNet-NOTE.txt` and the `source_provenance` column.
126
 
127
- ## Licenses
128
 
129
- This bundle aggregates several sources, each under its own license:
 
 
 
 
 
 
 
130
 
131
  | split / source | license | file |
132
  | --- | --- | --- |
133
- | Synthetic streams (substream_text, substream_video) | CC BY 4.0, Anonymous Authors 2026 | `LICENSES/synthetic-and-easyhuman.txt` |
134
- | EasyHuman (easyhuman_text, easyhuman_video) | CC BY 4.0, Anonymous Authors 2026 | `LICENSES/synthetic-and-easyhuman.txt` |
135
- | EPIC-Kitchens-100 derived clips (natural_video) | CC BY-NC 4.0 (research use only) | `LICENSES/EPIC-Kitchens-100-CC-BY-NC-4.0.txt` |
136
- | SoccerNet derived clips (natural_video, provenance only) | NDA-gated source; not redistributed | `LICENSES/SoccerNet-NOTE.txt` |
137
 
138
- The top-level `license: cc-by-4.0` tag in this card refers to the
139
  Anonymous-Authors-owned splits (synthetic + EasyHuman) and to this card,
140
- metadata, and code only. EPIC-Kitchens-100 derivatives remain CC BY-NC 4.0;
141
- SoccerNet provenance metadata is shared under fair-use research-citation
142
- conventions and the underlying clips are not included.
143
 
144
  ## Citation
145
 
146
- Cite as: Anonymous, "Substream Recollection," 2026 (anonymized for review).
 
15
  - benchmark
16
  - video-llm
17
  configs:
18
+ - config_name: text
19
+ data_files: text/questions.parquet
20
+ - config_name: synthetic_video
21
+ data_files: synthetic_video/questions.parquet
 
 
 
 
22
  - config_name: natural_video
23
+ data_files: natural_video/questions.parquet
24
  dataset_info:
25
+ - config_name: text
26
  splits:
27
  - name: train
28
+ num_examples: 8128
29
+ - config_name: synthetic_video
30
  splits:
31
  - name: train
32
+ num_examples: 6304
 
 
 
 
 
 
 
 
33
  - config_name: natural_video
34
  splits:
35
  - name: train
 
38
 
39
  # Substream Recollection
40
 
41
+ A controlled benchmark for substream-membership recall in long-context VLMs and
42
  LLMs. Each row is a `(stream, probe, label)` tuple: the model sees a long input
43
  stream and a short probe, and must answer "yes" or "no" — did the probe occur
44
  inside the stream?
45
 
46
+ This release reorganizes the original five-split bundle into three top-level
47
+ splits keyed by modality + source:
48
+
49
+ | split | rows | content |
50
+ | --- | --- | --- |
51
+ | `text` | 8,128 | text-modality questions for the synthetic substream benchmark *and* EasyHuman; the `subset` column distinguishes them. |
52
+ | `synthetic_video` | 6,304 | rendered synthetic substream videos *and* EasyHuman 3-belt videos; `subset` distinguishes them. |
53
+ | `natural_video` | 1,028 | EPIC-Kitchens-100 derived clips and SoccerNet provenance metadata. |
54
+
55
+ ## Directory layout
56
+
57
+ ```
58
+ anonstreammem/substream-recollection/
59
+ ├── README.md
60
+ ├── metadata.json # Croissant 1.0 covering all 3 RecordSets
61
+ ├── LICENSES/
62
+ │ ├── EPIC-Kitchens-100-CC-BY-NC-4.0.txt
63
+ │ ├── synthetic-and-easyhuman.txt
64
+ │ └── SoccerNet-NOTE.txt
65
+ ├── text/
66
+ │ ├── questions.parquet # subset in {substream, easyhuman}
67
+ │ └── questions.json # NDJSON copy of the parquet
68
+ ├── synthetic_video/
69
+ │ ├── questions.parquet # subset in {substream, easyhuman}
70
+ │ ├── questions.json
71
+ │ └── videos/
72
+ │ ├── L_8_frames/
73
+ │ ├── L_16_frames/
74
+ │ ├── L_32_frames/
75
+ │ ├── L_64_frames/
76
+ │ ├── L_128_frames/
77
+ │ ├── L_256_frames/
78
+ │ ├── L_512_frames/
79
+ │ ├── L_1024_frames/
80
+ │ └── easyhuman_L_256_frames/
81
+ └── natural_video/
82
+ ├── questions.parquet
83
+ ├── questions.json
84
+ └── videos/
85
+ ├── nat_8_frames/ # was exact1fps_short
86
+ ├── nat_16_frames/ # was exact1fps_B1
87
+ ├── nat_64_frames/ # was exact1fps_B2
88
+ ├── nat_128_frames/ # was exact1fps_B3
89
+ ├── nat_512_frames/ # was exact1fps_B4
90
+ └── nat_1024_frames/ # was exact1fps_B5
91
+ ```
92
+
93
+ Inside each `synthetic_video/videos/L_<L>_frames/<bucket>/` you'll find the
94
+ parent videos (e.g. `video_1_v0.mp4`) plus a `clips/` subfolder with the probe
95
+ clips. EasyHuman uses a flatter layout with no inner bucket directory.
96
+
97
+ ## Loading
98
+
99
+ ```python
100
+ from datasets import load_dataset
101
+
102
+ text = load_dataset("anonstreammem/substream-recollection", "text")["train"]
103
+ synthv = load_dataset("anonstreammem/substream-recollection", "synthetic_video")["train"]
104
+ natv = load_dataset("anonstreammem/substream-recollection", "natural_video")["train"]
105
+ ```
106
+
107
+ Video files are referenced by relative `video_path` / `clip_path` in each
108
+ parquet. To resolve them locally, snapshot the repo:
109
+
110
+ ```python
111
+ from huggingface_hub import snapshot_download
112
+ root = snapshot_download(
113
+ "anonstreammem/substream-recollection",
114
+ repo_type="dataset",
115
+ allow_patterns=["synthetic_video/**", "natural_video/**", "text/**"],
116
+ )
117
+ ```
118
+
119
+ Then `Path(root) / row["video_path"]` resolves to the actual mp4.
120
 
121
  ## Schema
122
 
123
+ All three parquets share most columns. The text and synthetic_video parquets
124
+ also carry `bucket` (original bucket id, e.g. `UNIFORM_EVAL_L0128_ELOW`,
125
+ `len_256`) and `source_split` (the pre-merge split label). `natural_video` adds
126
+ a `source_provenance` JSON column for SoccerNet rows.
127
 
128
  | column | type | notes |
129
  | --- | --- | --- |
130
+ | `subset` | str | `substream` / `easyhuman` (text + synthetic_video) or bucket id like `L=128` (natural_video) |
131
+ | `bucket` | str | (text + synthetic_video only) original bucket id |
132
+ | `source_split` | str | pre-merge split label |
133
  | `length_L` | int64 | normalized stream length |
134
  | `entropy_band` | str | `low` / `medium` / `uniform` / `natural` / `easyhuman` |
135
  | `modality` | str | `text` or `video` |
136
  | `question_variant` | str | `sequential` / `spatial` / `binary_natural` / `easyhuman_binary` |
137
+ | `video_index` | int64 | per-bucket video id |
138
  | `question_index` | int64 | per-video question position |
139
+ | `question_text` | str | prompt the model sees |
140
+ | `candidate` | str (JSON) | probe metadata; null when N/A |
141
  | `answer` | str | ground-truth label (`yes` / `no`) |
142
+ | `clip_path` | str | relative path to the probe clip |
143
+ | `video_path` | str | relative path to the parent video |
144
  | `dataset_source` | str | `synthetic` / `easyhuman` / `epic-kitchens-100` / `soccernet` |
145
+ | `input_sequence` | str | comma-joined input symbols (text-mode payload) |
146
+ | `source_class` | str | natural-only event class label |
147
  | `source_dataset` | str | natural-only origin tag |
148
+ | `video_id` | str | canonical video id |
149
  | `question_id` | str | canonical question id |
150
+ | `source_provenance` | str (JSON) | natural_video only; SoccerNet provenance |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
151
 
152
+ ## SoccerNet rows
153
 
154
+ The 78 SoccerNet-derived rows (event class `red_card`) appear in
155
+ `natural_video/questions.parquet` but the underlying mp4s are **not**
156
+ redistributed here — the SoccerNet source is NDA-gated. Use the
157
+ `source_provenance` JSON column to pull the originals from
158
+ [https://www.soccer-net.org/data](https://www.soccer-net.org/data) after
159
+ signing the SoccerNet NDA. See `LICENSES/SoccerNet-NOTE.txt`.
160
+
161
+ ## Licenses
162
 
163
  | split / source | license | file |
164
  | --- | --- | --- |
165
+ | Synthetic streams (`text/subset=substream`, `synthetic_video/subset=substream`) | CC BY 4.0, Anonymous Authors 2026 | `LICENSES/synthetic-and-easyhuman.txt` |
166
+ | EasyHuman (`text/subset=easyhuman`, `synthetic_video/subset=easyhuman`) | CC BY 4.0, Anonymous Authors 2026 | `LICENSES/synthetic-and-easyhuman.txt` |
167
+ | EPIC-Kitchens-100 derived clips (`natural_video`) | CC BY-NC 4.0 (research use only) | `LICENSES/EPIC-Kitchens-100-CC-BY-NC-4.0.txt` |
168
+ | SoccerNet derived rows (`natural_video`, provenance only) | NDA-gated source; not redistributed | `LICENSES/SoccerNet-NOTE.txt` |
169
 
170
+ The top-level `license: cc-by-4.0` tag on this card refers to the
171
  Anonymous-Authors-owned splits (synthetic + EasyHuman) and to this card,
172
+ metadata, and code only. EPIC-Kitchens-100 derivatives remain CC BY-NC 4.0.
 
 
173
 
174
  ## Citation
175
 
176
+ Anonymous, "Substream Recollection," 2026 (anonymized for review).
_split_sizes.json DELETED
@@ -1,34 +0,0 @@
1
- {
2
- "splits": {
3
- "easyhuman_text": {
4
- "num_rows": 448,
5
- "files": [
6
- "easyhuman_text.parquet"
7
- ]
8
- },
9
- "easyhuman_video": {
10
- "num_rows": 224,
11
- "files": [
12
- "easyhuman_video.parquet"
13
- ]
14
- },
15
- "substream_text": {
16
- "num_rows": 7680,
17
- "files": [
18
- "substream_text.parquet"
19
- ]
20
- },
21
- "substream_video": {
22
- "num_rows": 6080,
23
- "files": [
24
- "substream_video.parquet"
25
- ]
26
- },
27
- "natural_video": {
28
- "num_rows": 1028,
29
- "files": [
30
- "natural_video.parquet"
31
- ]
32
- }
33
- }
34
- }
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
metadata.json CHANGED
@@ -47,8 +47,8 @@
47
  "transform": "cr:transform"
48
  },
49
  "@type": "sc:Dataset",
50
- "name": "substream-recollection",
51
- "description": "Substream Recollection is a controlled benchmark for testing whether long-context VLMs / LLMs can recall whether a short probe substream occurred inside a longer parent stream. It pairs synthetic streams (low / medium / uniform entropy, lengths L=8..4096) with EasyHuman rendered 3-belt videos and a natural-video sub-benchmark drawn from EPIC-Kitchens-100 and SoccerNet.",
52
  "conformsTo": "http://mlcommons.org/croissant/1.0",
53
  "citeAs": "Anonymous, \"Substream Recollection,\" 2026 (anonymized for review).",
54
  "keywords": [
@@ -59,1539 +59,133 @@
59
  "vision-language-models"
60
  ],
61
  "license": "https://creativecommons.org/licenses/by/4.0/",
62
- "url": "https://huggingface.co/datasets/substream-recollection",
63
- "version": "1.0.0",
 
 
 
 
 
 
 
 
 
64
  "distribution": [
65
  {
66
  "@type": "cr:FileObject",
67
- "@id": "easyhuman_text",
68
- "name": "easyhuman_text",
69
- "description": "Parquet shard for split easyhuman_text.",
70
- "contentSize": "35488",
71
- "contentUrl": "easyhuman_text.parquet",
72
  "encodingFormat": "application/x-parquet",
73
- "sha256": "c999b026d494a507ded2d452c8924979961fe2c487ab742e4ccae45461495cfc",
74
- "license": "https://creativecommons.org/licenses/by/4.0/"
75
  },
76
  {
77
  "@type": "cr:FileObject",
78
- "@id": "easyhuman_video",
79
- "name": "easyhuman_video",
80
- "description": "Parquet shard for split easyhuman_video.",
81
- "contentSize": "20536",
82
- "contentUrl": "easyhuman_video.parquet",
83
  "encodingFormat": "application/x-parquet",
84
- "sha256": "3a734556d53f8144cb3081c221b77a72cb25c7f2411009d3c0f934a337757085",
85
- "license": "https://creativecommons.org/licenses/by/4.0/"
86
  },
87
  {
88
  "@type": "cr:FileObject",
89
- "@id": "substream_text",
90
- "name": "substream_text",
91
- "description": "Parquet shard for split substream_text.",
92
- "contentSize": "462931",
93
- "contentUrl": "substream_text.parquet",
94
  "encodingFormat": "application/x-parquet",
95
- "sha256": "25bec3762b4788ea884db85dcbebdbdf1c6c6b309dea649aa05cd0306f1bfbed",
96
- "license": "https://creativecommons.org/licenses/by/4.0/"
97
- },
98
- {
99
- "@type": "cr:FileObject",
100
- "@id": "substream_video",
101
- "name": "substream_video",
102
- "description": "Parquet shard for split substream_video.",
103
- "contentSize": "270127",
104
- "contentUrl": "substream_video.parquet",
105
- "encodingFormat": "application/x-parquet",
106
- "sha256": "3175ad85bc3b465750161844529e55111c985fd00750b8fbb260950ed1cc5d55",
107
- "license": "https://creativecommons.org/licenses/by/4.0/"
108
- },
109
- {
110
- "@type": "cr:FileObject",
111
- "@id": "natural_video",
112
- "name": "natural_video",
113
- "description": "Parquet shard for split natural_video.",
114
- "contentSize": "26698",
115
- "contentUrl": "natural_video.parquet",
116
- "encodingFormat": "application/x-parquet",
117
- "sha256": "1e1084463aff037aee62080daf4b6a995d447ffe673e81eb6b13cdee7b69b149",
118
- "license": "https://creativecommons.org/licenses/by-nc/4.0/"
119
  }
120
  ],
121
  "recordSet": [
122
  {
123
  "@type": "cr:RecordSet",
124
- "@id": "records/easyhuman_text",
125
- "name": "easyhuman_text",
126
- "description": "Substream Recollection split: easyhuman_text.",
127
- "field": [
128
- {
129
- "@type": "cr:Field",
130
- "@id": "records/easyhuman_text/split",
131
- "name": "split",
132
- "description": "Top-level split label.",
133
- "dataType": "sc:Text",
134
- "source": {
135
- "fileObject": {
136
- "@id": "easyhuman_text"
137
- },
138
- "extract": {
139
- "column": "split"
140
- }
141
- }
142
- },
143
- {
144
- "@type": "cr:Field",
145
- "@id": "records/easyhuman_text/subset",
146
- "name": "subset",
147
- "description": "Bucket-level subset id (e.g. UNIFORM_EVAL_L0128_ELOW, len_1024, L=128).",
148
- "dataType": "sc:Text",
149
- "source": {
150
- "fileObject": {
151
- "@id": "easyhuman_text"
152
- },
153
- "extract": {
154
- "column": "subset"
155
- }
156
- }
157
- },
158
- {
159
- "@type": "cr:Field",
160
- "@id": "records/easyhuman_text/length_L",
161
- "name": "length_L",
162
- "description": "Substream length L (frames or characters), normalized.",
163
- "dataType": "sc:Integer",
164
- "source": {
165
- "fileObject": {
166
- "@id": "easyhuman_text"
167
- },
168
- "extract": {
169
- "column": "length_L"
170
- }
171
- }
172
- },
173
- {
174
- "@type": "cr:Field",
175
- "@id": "records/easyhuman_text/entropy_band",
176
- "name": "entropy_band",
177
- "description": "low / medium / uniform / natural / easyhuman.",
178
- "dataType": "sc:Text",
179
- "source": {
180
- "fileObject": {
181
- "@id": "easyhuman_text"
182
- },
183
- "extract": {
184
- "column": "entropy_band"
185
- }
186
- }
187
- },
188
- {
189
- "@type": "cr:Field",
190
- "@id": "records/easyhuman_text/modality",
191
- "name": "modality",
192
- "description": "text or video.",
193
- "dataType": "sc:Text",
194
- "source": {
195
- "fileObject": {
196
- "@id": "easyhuman_text"
197
- },
198
- "extract": {
199
- "column": "modality"
200
- }
201
- }
202
- },
203
- {
204
- "@type": "cr:Field",
205
- "@id": "records/easyhuman_text/question_variant",
206
- "name": "question_variant",
207
- "description": "sequential, spatial, binary_natural, easyhuman_binary.",
208
- "dataType": "sc:Text",
209
- "source": {
210
- "fileObject": {
211
- "@id": "easyhuman_text"
212
- },
213
- "extract": {
214
- "column": "question_variant"
215
- }
216
- }
217
- },
218
- {
219
- "@type": "cr:Field",
220
- "@id": "records/easyhuman_text/video_index",
221
- "name": "video_index",
222
- "description": "Per-bucket video index (1-based).",
223
- "dataType": "sc:Integer",
224
- "source": {
225
- "fileObject": {
226
- "@id": "easyhuman_text"
227
- },
228
- "extract": {
229
- "column": "video_index"
230
- }
231
- }
232
- },
233
- {
234
- "@type": "cr:Field",
235
- "@id": "records/easyhuman_text/question_index",
236
- "name": "question_index",
237
- "description": "Per-video question index (0-based).",
238
- "dataType": "sc:Integer",
239
- "source": {
240
- "fileObject": {
241
- "@id": "easyhuman_text"
242
- },
243
- "extract": {
244
- "column": "question_index"
245
- }
246
- }
247
- },
248
- {
249
- "@type": "cr:Field",
250
- "@id": "records/easyhuman_text/question_text",
251
- "name": "question_text",
252
- "description": "Prompt the model sees (no schema scaffolding).",
253
- "dataType": "sc:Text",
254
- "source": {
255
- "fileObject": {
256
- "@id": "easyhuman_text"
257
- },
258
- "extract": {
259
- "column": "question_text"
260
- }
261
- }
262
- },
263
- {
264
- "@type": "cr:Field",
265
- "@id": "records/easyhuman_text/candidate",
266
- "name": "candidate",
267
- "description": "JSON-encoded probe metadata (sequence, clip_start, ...). Null when N/A.",
268
- "dataType": "sc:Text",
269
- "source": {
270
- "fileObject": {
271
- "@id": "easyhuman_text"
272
- },
273
- "extract": {
274
- "column": "candidate"
275
- }
276
- }
277
- },
278
- {
279
- "@type": "cr:Field",
280
- "@id": "records/easyhuman_text/answer",
281
- "name": "answer",
282
- "description": "Ground-truth label (yes/no for binary questions).",
283
- "dataType": "sc:Text",
284
- "source": {
285
- "fileObject": {
286
- "@id": "easyhuman_text"
287
- },
288
- "extract": {
289
- "column": "answer"
290
- }
291
- }
292
- },
293
- {
294
- "@type": "cr:Field",
295
- "@id": "records/easyhuman_text/clip_path",
296
- "name": "clip_path",
297
- "description": "Relative path to the probe clip file (where applicable).",
298
- "dataType": "sc:Text",
299
- "source": {
300
- "fileObject": {
301
- "@id": "easyhuman_text"
302
- },
303
- "extract": {
304
- "column": "clip_path"
305
- }
306
- }
307
- },
308
- {
309
- "@type": "cr:Field",
310
- "@id": "records/easyhuman_text/video_path",
311
- "name": "video_path",
312
- "description": "Relative path to the main video (where applicable).",
313
- "dataType": "sc:Text",
314
- "source": {
315
- "fileObject": {
316
- "@id": "easyhuman_text"
317
- },
318
- "extract": {
319
- "column": "video_path"
320
- }
321
- }
322
- },
323
- {
324
- "@type": "cr:Field",
325
- "@id": "records/easyhuman_text/dataset_source",
326
- "name": "dataset_source",
327
- "description": "synthetic, easyhuman, epic-kitchens-100, soccernet.",
328
- "dataType": "sc:Text",
329
- "source": {
330
- "fileObject": {
331
- "@id": "easyhuman_text"
332
- },
333
- "extract": {
334
- "column": "dataset_source"
335
- }
336
- }
337
- },
338
- {
339
- "@type": "cr:Field",
340
- "@id": "records/easyhuman_text/input_sequence",
341
- "name": "input_sequence",
342
- "description": "Comma-joined input symbols (text-mode payload).",
343
- "dataType": "sc:Text",
344
- "source": {
345
- "fileObject": {
346
- "@id": "easyhuman_text"
347
- },
348
- "extract": {
349
- "column": "input_sequence"
350
- }
351
- }
352
- },
353
- {
354
- "@type": "cr:Field",
355
- "@id": "records/easyhuman_text/source_class",
356
- "name": "source_class",
357
- "description": "Original event class label (natural-only).",
358
- "dataType": "sc:Text",
359
- "source": {
360
- "fileObject": {
361
- "@id": "easyhuman_text"
362
- },
363
- "extract": {
364
- "column": "source_class"
365
- }
366
- }
367
- },
368
- {
369
- "@type": "cr:Field",
370
- "@id": "records/easyhuman_text/source_dataset",
371
- "name": "source_dataset",
372
- "description": "Original source dataset (natural-only).",
373
- "dataType": "sc:Text",
374
- "source": {
375
- "fileObject": {
376
- "@id": "easyhuman_text"
377
- },
378
- "extract": {
379
- "column": "source_dataset"
380
- }
381
- }
382
- },
383
- {
384
- "@type": "cr:Field",
385
- "@id": "records/easyhuman_text/video_id",
386
- "name": "video_id",
387
- "description": "Canonical video identifier (joins to state files).",
388
- "dataType": "sc:Text",
389
- "source": {
390
- "fileObject": {
391
- "@id": "easyhuman_text"
392
- },
393
- "extract": {
394
- "column": "video_id"
395
- }
396
- }
397
- },
398
- {
399
- "@type": "cr:Field",
400
- "@id": "records/easyhuman_text/question_id",
401
- "name": "question_id",
402
- "description": "Canonical question identifier.",
403
- "dataType": "sc:Text",
404
- "source": {
405
- "fileObject": {
406
- "@id": "easyhuman_text"
407
- },
408
- "extract": {
409
- "column": "question_id"
410
- }
411
- }
412
- }
413
- ]
414
- },
415
- {
416
- "@type": "cr:RecordSet",
417
- "@id": "records/easyhuman_video",
418
- "name": "easyhuman_video",
419
- "description": "Substream Recollection split: easyhuman_video.",
420
- "field": [
421
- {
422
- "@type": "cr:Field",
423
- "@id": "records/easyhuman_video/split",
424
- "name": "split",
425
- "description": "Top-level split label.",
426
- "dataType": "sc:Text",
427
- "source": {
428
- "fileObject": {
429
- "@id": "easyhuman_video"
430
- },
431
- "extract": {
432
- "column": "split"
433
- }
434
- }
435
- },
436
- {
437
- "@type": "cr:Field",
438
- "@id": "records/easyhuman_video/subset",
439
- "name": "subset",
440
- "description": "Bucket-level subset id (e.g. UNIFORM_EVAL_L0128_ELOW, len_1024, L=128).",
441
- "dataType": "sc:Text",
442
- "source": {
443
- "fileObject": {
444
- "@id": "easyhuman_video"
445
- },
446
- "extract": {
447
- "column": "subset"
448
- }
449
- }
450
- },
451
- {
452
- "@type": "cr:Field",
453
- "@id": "records/easyhuman_video/length_L",
454
- "name": "length_L",
455
- "description": "Substream length L (frames or characters), normalized.",
456
- "dataType": "sc:Integer",
457
- "source": {
458
- "fileObject": {
459
- "@id": "easyhuman_video"
460
- },
461
- "extract": {
462
- "column": "length_L"
463
- }
464
- }
465
- },
466
- {
467
- "@type": "cr:Field",
468
- "@id": "records/easyhuman_video/entropy_band",
469
- "name": "entropy_band",
470
- "description": "low / medium / uniform / natural / easyhuman.",
471
- "dataType": "sc:Text",
472
- "source": {
473
- "fileObject": {
474
- "@id": "easyhuman_video"
475
- },
476
- "extract": {
477
- "column": "entropy_band"
478
- }
479
- }
480
- },
481
- {
482
- "@type": "cr:Field",
483
- "@id": "records/easyhuman_video/modality",
484
- "name": "modality",
485
- "description": "text or video.",
486
- "dataType": "sc:Text",
487
- "source": {
488
- "fileObject": {
489
- "@id": "easyhuman_video"
490
- },
491
- "extract": {
492
- "column": "modality"
493
- }
494
- }
495
- },
496
- {
497
- "@type": "cr:Field",
498
- "@id": "records/easyhuman_video/question_variant",
499
- "name": "question_variant",
500
- "description": "sequential, spatial, binary_natural, easyhuman_binary.",
501
- "dataType": "sc:Text",
502
- "source": {
503
- "fileObject": {
504
- "@id": "easyhuman_video"
505
- },
506
- "extract": {
507
- "column": "question_variant"
508
- }
509
- }
510
- },
511
- {
512
- "@type": "cr:Field",
513
- "@id": "records/easyhuman_video/video_index",
514
- "name": "video_index",
515
- "description": "Per-bucket video index (1-based).",
516
- "dataType": "sc:Integer",
517
- "source": {
518
- "fileObject": {
519
- "@id": "easyhuman_video"
520
- },
521
- "extract": {
522
- "column": "video_index"
523
- }
524
- }
525
- },
526
- {
527
- "@type": "cr:Field",
528
- "@id": "records/easyhuman_video/question_index",
529
- "name": "question_index",
530
- "description": "Per-video question index (0-based).",
531
- "dataType": "sc:Integer",
532
- "source": {
533
- "fileObject": {
534
- "@id": "easyhuman_video"
535
- },
536
- "extract": {
537
- "column": "question_index"
538
- }
539
- }
540
- },
541
- {
542
- "@type": "cr:Field",
543
- "@id": "records/easyhuman_video/question_text",
544
- "name": "question_text",
545
- "description": "Prompt the model sees (no schema scaffolding).",
546
- "dataType": "sc:Text",
547
- "source": {
548
- "fileObject": {
549
- "@id": "easyhuman_video"
550
- },
551
- "extract": {
552
- "column": "question_text"
553
- }
554
- }
555
- },
556
- {
557
- "@type": "cr:Field",
558
- "@id": "records/easyhuman_video/candidate",
559
- "name": "candidate",
560
- "description": "JSON-encoded probe metadata (sequence, clip_start, ...). Null when N/A.",
561
- "dataType": "sc:Text",
562
- "source": {
563
- "fileObject": {
564
- "@id": "easyhuman_video"
565
- },
566
- "extract": {
567
- "column": "candidate"
568
- }
569
- }
570
- },
571
- {
572
- "@type": "cr:Field",
573
- "@id": "records/easyhuman_video/answer",
574
- "name": "answer",
575
- "description": "Ground-truth label (yes/no for binary questions).",
576
- "dataType": "sc:Text",
577
- "source": {
578
- "fileObject": {
579
- "@id": "easyhuman_video"
580
- },
581
- "extract": {
582
- "column": "answer"
583
- }
584
- }
585
- },
586
- {
587
- "@type": "cr:Field",
588
- "@id": "records/easyhuman_video/clip_path",
589
- "name": "clip_path",
590
- "description": "Relative path to the probe clip file (where applicable).",
591
- "dataType": "sc:Text",
592
- "source": {
593
- "fileObject": {
594
- "@id": "easyhuman_video"
595
- },
596
- "extract": {
597
- "column": "clip_path"
598
- }
599
- }
600
- },
601
- {
602
- "@type": "cr:Field",
603
- "@id": "records/easyhuman_video/video_path",
604
- "name": "video_path",
605
- "description": "Relative path to the main video (where applicable).",
606
- "dataType": "sc:Text",
607
- "source": {
608
- "fileObject": {
609
- "@id": "easyhuman_video"
610
- },
611
- "extract": {
612
- "column": "video_path"
613
- }
614
- }
615
- },
616
- {
617
- "@type": "cr:Field",
618
- "@id": "records/easyhuman_video/dataset_source",
619
- "name": "dataset_source",
620
- "description": "synthetic, easyhuman, epic-kitchens-100, soccernet.",
621
- "dataType": "sc:Text",
622
- "source": {
623
- "fileObject": {
624
- "@id": "easyhuman_video"
625
- },
626
- "extract": {
627
- "column": "dataset_source"
628
- }
629
- }
630
- },
631
- {
632
- "@type": "cr:Field",
633
- "@id": "records/easyhuman_video/input_sequence",
634
- "name": "input_sequence",
635
- "description": "Comma-joined input symbols (text-mode payload).",
636
- "dataType": "sc:Text",
637
- "source": {
638
- "fileObject": {
639
- "@id": "easyhuman_video"
640
- },
641
- "extract": {
642
- "column": "input_sequence"
643
- }
644
- }
645
- },
646
- {
647
- "@type": "cr:Field",
648
- "@id": "records/easyhuman_video/source_class",
649
- "name": "source_class",
650
- "description": "Original event class label (natural-only).",
651
- "dataType": "sc:Text",
652
- "source": {
653
- "fileObject": {
654
- "@id": "easyhuman_video"
655
- },
656
- "extract": {
657
- "column": "source_class"
658
- }
659
- }
660
- },
661
- {
662
- "@type": "cr:Field",
663
- "@id": "records/easyhuman_video/source_dataset",
664
- "name": "source_dataset",
665
- "description": "Original source dataset (natural-only).",
666
- "dataType": "sc:Text",
667
- "source": {
668
- "fileObject": {
669
- "@id": "easyhuman_video"
670
- },
671
- "extract": {
672
- "column": "source_dataset"
673
- }
674
- }
675
- },
676
- {
677
- "@type": "cr:Field",
678
- "@id": "records/easyhuman_video/video_id",
679
- "name": "video_id",
680
- "description": "Canonical video identifier (joins to state files).",
681
- "dataType": "sc:Text",
682
- "source": {
683
- "fileObject": {
684
- "@id": "easyhuman_video"
685
- },
686
- "extract": {
687
- "column": "video_id"
688
- }
689
- }
690
- },
691
- {
692
- "@type": "cr:Field",
693
- "@id": "records/easyhuman_video/question_id",
694
- "name": "question_id",
695
- "description": "Canonical question identifier.",
696
- "dataType": "sc:Text",
697
- "source": {
698
- "fileObject": {
699
- "@id": "easyhuman_video"
700
- },
701
- "extract": {
702
- "column": "question_id"
703
- }
704
- }
705
- }
706
- ]
707
- },
708
- {
709
- "@type": "cr:RecordSet",
710
- "@id": "records/substream_text",
711
- "name": "substream_text",
712
- "description": "Substream Recollection split: substream_text.",
713
  "field": [
714
- {
715
- "@type": "cr:Field",
716
- "@id": "records/substream_text/split",
717
- "name": "split",
718
- "description": "Top-level split label.",
719
- "dataType": "sc:Text",
720
- "source": {
721
- "fileObject": {
722
- "@id": "substream_text"
723
- },
724
- "extract": {
725
- "column": "split"
726
- }
727
- }
728
- },
729
- {
730
- "@type": "cr:Field",
731
- "@id": "records/substream_text/subset",
732
- "name": "subset",
733
- "description": "Bucket-level subset id (e.g. UNIFORM_EVAL_L0128_ELOW, len_1024, L=128).",
734
- "dataType": "sc:Text",
735
- "source": {
736
- "fileObject": {
737
- "@id": "substream_text"
738
- },
739
- "extract": {
740
- "column": "subset"
741
- }
742
- }
743
- },
744
- {
745
- "@type": "cr:Field",
746
- "@id": "records/substream_text/length_L",
747
- "name": "length_L",
748
- "description": "Substream length L (frames or characters), normalized.",
749
- "dataType": "sc:Integer",
750
- "source": {
751
- "fileObject": {
752
- "@id": "substream_text"
753
- },
754
- "extract": {
755
- "column": "length_L"
756
- }
757
- }
758
- },
759
- {
760
- "@type": "cr:Field",
761
- "@id": "records/substream_text/entropy_band",
762
- "name": "entropy_band",
763
- "description": "low / medium / uniform / natural / easyhuman.",
764
- "dataType": "sc:Text",
765
- "source": {
766
- "fileObject": {
767
- "@id": "substream_text"
768
- },
769
- "extract": {
770
- "column": "entropy_band"
771
- }
772
- }
773
- },
774
- {
775
- "@type": "cr:Field",
776
- "@id": "records/substream_text/modality",
777
- "name": "modality",
778
- "description": "text or video.",
779
- "dataType": "sc:Text",
780
- "source": {
781
- "fileObject": {
782
- "@id": "substream_text"
783
- },
784
- "extract": {
785
- "column": "modality"
786
- }
787
- }
788
- },
789
- {
790
- "@type": "cr:Field",
791
- "@id": "records/substream_text/question_variant",
792
- "name": "question_variant",
793
- "description": "sequential, spatial, binary_natural, easyhuman_binary.",
794
- "dataType": "sc:Text",
795
- "source": {
796
- "fileObject": {
797
- "@id": "substream_text"
798
- },
799
- "extract": {
800
- "column": "question_variant"
801
- }
802
- }
803
- },
804
- {
805
- "@type": "cr:Field",
806
- "@id": "records/substream_text/video_index",
807
- "name": "video_index",
808
- "description": "Per-bucket video index (1-based).",
809
- "dataType": "sc:Integer",
810
- "source": {
811
- "fileObject": {
812
- "@id": "substream_text"
813
- },
814
- "extract": {
815
- "column": "video_index"
816
- }
817
- }
818
- },
819
- {
820
- "@type": "cr:Field",
821
- "@id": "records/substream_text/question_index",
822
- "name": "question_index",
823
- "description": "Per-video question index (0-based).",
824
- "dataType": "sc:Integer",
825
- "source": {
826
- "fileObject": {
827
- "@id": "substream_text"
828
- },
829
- "extract": {
830
- "column": "question_index"
831
- }
832
- }
833
- },
834
- {
835
- "@type": "cr:Field",
836
- "@id": "records/substream_text/question_text",
837
- "name": "question_text",
838
- "description": "Prompt the model sees (no schema scaffolding).",
839
- "dataType": "sc:Text",
840
- "source": {
841
- "fileObject": {
842
- "@id": "substream_text"
843
- },
844
- "extract": {
845
- "column": "question_text"
846
- }
847
- }
848
- },
849
- {
850
- "@type": "cr:Field",
851
- "@id": "records/substream_text/candidate",
852
- "name": "candidate",
853
- "description": "JSON-encoded probe metadata (sequence, clip_start, ...). Null when N/A.",
854
- "dataType": "sc:Text",
855
- "source": {
856
- "fileObject": {
857
- "@id": "substream_text"
858
- },
859
- "extract": {
860
- "column": "candidate"
861
- }
862
- }
863
- },
864
- {
865
- "@type": "cr:Field",
866
- "@id": "records/substream_text/answer",
867
- "name": "answer",
868
- "description": "Ground-truth label (yes/no for binary questions).",
869
- "dataType": "sc:Text",
870
- "source": {
871
- "fileObject": {
872
- "@id": "substream_text"
873
- },
874
- "extract": {
875
- "column": "answer"
876
- }
877
- }
878
- },
879
- {
880
- "@type": "cr:Field",
881
- "@id": "records/substream_text/clip_path",
882
- "name": "clip_path",
883
- "description": "Relative path to the probe clip file (where applicable).",
884
- "dataType": "sc:Text",
885
- "source": {
886
- "fileObject": {
887
- "@id": "substream_text"
888
- },
889
- "extract": {
890
- "column": "clip_path"
891
- }
892
- }
893
- },
894
- {
895
- "@type": "cr:Field",
896
- "@id": "records/substream_text/video_path",
897
- "name": "video_path",
898
- "description": "Relative path to the main video (where applicable).",
899
- "dataType": "sc:Text",
900
- "source": {
901
- "fileObject": {
902
- "@id": "substream_text"
903
- },
904
- "extract": {
905
- "column": "video_path"
906
- }
907
- }
908
- },
909
- {
910
- "@type": "cr:Field",
911
- "@id": "records/substream_text/dataset_source",
912
- "name": "dataset_source",
913
- "description": "synthetic, easyhuman, epic-kitchens-100, soccernet.",
914
- "dataType": "sc:Text",
915
- "source": {
916
- "fileObject": {
917
- "@id": "substream_text"
918
- },
919
- "extract": {
920
- "column": "dataset_source"
921
- }
922
- }
923
- },
924
- {
925
- "@type": "cr:Field",
926
- "@id": "records/substream_text/input_sequence",
927
- "name": "input_sequence",
928
- "description": "Comma-joined input symbols (text-mode payload).",
929
- "dataType": "sc:Text",
930
- "source": {
931
- "fileObject": {
932
- "@id": "substream_text"
933
- },
934
- "extract": {
935
- "column": "input_sequence"
936
- }
937
- }
938
- },
939
- {
940
- "@type": "cr:Field",
941
- "@id": "records/substream_text/source_class",
942
- "name": "source_class",
943
- "description": "Original event class label (natural-only).",
944
- "dataType": "sc:Text",
945
- "source": {
946
- "fileObject": {
947
- "@id": "substream_text"
948
- },
949
- "extract": {
950
- "column": "source_class"
951
- }
952
- }
953
- },
954
- {
955
- "@type": "cr:Field",
956
- "@id": "records/substream_text/source_dataset",
957
- "name": "source_dataset",
958
- "description": "Original source dataset (natural-only).",
959
- "dataType": "sc:Text",
960
- "source": {
961
- "fileObject": {
962
- "@id": "substream_text"
963
- },
964
- "extract": {
965
- "column": "source_dataset"
966
- }
967
- }
968
- },
969
- {
970
- "@type": "cr:Field",
971
- "@id": "records/substream_text/video_id",
972
- "name": "video_id",
973
- "description": "Canonical video identifier (joins to state files).",
974
- "dataType": "sc:Text",
975
- "source": {
976
- "fileObject": {
977
- "@id": "substream_text"
978
- },
979
- "extract": {
980
- "column": "video_id"
981
- }
982
- }
983
- },
984
- {
985
- "@type": "cr:Field",
986
- "@id": "records/substream_text/question_id",
987
- "name": "question_id",
988
- "description": "Canonical question identifier.",
989
- "dataType": "sc:Text",
990
- "source": {
991
- "fileObject": {
992
- "@id": "substream_text"
993
- },
994
- "extract": {
995
- "column": "question_id"
996
- }
997
- }
998
- }
999
  ]
1000
  },
1001
  {
1002
  "@type": "cr:RecordSet",
1003
- "@id": "records/substream_video",
1004
- "name": "substream_video",
1005
- "description": "Substream Recollection split: substream_video.",
1006
  "field": [
1007
- {
1008
- "@type": "cr:Field",
1009
- "@id": "records/substream_video/split",
1010
- "name": "split",
1011
- "description": "Top-level split label.",
1012
- "dataType": "sc:Text",
1013
- "source": {
1014
- "fileObject": {
1015
- "@id": "substream_video"
1016
- },
1017
- "extract": {
1018
- "column": "split"
1019
- }
1020
- }
1021
- },
1022
- {
1023
- "@type": "cr:Field",
1024
- "@id": "records/substream_video/subset",
1025
- "name": "subset",
1026
- "description": "Bucket-level subset id (e.g. UNIFORM_EVAL_L0128_ELOW, len_1024, L=128).",
1027
- "dataType": "sc:Text",
1028
- "source": {
1029
- "fileObject": {
1030
- "@id": "substream_video"
1031
- },
1032
- "extract": {
1033
- "column": "subset"
1034
- }
1035
- }
1036
- },
1037
- {
1038
- "@type": "cr:Field",
1039
- "@id": "records/substream_video/length_L",
1040
- "name": "length_L",
1041
- "description": "Substream length L (frames or characters), normalized.",
1042
- "dataType": "sc:Integer",
1043
- "source": {
1044
- "fileObject": {
1045
- "@id": "substream_video"
1046
- },
1047
- "extract": {
1048
- "column": "length_L"
1049
- }
1050
- }
1051
- },
1052
- {
1053
- "@type": "cr:Field",
1054
- "@id": "records/substream_video/entropy_band",
1055
- "name": "entropy_band",
1056
- "description": "low / medium / uniform / natural / easyhuman.",
1057
- "dataType": "sc:Text",
1058
- "source": {
1059
- "fileObject": {
1060
- "@id": "substream_video"
1061
- },
1062
- "extract": {
1063
- "column": "entropy_band"
1064
- }
1065
- }
1066
- },
1067
- {
1068
- "@type": "cr:Field",
1069
- "@id": "records/substream_video/modality",
1070
- "name": "modality",
1071
- "description": "text or video.",
1072
- "dataType": "sc:Text",
1073
- "source": {
1074
- "fileObject": {
1075
- "@id": "substream_video"
1076
- },
1077
- "extract": {
1078
- "column": "modality"
1079
- }
1080
- }
1081
- },
1082
- {
1083
- "@type": "cr:Field",
1084
- "@id": "records/substream_video/question_variant",
1085
- "name": "question_variant",
1086
- "description": "sequential, spatial, binary_natural, easyhuman_binary.",
1087
- "dataType": "sc:Text",
1088
- "source": {
1089
- "fileObject": {
1090
- "@id": "substream_video"
1091
- },
1092
- "extract": {
1093
- "column": "question_variant"
1094
- }
1095
- }
1096
- },
1097
- {
1098
- "@type": "cr:Field",
1099
- "@id": "records/substream_video/video_index",
1100
- "name": "video_index",
1101
- "description": "Per-bucket video index (1-based).",
1102
- "dataType": "sc:Integer",
1103
- "source": {
1104
- "fileObject": {
1105
- "@id": "substream_video"
1106
- },
1107
- "extract": {
1108
- "column": "video_index"
1109
- }
1110
- }
1111
- },
1112
- {
1113
- "@type": "cr:Field",
1114
- "@id": "records/substream_video/question_index",
1115
- "name": "question_index",
1116
- "description": "Per-video question index (0-based).",
1117
- "dataType": "sc:Integer",
1118
- "source": {
1119
- "fileObject": {
1120
- "@id": "substream_video"
1121
- },
1122
- "extract": {
1123
- "column": "question_index"
1124
- }
1125
- }
1126
- },
1127
- {
1128
- "@type": "cr:Field",
1129
- "@id": "records/substream_video/question_text",
1130
- "name": "question_text",
1131
- "description": "Prompt the model sees (no schema scaffolding).",
1132
- "dataType": "sc:Text",
1133
- "source": {
1134
- "fileObject": {
1135
- "@id": "substream_video"
1136
- },
1137
- "extract": {
1138
- "column": "question_text"
1139
- }
1140
- }
1141
- },
1142
- {
1143
- "@type": "cr:Field",
1144
- "@id": "records/substream_video/candidate",
1145
- "name": "candidate",
1146
- "description": "JSON-encoded probe metadata (sequence, clip_start, ...). Null when N/A.",
1147
- "dataType": "sc:Text",
1148
- "source": {
1149
- "fileObject": {
1150
- "@id": "substream_video"
1151
- },
1152
- "extract": {
1153
- "column": "candidate"
1154
- }
1155
- }
1156
- },
1157
- {
1158
- "@type": "cr:Field",
1159
- "@id": "records/substream_video/answer",
1160
- "name": "answer",
1161
- "description": "Ground-truth label (yes/no for binary questions).",
1162
- "dataType": "sc:Text",
1163
- "source": {
1164
- "fileObject": {
1165
- "@id": "substream_video"
1166
- },
1167
- "extract": {
1168
- "column": "answer"
1169
- }
1170
- }
1171
- },
1172
- {
1173
- "@type": "cr:Field",
1174
- "@id": "records/substream_video/clip_path",
1175
- "name": "clip_path",
1176
- "description": "Relative path to the probe clip file (where applicable).",
1177
- "dataType": "sc:Text",
1178
- "source": {
1179
- "fileObject": {
1180
- "@id": "substream_video"
1181
- },
1182
- "extract": {
1183
- "column": "clip_path"
1184
- }
1185
- }
1186
- },
1187
- {
1188
- "@type": "cr:Field",
1189
- "@id": "records/substream_video/video_path",
1190
- "name": "video_path",
1191
- "description": "Relative path to the main video (where applicable).",
1192
- "dataType": "sc:Text",
1193
- "source": {
1194
- "fileObject": {
1195
- "@id": "substream_video"
1196
- },
1197
- "extract": {
1198
- "column": "video_path"
1199
- }
1200
- }
1201
- },
1202
- {
1203
- "@type": "cr:Field",
1204
- "@id": "records/substream_video/dataset_source",
1205
- "name": "dataset_source",
1206
- "description": "synthetic, easyhuman, epic-kitchens-100, soccernet.",
1207
- "dataType": "sc:Text",
1208
- "source": {
1209
- "fileObject": {
1210
- "@id": "substream_video"
1211
- },
1212
- "extract": {
1213
- "column": "dataset_source"
1214
- }
1215
- }
1216
- },
1217
- {
1218
- "@type": "cr:Field",
1219
- "@id": "records/substream_video/input_sequence",
1220
- "name": "input_sequence",
1221
- "description": "Comma-joined input symbols (text-mode payload).",
1222
- "dataType": "sc:Text",
1223
- "source": {
1224
- "fileObject": {
1225
- "@id": "substream_video"
1226
- },
1227
- "extract": {
1228
- "column": "input_sequence"
1229
- }
1230
- }
1231
- },
1232
- {
1233
- "@type": "cr:Field",
1234
- "@id": "records/substream_video/source_class",
1235
- "name": "source_class",
1236
- "description": "Original event class label (natural-only).",
1237
- "dataType": "sc:Text",
1238
- "source": {
1239
- "fileObject": {
1240
- "@id": "substream_video"
1241
- },
1242
- "extract": {
1243
- "column": "source_class"
1244
- }
1245
- }
1246
- },
1247
- {
1248
- "@type": "cr:Field",
1249
- "@id": "records/substream_video/source_dataset",
1250
- "name": "source_dataset",
1251
- "description": "Original source dataset (natural-only).",
1252
- "dataType": "sc:Text",
1253
- "source": {
1254
- "fileObject": {
1255
- "@id": "substream_video"
1256
- },
1257
- "extract": {
1258
- "column": "source_dataset"
1259
- }
1260
- }
1261
- },
1262
- {
1263
- "@type": "cr:Field",
1264
- "@id": "records/substream_video/video_id",
1265
- "name": "video_id",
1266
- "description": "Canonical video identifier (joins to state files).",
1267
- "dataType": "sc:Text",
1268
- "source": {
1269
- "fileObject": {
1270
- "@id": "substream_video"
1271
- },
1272
- "extract": {
1273
- "column": "video_id"
1274
- }
1275
- }
1276
- },
1277
- {
1278
- "@type": "cr:Field",
1279
- "@id": "records/substream_video/question_id",
1280
- "name": "question_id",
1281
- "description": "Canonical question identifier.",
1282
- "dataType": "sc:Text",
1283
- "source": {
1284
- "fileObject": {
1285
- "@id": "substream_video"
1286
- },
1287
- "extract": {
1288
- "column": "question_id"
1289
- }
1290
- }
1291
- }
1292
  ]
1293
  },
1294
  {
1295
  "@type": "cr:RecordSet",
1296
- "@id": "records/natural_video",
1297
  "name": "natural_video",
1298
- "description": "Substream Recollection split: natural_video.",
1299
  "field": [
1300
- {
1301
- "@type": "cr:Field",
1302
- "@id": "records/natural_video/split",
1303
- "name": "split",
1304
- "description": "Top-level split label.",
1305
- "dataType": "sc:Text",
1306
- "source": {
1307
- "fileObject": {
1308
- "@id": "natural_video"
1309
- },
1310
- "extract": {
1311
- "column": "split"
1312
- }
1313
- }
1314
- },
1315
- {
1316
- "@type": "cr:Field",
1317
- "@id": "records/natural_video/subset",
1318
- "name": "subset",
1319
- "description": "Bucket-level subset id (e.g. UNIFORM_EVAL_L0128_ELOW, len_1024, L=128).",
1320
- "dataType": "sc:Text",
1321
- "source": {
1322
- "fileObject": {
1323
- "@id": "natural_video"
1324
- },
1325
- "extract": {
1326
- "column": "subset"
1327
- }
1328
- }
1329
- },
1330
- {
1331
- "@type": "cr:Field",
1332
- "@id": "records/natural_video/length_L",
1333
- "name": "length_L",
1334
- "description": "Substream length L (frames or characters), normalized.",
1335
- "dataType": "sc:Integer",
1336
- "source": {
1337
- "fileObject": {
1338
- "@id": "natural_video"
1339
- },
1340
- "extract": {
1341
- "column": "length_L"
1342
- }
1343
- }
1344
- },
1345
- {
1346
- "@type": "cr:Field",
1347
- "@id": "records/natural_video/entropy_band",
1348
- "name": "entropy_band",
1349
- "description": "low / medium / uniform / natural / easyhuman.",
1350
- "dataType": "sc:Text",
1351
- "source": {
1352
- "fileObject": {
1353
- "@id": "natural_video"
1354
- },
1355
- "extract": {
1356
- "column": "entropy_band"
1357
- }
1358
- }
1359
- },
1360
- {
1361
- "@type": "cr:Field",
1362
- "@id": "records/natural_video/modality",
1363
- "name": "modality",
1364
- "description": "text or video.",
1365
- "dataType": "sc:Text",
1366
- "source": {
1367
- "fileObject": {
1368
- "@id": "natural_video"
1369
- },
1370
- "extract": {
1371
- "column": "modality"
1372
- }
1373
- }
1374
- },
1375
- {
1376
- "@type": "cr:Field",
1377
- "@id": "records/natural_video/question_variant",
1378
- "name": "question_variant",
1379
- "description": "sequential, spatial, binary_natural, easyhuman_binary.",
1380
- "dataType": "sc:Text",
1381
- "source": {
1382
- "fileObject": {
1383
- "@id": "natural_video"
1384
- },
1385
- "extract": {
1386
- "column": "question_variant"
1387
- }
1388
- }
1389
- },
1390
- {
1391
- "@type": "cr:Field",
1392
- "@id": "records/natural_video/video_index",
1393
- "name": "video_index",
1394
- "description": "Per-bucket video index (1-based).",
1395
- "dataType": "sc:Integer",
1396
- "source": {
1397
- "fileObject": {
1398
- "@id": "natural_video"
1399
- },
1400
- "extract": {
1401
- "column": "video_index"
1402
- }
1403
- }
1404
- },
1405
- {
1406
- "@type": "cr:Field",
1407
- "@id": "records/natural_video/question_index",
1408
- "name": "question_index",
1409
- "description": "Per-video question index (0-based).",
1410
- "dataType": "sc:Integer",
1411
- "source": {
1412
- "fileObject": {
1413
- "@id": "natural_video"
1414
- },
1415
- "extract": {
1416
- "column": "question_index"
1417
- }
1418
- }
1419
- },
1420
- {
1421
- "@type": "cr:Field",
1422
- "@id": "records/natural_video/question_text",
1423
- "name": "question_text",
1424
- "description": "Prompt the model sees (no schema scaffolding).",
1425
- "dataType": "sc:Text",
1426
- "source": {
1427
- "fileObject": {
1428
- "@id": "natural_video"
1429
- },
1430
- "extract": {
1431
- "column": "question_text"
1432
- }
1433
- }
1434
- },
1435
- {
1436
- "@type": "cr:Field",
1437
- "@id": "records/natural_video/candidate",
1438
- "name": "candidate",
1439
- "description": "JSON-encoded probe metadata (sequence, clip_start, ...). Null when N/A.",
1440
- "dataType": "sc:Text",
1441
- "source": {
1442
- "fileObject": {
1443
- "@id": "natural_video"
1444
- },
1445
- "extract": {
1446
- "column": "candidate"
1447
- }
1448
- }
1449
- },
1450
- {
1451
- "@type": "cr:Field",
1452
- "@id": "records/natural_video/answer",
1453
- "name": "answer",
1454
- "description": "Ground-truth label (yes/no for binary questions).",
1455
- "dataType": "sc:Text",
1456
- "source": {
1457
- "fileObject": {
1458
- "@id": "natural_video"
1459
- },
1460
- "extract": {
1461
- "column": "answer"
1462
- }
1463
- }
1464
- },
1465
- {
1466
- "@type": "cr:Field",
1467
- "@id": "records/natural_video/clip_path",
1468
- "name": "clip_path",
1469
- "description": "Relative path to the probe clip file (where applicable).",
1470
- "dataType": "sc:Text",
1471
- "source": {
1472
- "fileObject": {
1473
- "@id": "natural_video"
1474
- },
1475
- "extract": {
1476
- "column": "clip_path"
1477
- }
1478
- }
1479
- },
1480
- {
1481
- "@type": "cr:Field",
1482
- "@id": "records/natural_video/video_path",
1483
- "name": "video_path",
1484
- "description": "Relative path to the main video (where applicable).",
1485
- "dataType": "sc:Text",
1486
- "source": {
1487
- "fileObject": {
1488
- "@id": "natural_video"
1489
- },
1490
- "extract": {
1491
- "column": "video_path"
1492
- }
1493
- }
1494
- },
1495
- {
1496
- "@type": "cr:Field",
1497
- "@id": "records/natural_video/dataset_source",
1498
- "name": "dataset_source",
1499
- "description": "synthetic, easyhuman, epic-kitchens-100, soccernet.",
1500
- "dataType": "sc:Text",
1501
- "source": {
1502
- "fileObject": {
1503
- "@id": "natural_video"
1504
- },
1505
- "extract": {
1506
- "column": "dataset_source"
1507
- }
1508
- }
1509
- },
1510
- {
1511
- "@type": "cr:Field",
1512
- "@id": "records/natural_video/input_sequence",
1513
- "name": "input_sequence",
1514
- "description": "Comma-joined input symbols (text-mode payload).",
1515
- "dataType": "sc:Text",
1516
- "source": {
1517
- "fileObject": {
1518
- "@id": "natural_video"
1519
- },
1520
- "extract": {
1521
- "column": "input_sequence"
1522
- }
1523
- }
1524
- },
1525
- {
1526
- "@type": "cr:Field",
1527
- "@id": "records/natural_video/source_class",
1528
- "name": "source_class",
1529
- "description": "Original event class label (natural-only).",
1530
- "dataType": "sc:Text",
1531
- "source": {
1532
- "fileObject": {
1533
- "@id": "natural_video"
1534
- },
1535
- "extract": {
1536
- "column": "source_class"
1537
- }
1538
- }
1539
- },
1540
- {
1541
- "@type": "cr:Field",
1542
- "@id": "records/natural_video/source_dataset",
1543
- "name": "source_dataset",
1544
- "description": "Original source dataset (natural-only).",
1545
- "dataType": "sc:Text",
1546
- "source": {
1547
- "fileObject": {
1548
- "@id": "natural_video"
1549
- },
1550
- "extract": {
1551
- "column": "source_dataset"
1552
- }
1553
- }
1554
- },
1555
- {
1556
- "@type": "cr:Field",
1557
- "@id": "records/natural_video/video_id",
1558
- "name": "video_id",
1559
- "description": "Canonical video identifier (joins to state files).",
1560
- "dataType": "sc:Text",
1561
- "source": {
1562
- "fileObject": {
1563
- "@id": "natural_video"
1564
- },
1565
- "extract": {
1566
- "column": "video_id"
1567
- }
1568
- }
1569
- },
1570
- {
1571
- "@type": "cr:Field",
1572
- "@id": "records/natural_video/question_id",
1573
- "name": "question_id",
1574
- "description": "Canonical question identifier.",
1575
- "dataType": "sc:Text",
1576
- "source": {
1577
- "fileObject": {
1578
- "@id": "natural_video"
1579
- },
1580
- "extract": {
1581
- "column": "question_id"
1582
- }
1583
- }
1584
- }
1585
  ]
1586
  }
1587
- ],
1588
- "creator": {
1589
- "@type": "sc:Organization",
1590
- "name": "Anonymous Authors"
1591
- },
1592
- "author": {
1593
- "@type": "sc:Person",
1594
- "name": "Anonymous Authors"
1595
- },
1596
- "datePublished": "2026-05-03"
1597
- }
 
47
  "transform": "cr:transform"
48
  },
49
  "@type": "sc:Dataset",
50
+ "name": "Substream Recollection",
51
+ "description": "Substream Recollection is a controlled benchmark for testing whether long-context VLMs / LLMs can recall whether a short probe substream occurred inside a longer parent stream. It pairs synthetic streams (low / medium / uniform entropy, lengths L=8..4096) with EasyHuman rendered 3-belt videos and a natural-video sub-benchmark drawn from EPIC-Kitchens-100 and SoccerNet. The dataset is organized into three top-level splits: text (text-modality questions), synthetic_video (rendered synthetic and EasyHuman videos), and natural_video (real-world videos).",
52
  "conformsTo": "http://mlcommons.org/croissant/1.0",
53
  "citeAs": "Anonymous, \"Substream Recollection,\" 2026 (anonymized for review).",
54
  "keywords": [
 
59
  "vision-language-models"
60
  ],
61
  "license": "https://creativecommons.org/licenses/by/4.0/",
62
+ "url": "https://huggingface.co/datasets/anonstreammem/substream-recollection",
63
+ "version": "2.0.0",
64
+ "creator": {
65
+ "@type": "sc:Organization",
66
+ "name": "Anonymous Authors"
67
+ },
68
+ "author": {
69
+ "@type": "sc:Person",
70
+ "name": "Anonymous Authors"
71
+ },
72
+ "datePublished": "2026-05-03",
73
  "distribution": [
74
  {
75
  "@type": "cr:FileObject",
76
+ "@id": "text-parquet",
77
+ "name": "text-parquet",
78
+ "description": "Parquet file for the text split (substream_text + easyhuman_text merged; subset column distinguishes them).",
79
+ "contentUrl": "text/questions.parquet",
 
80
  "encodingFormat": "application/x-parquet",
81
+ "contentSize": "495194",
82
+ "sha256": "b563967fedef8399c0725213d12a4796f6f4d47962c3391867723724bc966bcb"
83
  },
84
  {
85
  "@type": "cr:FileObject",
86
+ "@id": "synthetic_video-parquet",
87
+ "name": "synthetic_video-parquet",
88
+ "description": "Parquet file for the synthetic_video split (substream_video + easyhuman_video merged; subset column distinguishes them).",
89
+ "contentUrl": "synthetic_video/questions.parquet",
 
90
  "encodingFormat": "application/x-parquet",
91
+ "contentSize": "297414",
92
+ "sha256": "8410ae2c1e8f3883ebb77571698c992d02bc1756a7369f993817503810e3b2b7"
93
  },
94
  {
95
  "@type": "cr:FileObject",
96
+ "@id": "natural_video-parquet",
97
+ "name": "natural_video-parquet",
98
+ "description": "Parquet file for the natural_video split (EPIC-Kitchens-100 derived clips and SoccerNet provenance metadata).",
99
+ "contentUrl": "natural_video/questions.parquet",
 
100
  "encodingFormat": "application/x-parquet",
101
+ "contentSize": "32728",
102
+ "sha256": "f3c4b0309a0977571c9c51d976a201430a1a5ff4c947007cdccdeb4f96ffd140"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
103
  }
104
  ],
105
  "recordSet": [
106
  {
107
  "@type": "cr:RecordSet",
108
+ "@id": "text",
109
+ "name": "text",
110
+ "description": "Text-modality questions for both the synthetic substream benchmark and EasyHuman. The subset column is 'substream' or 'easyhuman'; the bucket column preserves the original bucket id (e.g. UNIFORM_EVAL_L0128_ELOW, len_256). License: CC BY 4.0.",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
111
  "field": [
112
+ {"@type": "cr:Field", "@id": "text/subset", "name": "subset", "description": "'substream' or 'easyhuman' (which sub-benchmark the row belongs to).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "subset"}}},
113
+ {"@type": "cr:Field", "@id": "text/bucket", "name": "bucket", "description": "Original bucket id (e.g. UNIFORM_EVAL_L0128_ELOW, len_256).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "bucket"}}},
114
+ {"@type": "cr:Field", "@id": "text/source_split", "name": "source_split", "description": "Original split label before the merge ('substream_text' or 'easyhuman_text').", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "source_split"}}},
115
+ {"@type": "cr:Field", "@id": "text/length_L", "name": "length_L", "description": "Substream length L (number of symbols).", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "length_L"}}},
116
+ {"@type": "cr:Field", "@id": "text/entropy_band", "name": "entropy_band", "description": "low / medium / uniform / easyhuman.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "entropy_band"}}},
117
+ {"@type": "cr:Field", "@id": "text/modality", "name": "modality", "description": "'text'.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "modality"}}},
118
+ {"@type": "cr:Field", "@id": "text/question_variant", "name": "question_variant", "description": "sequential / spatial / easyhuman_binary.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "question_variant"}}},
119
+ {"@type": "cr:Field", "@id": "text/video_index", "name": "video_index", "description": "Per-bucket video index (1-based).", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "video_index"}}},
120
+ {"@type": "cr:Field", "@id": "text/question_index", "name": "question_index", "description": "Per-video question index (0-based).", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "question_index"}}},
121
+ {"@type": "cr:Field", "@id": "text/question_text", "name": "question_text", "description": "Prompt the model sees (no schema scaffolding).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "question_text"}}},
122
+ {"@type": "cr:Field", "@id": "text/candidate", "name": "candidate", "description": "JSON-encoded probe metadata (sequence, clip_start, ...). Null when N/A.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "candidate"}}},
123
+ {"@type": "cr:Field", "@id": "text/answer", "name": "answer", "description": "Ground-truth label (yes/no).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "answer"}}},
124
+ {"@type": "cr:Field", "@id": "text/clip_path", "name": "clip_path", "description": "Always null for text rows (kept for cross-split schema parity).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "clip_path"}}},
125
+ {"@type": "cr:Field", "@id": "text/video_path", "name": "video_path", "description": "Always null for text rows (kept for cross-split schema parity).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "video_path"}}},
126
+ {"@type": "cr:Field", "@id": "text/dataset_source", "name": "dataset_source", "description": "synthetic / easyhuman.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "dataset_source"}}},
127
+ {"@type": "cr:Field", "@id": "text/input_sequence", "name": "input_sequence", "description": "Comma-joined input symbols (text-mode payload).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "input_sequence"}}},
128
+ {"@type": "cr:Field", "@id": "text/source_class", "name": "source_class", "description": "Always null for text split.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "source_class"}}},
129
+ {"@type": "cr:Field", "@id": "text/source_dataset", "name": "source_dataset", "description": "Always null for text split.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "source_dataset"}}},
130
+ {"@type": "cr:Field", "@id": "text/video_id", "name": "video_id", "description": "Canonical video identifier.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "video_id"}}},
131
+ {"@type": "cr:Field", "@id": "text/question_id", "name": "question_id", "description": "Canonical question identifier.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "text-parquet"}, "extract": {"column": "question_id"}}}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
132
  ]
133
  },
134
  {
135
  "@type": "cr:RecordSet",
136
+ "@id": "synthetic_video",
137
+ "name": "synthetic_video",
138
+ "description": "Video-modality questions over rendered synthetic streams (substream_video) and EasyHuman 3-belt videos (easyhuman_video). The subset column is 'substream' or 'easyhuman'; bucket preserves the original bucket id; video_path / clip_path are relative to this directory. License: CC BY 4.0.",
139
  "field": [
140
+ {"@type": "cr:Field", "@id": "synthetic_video/subset", "name": "subset", "description": "'substream' or 'easyhuman'.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "subset"}}},
141
+ {"@type": "cr:Field", "@id": "synthetic_video/bucket", "name": "bucket", "description": "Original bucket id (e.g. UNIFORM_EVAL_L0128_ELOW, len_256).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "bucket"}}},
142
+ {"@type": "cr:Field", "@id": "synthetic_video/source_split", "name": "source_split", "description": "Original split label before the merge.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "source_split"}}},
143
+ {"@type": "cr:Field", "@id": "synthetic_video/length_L", "name": "length_L", "description": "Substream length L (frames).", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "length_L"}}},
144
+ {"@type": "cr:Field", "@id": "synthetic_video/entropy_band", "name": "entropy_band", "description": "low / medium / uniform / easyhuman.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "entropy_band"}}},
145
+ {"@type": "cr:Field", "@id": "synthetic_video/modality", "name": "modality", "description": "'video'.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "modality"}}},
146
+ {"@type": "cr:Field", "@id": "synthetic_video/question_variant", "name": "question_variant", "description": "sequential / spatial / easyhuman_binary.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "question_variant"}}},
147
+ {"@type": "cr:Field", "@id": "synthetic_video/video_index", "name": "video_index", "description": "Per-bucket video index (1-based).", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "video_index"}}},
148
+ {"@type": "cr:Field", "@id": "synthetic_video/question_index", "name": "question_index", "description": "Per-video question index (0-based).", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "question_index"}}},
149
+ {"@type": "cr:Field", "@id": "synthetic_video/question_text", "name": "question_text", "description": "Prompt the model sees.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "question_text"}}},
150
+ {"@type": "cr:Field", "@id": "synthetic_video/candidate", "name": "candidate", "description": "JSON-encoded probe metadata.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "candidate"}}},
151
+ {"@type": "cr:Field", "@id": "synthetic_video/answer", "name": "answer", "description": "Ground-truth label (yes/no).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "answer"}}},
152
+ {"@type": "cr:Field", "@id": "synthetic_video/clip_path", "name": "clip_path", "description": "Relative path to the probe clip mp4 (under synthetic_video/).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "clip_path"}}},
153
+ {"@type": "cr:Field", "@id": "synthetic_video/video_path", "name": "video_path", "description": "Relative path to the parent video mp4 (under synthetic_video/).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "video_path"}}},
154
+ {"@type": "cr:Field", "@id": "synthetic_video/dataset_source", "name": "dataset_source", "description": "synthetic / easyhuman.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "dataset_source"}}},
155
+ {"@type": "cr:Field", "@id": "synthetic_video/input_sequence", "name": "input_sequence", "description": "Comma-joined input symbols.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "input_sequence"}}},
156
+ {"@type": "cr:Field", "@id": "synthetic_video/source_class", "name": "source_class", "description": "Always null for synthetic_video split.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "source_class"}}},
157
+ {"@type": "cr:Field", "@id": "synthetic_video/source_dataset", "name": "source_dataset", "description": "Always null for synthetic_video split.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "source_dataset"}}},
158
+ {"@type": "cr:Field", "@id": "synthetic_video/video_id", "name": "video_id", "description": "Canonical video identifier.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "video_id"}}},
159
+ {"@type": "cr:Field", "@id": "synthetic_video/question_id", "name": "question_id", "description": "Canonical question identifier.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "synthetic_video-parquet"}, "extract": {"column": "question_id"}}}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
160
  ]
161
  },
162
  {
163
  "@type": "cr:RecordSet",
164
+ "@id": "natural_video",
165
  "name": "natural_video",
166
+ "description": "Natural-video benchmark drawn from EPIC-Kitchens-100 and SoccerNet at L in {8,16,64,128,512,1024}. EPIC-Kitchens-100 derived clips ship in this repo (license: CC BY-NC 4.0). SoccerNet rows do NOT ship the underlying mp4s (NDA-gated source); the source_provenance column carries provenance metadata so the originals can be obtained from soccer-net.org after signing the SoccerNet NDA. Mixed license: CC BY-NC 4.0 (EPIC-Kitchens-100 derivatives) and SoccerNet NDA terms (provenance metadata only).",
167
  "field": [
168
+ {"@type": "cr:Field", "@id": "natural_video/subset", "name": "subset", "description": "Bucket id (e.g. L=8, L=128).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "subset"}}},
169
+ {"@type": "cr:Field", "@id": "natural_video/source_split", "name": "source_split", "description": "Always 'natural_video'.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "source_split"}}},
170
+ {"@type": "cr:Field", "@id": "natural_video/length_L", "name": "length_L", "description": "Stream length L.", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "length_L"}}},
171
+ {"@type": "cr:Field", "@id": "natural_video/entropy_band", "name": "entropy_band", "description": "'natural'.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "entropy_band"}}},
172
+ {"@type": "cr:Field", "@id": "natural_video/modality", "name": "modality", "description": "'video'.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "modality"}}},
173
+ {"@type": "cr:Field", "@id": "natural_video/question_variant", "name": "question_variant", "description": "'binary_natural'.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "question_variant"}}},
174
+ {"@type": "cr:Field", "@id": "natural_video/video_index", "name": "video_index", "description": "Per-bucket video index.", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "video_index"}}},
175
+ {"@type": "cr:Field", "@id": "natural_video/question_index", "name": "question_index", "description": "Per-video question index.", "dataType": "sc:Integer", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "question_index"}}},
176
+ {"@type": "cr:Field", "@id": "natural_video/question_text", "name": "question_text", "description": "Prompt the model sees.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "question_text"}}},
177
+ {"@type": "cr:Field", "@id": "natural_video/candidate", "name": "candidate", "description": "JSON-encoded probe metadata. Null for natural rows.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "candidate"}}},
178
+ {"@type": "cr:Field", "@id": "natural_video/answer", "name": "answer", "description": "Ground-truth label (yes/no).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "answer"}}},
179
+ {"@type": "cr:Field", "@id": "natural_video/clip_path", "name": "clip_path", "description": "Always null for natural rows.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "clip_path"}}},
180
+ {"@type": "cr:Field", "@id": "natural_video/video_path", "name": "video_path", "description": "Relative path to the video mp4 under natural_video/. Null/missing for SoccerNet rows.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "video_path"}}},
181
+ {"@type": "cr:Field", "@id": "natural_video/dataset_source", "name": "dataset_source", "description": "epic-kitchens-100 / soccernet.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "dataset_source"}}},
182
+ {"@type": "cr:Field", "@id": "natural_video/input_sequence", "name": "input_sequence", "description": "Always null for natural rows.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "input_sequence"}}},
183
+ {"@type": "cr:Field", "@id": "natural_video/source_class", "name": "source_class", "description": "Original event class label (e.g. clean_counter, red_card).", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "source_class"}}},
184
+ {"@type": "cr:Field", "@id": "natural_video/source_dataset", "name": "source_dataset", "description": "epic-kitchens-100 / soccernet.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "source_dataset"}}},
185
+ {"@type": "cr:Field", "@id": "natural_video/video_id", "name": "video_id", "description": "Canonical video identifier.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "video_id"}}},
186
+ {"@type": "cr:Field", "@id": "natural_video/question_id", "name": "question_id", "description": "Canonical question identifier.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "question_id"}}},
187
+ {"@type": "cr:Field", "@id": "natural_video/source_provenance", "name": "source_provenance", "description": "JSON describing the source match, half, time range, and license. Populated for SoccerNet rows; null elsewhere.", "dataType": "sc:Text", "source": {"fileObject": {"@id": "natural_video-parquet"}, "extract": {"column": "source_provenance"}}}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
188
  ]
189
  }
190
+ ]
191
+ }
 
 
 
 
 
 
 
 
 
natural_video/questions.json ADDED
The diff for this file is too large to render. See raw diff
 
easyhuman_text.parquet → natural_video/questions.parquet RENAMED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c999b026d494a507ded2d452c8924979961fe2c487ab742e4ccae45461495cfc
3
- size 35488
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3c4b0309a0977571c9c51d976a201430a1a5ff4c947007cdccdeb4f96ffd140
3
+ size 32728
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_neg_01.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_neg_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_pos_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/clean_counter_pos_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/cut_nontomato_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/cut_nontomato_neg_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_cupboard_neg_03.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_drawer_neg_02.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_drawer_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_neg_02.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_neg_03.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_pos_03.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_pos_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/open_fridge_pos_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_neg_02.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_neg_03.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_pos_02.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_pos_03.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_pos_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/pour_water_pos_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/rinse_X_neg_03.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/rinse_X_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/rinse_X_pos_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/rinse_X_pos_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/turn_on_tap_neg_02.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/turn_on_tap_neg_03.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/turn_on_tap_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/turn_on_tap_pos_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_neg_01.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_neg_02.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_neg_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_pos_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wash_plate_pos_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_neg_02.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_neg_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_neg_05.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_pos_04.mp4 RENAMED
File without changes
{exact1fps_B5/videos → natural_video/videos/nat_1024_frames}/wipe_surface_pos_05.mp4 RENAMED
File without changes
{exact1fps_B3/videos → natural_video/videos/nat_128_frames}/clean_counter_neg_01.mp4 RENAMED
File without changes
{exact1fps_B3/videos → natural_video/videos/nat_128_frames}/clean_counter_neg_02.mp4 RENAMED
File without changes