Complete verified EEF companion release with explicit action semantics and reader

.gitattributes

@@ -59,3 +59,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
 metadata/episode_outcomes.jsonl filter=lfs diff=lfs merge=lfs -text
+eef/episodes.jsonl filter=lfs diff=lfs merge=lfs -text

README.md

@@ -72,6 +72,68 @@ rows = np.fromfile("<key>.bin", dtype="<f8").reshape(metadata["num_steps"], 32)
 `dataset_info.json` summarizes the release. `shards.jsonl` provides archive sizes,
 episode counts, and SHA-256 checksums. Episode metadata contains only public data fields.
 
+## EEF action/state companions
+
+The optional `eef/` companion release covers every episode in this processed subset,
+using the same stable `episode_id` and exactly the same retained video frames. The
+original joint binaries and videos remain available. Companion TARs contain
+`<episode_id>.npz` arrays and `<episode_id>.json` frame/pose provenance.
+
+`state_pose` is `[T, 2, 7]` float32: XYZ in metres followed by quaternion **xyzw**,
+left arm then right arm. `command_pose` uses the same layout; when a recorded command
+stream exactly copies the state, the array is stored once and the metadata declares
+the alias. Gripper arrays use **0 = closed, 1 = open**. Boolean pose and gripper masks
+must be respected. `source_frame_index` preserves the video-to-source row mapping.
+
+Select action semantics explicitly in `eef/reader.py`:
+
+- `action_mode="command"`: the corresponding recorded-command pose, with no added
+  timestep shift. Copied source pose commands are excluded from supervision by default.
+- `action_mode="next_state"`: the achieved state at the next retained frame, including
+  gripper state. This is a derived target, not an original controller command. The
+  final target is masked; do not shift it again in another loader.
+
+AgiBot retains the source robot/ego-frame flange state poses. All 142,685 episodes
+have valid finite state-pose arrays. Native EEF commands exactly copy state poses
+in 141,318 episodes. Of the remaining episodes, 1,364 have shorter command streams
+and three have distinct dense commands. Sparse commands are aligned only at valid
+source indices; the fifty episodes with unresolved indices have pose-command masks
+set to false. Uncommanded rows are not silently padded with state poses.
+
+For the OpenWAM-Alpha future-state recipe, select `action_mode="next_state"`.
+`command` mode preserves the native command distinction and its missing-data masks;
+it cannot recover independent commands from the source copies. Gripper commands
+remain separate. The exact URDF root origin and dataset-wide visual projection
+accuracy have not been independently certified.
+
+The reader emits `[T, 80]` state/action arrays and masks by default, with the left
+20-D schema arm in dimensions `0:10` and the right arm in `34:44`; the unused slots
+are masked. Set `unify_action=False` for `[T, 20]`: each arm is XYZ + the first two
+rotation-matrix columns (6D rotation) + gripper. Poses use the frames named in each
+episode, not camera coordinates or a universal fingertip TCP.
+
+From a downloaded dataset root, join an existing episode's metadata to its companion:
+
+```python
+from eef.reader import load_index, load_indexed_episode
+
+eef_index = load_index("eef/episodes.jsonl")
+targets = load_indexed_episode(
+    ".", eef_index[metadata["episode_id"]], action_mode="next_state"
+)
+# targets["state"], targets["action"], targets["state_mask"], targets["action_mask"]
+```
+
+`eef/episodes.jsonl` supplies direct TAR offsets. `eef/shards.jsonl` supplies file
+sizes and SHA-256 checksums; `eef/info.json` records full coverage and verification.
+`eef/models.json` documents frames, kinematics and limitations. `eef/statistics.json`
+contains separate state, command and next-state statistics from the **training split
+only**; stored poses are unnormalized. Keep rotation-6D coordinates unnormalized.
+
+The portable reader matches the OpenWAM pose layout; integrating these companions
+with a training pipeline's video/window reader is still required. The original
+OpenWAM LeRobot reader does not directly consume these WebDataset companions.
+
 ## License and attribution
 
 This processed release retains the upstream **Creative Commons Attribution-NonCommercial-ShareAlike

dataset_info.json

@@ -19,5 +19,14 @@
       "hours": 231.44475925925926
     }
   ],
-  "format": "WebDataset tar shards; one .mp4, .bin and .json per episode"
+  "format": "WebDataset tar shards; one .mp4, .bin and .json per episode",
+  "eef": {
+    "info": "eef/info.json",
+    "episodes": 142685,
+    "frames": 251650007,
+    "bytes": 9842554880,
+    "index": "eef/episodes.jsonl",
+    "reader": "eef/reader.py",
+    "verified": true
+  }
 }

eef/episodes.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44238347e6ceb8754cb4245569e991e3bba0a8f8d351af79573f8f29833f8d2e
+size 38136483

eef/info.json

@@ -0,0 +1,27 @@
+{
+  "schema_version": 1,
+  "episodes": 142685,
+  "frames": 251650007,
+  "shards": 558,
+  "bytes": 9842554880,
+  "counts": {
+    "episodes": 142685,
+    "frames": 251650007,
+    "copied_command_episodes": 141318,
+    "unresolved_sparse_command_episodes": 50,
+    "command_review_episodes": 0,
+    "state_invalid_arm_frames": 0,
+    "command_invalid_arm_frames": 975216
+  },
+  "coverage_complete": true,
+  "all_shards_checksum_verified": true,
+  "all_frames_pose_schema_verified": true,
+  "data_index": "eef/episodes.jsonl",
+  "shard_manifest": "eef/shards.jsonl",
+  "statistics": "eef/statistics.json",
+  "reader": "eef/reader.py",
+  "pose_format": "float32 XYZ metres + xyzw quaternion, [frames,2,7]",
+  "actions": "Recorded-command poses remain separate from explicitly derived next-state targets. Select action_mode in the reader.",
+  "camera_reprojection_validation": "Source poses retained; earlier qualitative projection checks covered one episode, not a full visual validation.",
+  "models": "eef/models.json"
+}

eef/models.json

@@ -0,0 +1,18 @@
+{
+  "schema_version": 1,
+  "position_units": "metres",
+  "quaternion_order": "xyzw",
+  "endpoint_contract": "Rigid terminal-arm attachment independent of finger articulation; no universal gripper-centre TCP",
+  "per_episode_frame_names": "base_frames and endpoints in each companion JSON",
+  "camera_geometry": "Existing per-view camera metadata remains separate. Reprojection is not certified across all episodes.",
+  "AgiBot": {
+    "base_frame": "source_robot_ego",
+    "endpoint": "source robot arm flange",
+    "orientation": "xyzw",
+    "source": "https://huggingface.co/datasets/agibot-world/AgiBotWorld-Beta#explanation-of-proprioceptive-state",
+    "state_pose_source": "state/end/position and state/end/orientation",
+    "command_pose_source": "action/end/position and action/end/orientation",
+    "exact_URDF_root_origin_independently_verified": false,
+    "command_alignment": "Dense same-frame or explicit sparse source indices; unresolved indices are masked, not filled from states."
+  }
+}

eef/reader.py

@@ -0,0 +1,99 @@
+"""Read an EEF companion and construct explicitly selected OpenWAM-style targets.
+
+No temporal shift is applied to command actions. next_state is an explicitly
+derived, one-retained-frame-ahead achieved-state target, with terminal masking.
+"""
+import io
+import json
+from pathlib import Path
+import tarfile
+
+import numpy as np
+
+
+def load_index(path):
+    with open(path, encoding='utf-8') as stream:
+        return {row['episode_id']:row for row in map(json.loads,stream)}
+
+
+def load_indexed_episode(dataset_root, record, *, action_mode, unify_action=True, allow_copied_commands=False):
+    with (Path(dataset_root)/record['shard']).open('rb') as stream:
+        stream.seek(record['json_offset']);metadata=json.loads(stream.read(record['json_bytes']))
+        stream.seek(record['npz_offset']);data=stream.read(record['npz_bytes'])
+    if metadata['episode_id']!=record['episode_id']:
+        raise ValueError('EEF episode index does not match shard contents')
+    with np.load(io.BytesIO(data),allow_pickle=False) as archive:
+        arrays={key:archive[key] for key in archive.files}
+    result=build_arrays(arrays,metadata,action_mode=action_mode,unify_action=unify_action,
+                        allow_copied_commands=allow_copied_commands)
+    result['metadata']=metadata
+    return result
+
+
+def _pose20(pose, grip, pose_valid, grip_valid):
+    q = pose[..., 3:7]
+    x, y, z, w = np.moveaxis(q, -1, 0)
+    col0 = np.stack([1-2*(y*y+z*z), 2*(x*y+z*w), 2*(x*z-y*w)], axis=-1)
+    col1 = np.stack([2*(x*y-z*w), 1-2*(x*x+z*z), 2*(y*z+x*w)], axis=-1)
+    value = np.concatenate([pose[..., :3], col0, col1, grip[..., None]], axis=-1).reshape(len(pose), 20)
+    mask = np.concatenate([np.repeat(pose_valid[..., None], 9, axis=-1), grip_valid[..., None]], axis=-1).reshape(len(pose), 20)
+    return np.where(mask, value, 0).astype(np.float32), mask
+
+
+def build_arrays(arrays, metadata, *, action_mode, unify_action=True, allow_copied_commands=False):
+    if action_mode not in {'command', 'next_state'}:
+        raise ValueError('Explicitly choose action_mode=command or next_state')
+    state, state_mask = _pose20(arrays['state_pose'], arrays['state_gripper'],
+                               arrays['state_pose_valid'], arrays['state_gripper_valid'])
+    if action_mode == 'next_state':
+        action = np.zeros_like(state); action_mask = np.zeros_like(state_mask)
+        action[:-1] = state[1:]; action_mask[:-1] = state_mask[1:]
+    else:
+        pose = arrays['command_pose'] if 'command_pose' in arrays else arrays['state_pose']
+        pose_mask = arrays['command_pose_valid'].copy()
+        if not allow_copied_commands:
+            pose_mask[:, np.asarray(metadata.get('command_pose_is_state_copy', [False, False]), dtype=bool)] = False
+        action, action_mask = _pose20(pose, arrays['command_gripper'], pose_mask, arrays['command_gripper_valid'])
+        if metadata.get('command_quality_review_required', False):
+            action_mask[:]=False;action[:]=0
+    if unify_action:
+        def scatter(value):
+            out = np.zeros((len(value), 80), dtype=value.dtype)
+            out[:, :10] = value[:, :10]; out[:, 34:44] = value[:, 10:20]
+            return out
+        state, action, state_mask, action_mask = map(scatter, [state, action, state_mask, action_mask])
+    return {'state': state, 'action': action, 'state_mask': state_mask, 'action_mask': action_mask,
+            'episode_id': metadata['episode_id'], 'fps': metadata['fps'], 'action_mode': action_mode}
+
+
+def load_episode(shard_path, episode_id, *, action_mode, unify_action=True, allow_copied_commands=False):
+    with tarfile.open(shard_path, 'r') as shard:
+        metadata = json.load(shard.extractfile(episode_id+'.json'))
+        data = shard.extractfile(episode_id+'.npz').read()
+    with np.load(io.BytesIO(data), allow_pickle=False) as archive:
+        arrays = {key: archive[key] for key in archive.files}
+    result = build_arrays(arrays, metadata, action_mode=action_mode, unify_action=unify_action,
+                          allow_copied_commands=allow_copied_commands)
+    result['metadata'] = metadata
+    return result
+
+
+def iter_episodes(shard_path, *, action_mode, unify_action=True, allow_copied_commands=False):
+    """Stream companions without repeatedly scanning the TAR directory."""
+    with tarfile.open(shard_path, 'r|') as shard:
+        pending = {}
+        for member in shard:
+            if not member.isfile():
+                continue
+            key, suffix = member.name.rsplit('.', 1)
+            data = shard.extractfile(member).read()
+            if suffix == 'npz':
+                pending[key] = data
+            elif suffix == 'json':
+                metadata = json.loads(data)
+                with np.load(io.BytesIO(pending.pop(key)), allow_pickle=False) as archive:
+                    arrays = {name: archive[name] for name in archive.files}
+                result = build_arrays(arrays, metadata, action_mode=action_mode, unify_action=unify_action,
+                                      allow_copied_commands=allow_copied_commands)
+                result['metadata'] = metadata
+                yield result

eef/statistics.json

@@ -0,0 +1,358 @@
+{
+  "scope": "training split only; exact valid-frame moments in the 20-D layout",
+  "rotation_6d_dimensions": [
+    3,
+    4,
+    5,
+    6,
+    7,
+    8,
+    13,
+    14,
+    15,
+    16,
+    17,
+    18
+  ],
+  "normalization_note": "Keep rotation-6D coordinates unnormalized. Grippers already use 0=closed, 1=open. No normalization has been applied to stored poses.",
+  "robots": {
+    "AgiBot": {
+      "state": {
+        "count": [
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973,
+          226653973
+        ],
+        "mean": [
+          0.6631940055623855,
+          0.28903776350773996,
+          0.6848695329894116,
+          -0.7391199948147912,
+          -0.3534778529198422,
+          -0.03804945112269133,
+          -0.08077497391340434,
+          0.2133219088640874,
+          -0.7830034043435947,
+          0.6534640223098075,
+          0.6726976926892646,
+          -0.3059446934729797,
+          0.688528989931976,
+          0.6839481652597392,
+          -0.3573610420599532,
+          -0.03062644217731558,
+          -0.13480754501976044,
+          -0.2143777791881827,
+          -0.7010732802430177,
+          0.6166179499517503
+        ],
+        "std": [
+          0.07567273153139382,
+          0.09814911103510317,
+          0.1805050808541264,
+          0.3329455342053897,
+          0.3059807610387076,
+          0.3504715753468137,
+          0.3677712418642141,
+          0.3222813887201011,
+          0.3094411991090402,
+          0.4214729326040272,
+          0.07974220898004956,
+          0.09671544900978295,
+          0.17572542533404814,
+          0.3546607490349007,
+          0.33838840500453254,
+          0.40407803002313414,
+          0.4096834737141193,
+          0.39855031821611775,
+          0.34304874944311853,
+          0.4223983257720661
+        ],
+        "min": [
+          0.19337546825408936,
+          -0.21049153804779053,
+          0.028447948396205902,
+          -1.0,
+          -1.0000001192092896,
+          -0.9999998211860657,
+          -1.0,
+          -0.9999995231628418,
+          -1.0000001192092896,
+          0.0,
+          0.12499918043613434,
+          -0.7061007022857666,
+          0.04709611460566521,
+          -0.9999996423721313,
+          -1.0,
+          -1.0,
+          -1.0,
+          -1.000000238418579,
+          -1.0,
+          0.0
+        ],
+        "max": [
+          0.9986844062805176,
+          0.6781377792358398,
+          1.3417878150939941,
+          0.9999530911445618,
+          0.9999873042106628,
+          1.0,
+          0.9999998211860657,
+          1.0,
+          0.9999994039535522,
+          1.0,
+          0.9906591176986694,
+          0.17973941564559937,
+          1.427522897720337,
+          1.0,
+          0.9999999403953552,
+          1.0,
+          0.9999999403953552,
+          0.9999999403953552,
+          0.9999961256980896,
+          1.0
+        ]
+      },
+      "command": {
+        "count": [
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          226653973,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          2210797,
+          226653973
+        ],
+        "mean": [
+          0.656436681430082,
+          0.2608774757977724,
+          0.6076868538718683,
+          -0.7779111880564965,
+          -0.3161532700969761,
+          -0.05264641584912366,
+          -0.03332808070202698,
+          0.2400891553326224,
+          -0.8287253533851208,
+          0.6054802055936307,
+          0.6835084985622746,
+          -0.33080879645638805,
+          0.6251254181777653,
+          0.8083921816747113,
+          -0.3341265157002092,
+          0.09681055862627627,
+          -0.08063765990062455,
+          -0.3919620871317091,
+          -0.7650314796543002,
+          0.5700910382496689
+        ],
+        "std": [
+          0.0589226558832337,
+          0.09187304601342013,
+          0.16131641785495893,
+          0.25755233603542377,
+          0.3572854517753128,
+          0.3132787079059469,
+          0.34125909041301195,
+          0.22299815987875485,
+          0.2971107191411552,
+          0.48777969288176537,
+          0.07479180362774607,
+          0.09265278909304117,
+          0.15343198088570328,
+          0.20459239053342107,
+          0.3117762765455571,
+          0.29398431127163194,
+          0.30274027949713583,
+          0.31765654638206936,
+          0.24906378126293655,
+          0.49341158713438077
+        ],
+        "min": [
+          0.3886728882789612,
+          -0.07427632063627243,
+          0.0765778124332428,
+          -0.9999990463256836,
+          -1.0,
+          -0.9999237656593323,
+          -0.9869900941848755,
+          -0.7971467971801758,
+          -0.9999998211860657,
+          0.0,
+          0.3664100468158722,
+          -0.652158260345459,
+          0.04780077561736107,
+          -0.879040002822876,
+          -0.9999977350234985,
+          -0.9999943971633911,
+          -0.9996262192726135,
+          -0.9999500513076782,
+          -0.9999997019767761,
+          0.0
+        ],
+        "max": [
+          0.8646109700202942,
+          0.5344367623329163,
+          1.1397204399108887,
+          0.5227402448654175,
+          0.9996250867843628,
+          0.9939199686050415,
+          0.998774528503418,
+          0.9996106624603271,
+          0.7573980093002319,
+          1.0,
+          0.8876892924308777,
+          0.13359856605529785,
+          1.0214377641677856,
+          0.9999983310699463,
+          0.9942952394485474,
+          0.989382803440094,
+          0.9954850673675537,
+          0.9435998201370239,
+          0.9999963045120239,
+          1.0
+        ]
+      },
+      "next_state": {
+        "count": [
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443,
+          226525443
+        ],
+        "mean": [
+          0.6632270325653583,
+          0.2890253108023381,
+          0.6848328136441091,
+          -0.7390149227162651,
+          -0.353546572147465,
+          -0.03804223189876352,
+          -0.08081961877144499,
+          0.21336266056909922,
+          -0.7829138452007917,
+          0.6533071234561998,
+          0.6727351420221108,
+          -0.305938222147864,
+          0.6884945355287867,
+          0.6838114941815796,
+          -0.35744255912173295,
+          -0.030669607525611448,
+          -0.13488318989895443,
+          -0.21440027382037877,
+          -0.7009672427838373,
+          0.6164417004412756
+        ],
+        "std": [
+          0.07567314817055135,
+          0.09816994268643958,
+          0.18051513275671677,
+          0.3329858542465733,
+          0.30602261674838793,
+          0.35054975794861853,
+          0.3678330103951213,
+          0.3223495806659864,
+          0.30948360148360754,
+          0.4215084326821277,
+          0.07974123653445743,
+          0.09673429960193489,
+          0.17573403937952553,
+          0.3546907718824904,
+          0.3384391236199661,
+          0.40416512525486814,
+          0.4097494458114911,
+          0.39862385926750626,
+          0.34305740430139864,
+          0.42241980095692633
+        ],
+        "min": [
+          0.19337546825408936,
+          -0.21049153804779053,
+          0.028447948396205902,
+          -1.0,
+          -1.0000001192092896,
+          -0.9999998211860657,
+          -1.0,
+          -0.9999995231628418,
+          -1.0000001192092896,
+          0.0,
+          0.12499918043613434,
+          -0.7061007022857666,
+          0.04709611460566521,
+          -0.9999996423721313,
+          -1.0,
+          -1.0,
+          -1.0,
+          -1.000000238418579,
+          -1.0,
+          0.0
+        ],
+        "max": [
+          0.9986844062805176,
+          0.6781377792358398,
+          1.3417878150939941,
+          0.9999530911445618,
+          0.9999873042106628,
+          1.0,
+          0.9999998211860657,
+          1.0,
+          0.9999994039535522,
+          1.0,
+          0.9906591176986694,
+          0.17973941564559937,
+          1.427522897720337,
+          1.0,
+          0.9999999403953552,
+          1.0,
+          0.9999999403953552,
+          0.9999999403953552,
+          0.9999961256980896,
+          1.0
+        ]
+      }
+    }
+  }
+}