Skip to content

[Fix] Update dataset conversion for InternData-N1 VLN-PE v0.5 dataset format #288

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
kew6688 merged 3 commits into from
Feb 6, 2026
+392 −7
Merged

[Fix] Update dataset conversion for InternData-N1 VLN-PE v0.5 dataset format #288

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
5891be6
update lerobot dataset
Feb 4, 2026
4c93a3d
add compatibility doc
Feb 4, 2026
5d247ab
remove chinese comment
Feb 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 13 additions & 0 deletions docs/compatibility.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,13 @@
# Compatibility

## v0.3.1

### InternData-N1 update to v0.5

The InternData-N1 VLN-PE trajectory training dataset has been upgraded from `v0.1` to `v0.5`. This update introduces minor structural changes in the dataset layout and updates the LeRobot-to-LMDB conversion logic to match the new `v0.5` data structure.

The training pipeline now uses the new key name:
- `instruction_text` → `task`

The updated conversion logic is **not compatible** with InternData-N1 `v0.1`.

5 changes: 3 additions & 2 deletions internnav/dataset/cma_lerobot_dataset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@

from internnav.dataset.base import BaseDataset, ObservationsDict, _block_shuffle
from internnav.model.utils.feature_extract import extract_instruction_tokens
from internnav.utils.lerobot_as_lmdb import LerobotAsLmdb
from internnav.utils.loader import LerobotAsLmdb


class CMALerobotDataset(BaseDataset):
Expand Down Expand Up @@ -38,8 +38,9 @@ def __init__(
self.camera_name = self.config.il.camera_name

self.lerobot_as_lmdb = LerobotAsLmdb(self.lerobot_features_dir)
self.lmdb_keys = self.lerobot_as_lmdb.get_all_keys()
self.lmdb_keys = self.lerobot_as_lmdb.get_all_keys(allow_scan_list=['r2r']) # r2r / r2r_aliengo / r2r_flash
self.length = len(self.lmdb_keys)
print(f"total keys in traj_data: {len(self.lmdb_keys)}")

# For CMA-CLIP
self.use_clip_encoders = False
Expand Down
7 changes: 4 additions & 3 deletions internnav/dataset/rdp_lerobot_dataset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,7 +26,7 @@
from internnav.model.basemodel.LongCLIP.model import longclip
from internnav.model.utils.feature_extract import extract_instruction_tokens
from internnav.utils.geometry_utils import get_delta, normalize_data, to_local_coords
from internnav.utils.lerobot_as_lmdb import LerobotAsLmdb
from internnav.utils.loader import LerobotAsLmdb


def _convert_image_to_rgb(image):
Expand Down Expand Up @@ -103,8 +103,9 @@ def __init__(
self.to_pil = ToPILImage()
self.image_processor = _transform(n_px=224) # copy from clip-long
self.lerobot_as_lmdb = LerobotAsLmdb(self.lerobot_features_dir)
self.lmdb_keys = self.lerobot_as_lmdb.get_all_keys()
self.lmdb_keys = self.lerobot_as_lmdb.get_all_keys(allow_scan_list=['r2r']) # r2r / r2r_aliengo / r2r_flash
self.length = len(self.lmdb_keys)
print(f"total keys in traj_data: {len(self.lmdb_keys)}")

self.start = 0
self.end = self.length
Expand Down Expand Up @@ -192,7 +193,7 @@ def _load_next(self): # noqa: C901
episodes_in_json = data_to_load['episodes_in_json']

instructions = [
episodes_in_json[ep_idx]['instruction_text'][: self.config.model.text_encoder.max_length]
episodes_in_json[ep_idx]['task'][: self.config.model.text_encoder.max_length]
for ep_idx in range(len(episodes_in_json))
]

Expand Down
Loading