neuralset 0.0.2 fails to import against exca 0.5.26 — clean install broken on Colab (Python 3.12)

[Karre89]

Summary
A fresh, first-cell install of tribev2 in a clean Google Colab runtime installs neuralset 0.0.2 + exca 0.5.26, and from tribev2.demo_utils import TribeModel fails at import time inside neuralset. The exca dependency appears unpinned, so pip resolves an incompatible exca.

Environment

• Google Colab, fresh runtime, GPU
• Python 3.12
• Install (first cell, nothing before it): pip install "tribev2 @ git+https://github.com/facebookresearch/tribev2.git"
• Resolved versions: tribev2 0.1.0, neuralset 0.0.2, exca 0.5.26

Error 1 — on import:
AttributeError: module 'exca.steps.base' has no attribute 'NoValue'
neuralset/events/study.py, in StudyLoader():
def run(self, value: tp.Any = exca.steps.base.NoValue()) -> tp.Any:

Error 2 — after shimming a placeholder NoValue, a second break appears:
AttributeError: partially initialized module 'neuralset' has no attribute 'segments' (most likely due to a circular import)
neuralset/dataloader.py, in SegmentsMixin():
segments: tp.Sequence[ns.segments.Segment]

The core contradiction: neuralset 0.0.2 declares exca>=0.5.20, but its code calls exca.steps.base.NoValue, which is absent in current exca (0.5.26) AND in the pre-0.5 line (0.4.8). No exca version satisfies both the requirement and the source — neuralset appears to reference a removed/renamed exca API.

Question: what exca version is neuralset 0.0.2 built against? Could the exca dependency be pinned to a known-good version? That would unblock this immediately.

Thanks for releasing TRIBE v2 — keen to use it once the install resolves.

[deepika-a3i]

Hi I am also facing the same issue. Any resolutions for this?

`AttributeError Traceback (most recent call last)
/tmp/ipykernel_1818/3387571854.py in <cell line: 0>()
----> 1 from tribev2.demo_utils import TribeModel, download_file
2 from tribev2.plotting import PlotBrain
3 from pathlib import Path
4
5 CACHE_FOLDER = Path("./cache")

7 frames
/usr/local/lib/python3.12/dist-packages/neuralset/events/study.py in StudyLoader()
253 return chain._run(*args)
254
--> 255 def run(self, value: tp.Any = exca.steps.base.NoValue()) -> tp.Any:
256 """Execute -- bypass Step-level caching when transforms exist
257 (the internal chain handles it)."""

AttributeError: module 'exca.steps.base' has no attribute 'NoValue'`