Weight provenance: is lingbot-map initialized from / distilled from VGGT's checkpoint, or trained from scratch?

Open Beginner friendly
#107 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
2/5
Estimated time
1-3 hours
Newbie friendliness
68/100
Issue type
Documentation
Clarity
Mostly clear
Activity status
Active

Research direction

Read the README, model card, and LICENSE.txt first, then inspect any released-checkpoint and training references for provenance details. Done means the maintainers' answers are captured in a concise README or model-card note covering checkpoint initialization, training datasets, and whether Apache-2.0 applies to weights or only code.

Written by the indexing model from the issue text.

Description

Thank you for the great work and for releasing the code under Apache-2.0.

The README/model card lists VGGT and DINOv2 under 'builds upon several excellent open-source projects,' but it isn't clear from the README, model card, or LICENSE.txt whether this refers to:

(a) architectural inspiration only (LingBot-Map trained from scratch, unrelated model weights), or
(b) the released lingbot-map / lingbot-map-stage1 checkpoints being initialized from, fine-tuned from, or distilled using Meta's VGGT pretrained weights (which are released under a non-commercial research license) or DINOv2 weights, or
(c) any other dataset (ScanNet, Habitat, BlendedMVS, Co3D, etc.) that carries research-only/non-commercial terms.

Could you confirm:

  1. Were the released checkpoints initialized from, fine-tuned from, or distilled from VGGT's or DINOv2's pretrained weights, or trained entirely from scratch?
  2. What dataset(s) were used to train the released checkpoints, and do any of them carry non-commercial or research-only license terms?
  3. Does the Apache-2.0 LICENSE.txt cover the released model weights as well as the code, or only the code?

This matters for evaluating commercial use of the released checkpoints. Happy to help update the README/model card with a short provenance note if useful. Thanks!

Dominant language
Python
Stars
17.1k
Forks
1.9k
PR merge metrics
No merged PRs in 30d

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from Robbyant/lingbot-map

All issues in Robbyant/lingbot-map

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.