---
license: mit
tags:
- model-collaboration
- learning-to-defer
- vision-language
---

# Learned Routing for Specialist–VLM Collaboration

Artifacts for **When Should a Vision Specialist Defer? Budgeted Collaboration with Vision–Language Models**.

[Code and reproduction instructions](https://github.com/nohi191212/ModelCollaboration) · [Latest paper](releases/20260913_lr/main.pdf) · [Supplement](releases/20260913_lr/appendix.pdf)

## Latest release

Use [releases/20260913_lr](https://huggingface.co/nohi191212/ModelCollaboration/tree/main/releases/20260913_lr).

- **80 main-table router checkpoints:** 16 model pairs × five seeds, in `full_budget_20260913_weights.tar.gz`.
- Experiment code, configurations, histories and per-example outputs, including the other objectives and scoring experiments. Only the main-table router weights are included in this release.
- Eight specialist input caches covering train, validation and test splits, with an input manifest.
- Paper sources and numerical summaries in the code repository; latest PDFs and result data here.

To reconstruct Tables 1 and 3 without training, download `paper_replay_runs.tar.gz` and `paired_outcomes.tar.gz`, then follow the [reproduction instructions](https://github.com/nohi191212/ModelCollaboration/blob/main/docs/REPRODUCE.md). The replay checks 80 main runs, 320 Table 3 runs and 19,392 curve points.

Main checkpoints use validation selection across 0–100% calls. Table 3 preserves its earlier checkpoint selection settings, detailed in the supplement.

Root-level archives are retained for earlier releases. The latest release does not add supplementary router checkpoints.

## Data and licenses

Original code is MIT licensed. Upstream model and dataset licenses remain unchanged. This release provides derived experiment features, labels and outputs; obtain original images from their providers. ConstructionSite-derived data retain CC BY-NC 4.0 terms. See [model and dataset sources](https://github.com/nohi191212/ModelCollaboration/blob/main/docs/MODELS_AND_DATA.md).