Skip to content
OpenTrain AIFor AI Companies

Preventing Zero-Shot Transfer Degradation in Continual Learning of Vision-Language Models

Published Mar 1, 2023
arXiv PDF
Researcher verdict
Starting point
Use as implementation starting point
Benchmark evidence
Thin evidence
Verify before relying
Time to first repro
A few days
Plan setup time
Risk flags
1
Review before use

Results and benchmarks

Freshness tier: hot
Preventing Zero-Shot Transfer Degradation in Continual Learning of Vision-Language Models is the primary contribution described in this paper.
Task Dataset Metric Value Source
Computer vision MTIL Accuracy 89.0 paper-derived
Computer vision MNIST Accuracy 59.4 paper-derived
Computer vision Zero-shot Flowers 71.0 paper-derived

Audit each benchmark finding before selecting an implementation path. Evidence refs map to the disclosure below.

Implementation

Historical official implementation (not recommended for new builds)

Why this implementation
Confidence: low

aimagelab/mammoth is the closest maintained adjacent implementation (Partial overlap with paper title keywords). It is not paper-verified; validate algorithm and evaluation setup against the paper before trusting reported metrics. Community adoption signal: 834 GitHub stars.

Open thunderbeee/zscl
Reproduction risks
  • Adjacent implementations are not paper-verified
  • Recommended repository is adjacent and not paper-verified.
  • Adjacent implementation match confidence is low.
  • No direct maintained implementation is currently verified.
  • Only historical official repository was found: thunderbeee/zscl.
  • No maintained paper-verified implementation met reliability thresholds.

Reproduction readiness

Time to first repro: days
Last checked: Aug 23, 2026

Major work

No dependency manifest, manual reconstruction required

  • thunderbeee/zscl has no requirements.txt, environment.yml, pyproject.toml, or Dockerfile.
  • You will need to reverse-engineer dependencies from import statements in the source code.
  • Last push was 904 days ago.
Open thunderbeee/zscl

Hardware requirements

  • Expect multi-day setup/compute for meaningful reproduction based on current guidance.

Repositories and ecosystem

Closest related implementations

These are not paper-verified. Use them as reference points when no direct implementation is available.

  • aimagelab/mammoth Adjacent · Confidence: Low · 834 stars

    Partial overlap with paper title keywords

No additional verified repositories beyond the primary recommendation.

Hugging Face artifacts

No trustworthy direct or curated related Hugging Face artifacts were found yet. Use targeted searches to quickly locate candidate models, datasets, and demos.

Search models Search datasets Search spaces

Tip: start with models, then check datasets and spaces if you need evaluation data or demos.

Research context

Tasks

Computer vision

Methods

Transformer

Domains

Computer vision, Natural Language Processing

Evaluation and human feedback data

Open this paper in HFEPX to review benchmark signals, evaluation modes, and human-feedback protocol context.

Open in HFEPX
Explore similar papers

Jump to Paper2Code search queries derived from this paper's research context.

Computer vision Transformer Natural Language Processing

Data includes links from Papers with Code ( CC-BY-SA-4.0 ).