A new site called Are You in the Weights? appeared on Hacker News with 299 points and 161 comments. The tool at https://www.intheweights.com/ lets users query whether specific AI models have publicly listed weights.
What It Is
The site indexes model weights from major hubs and papers. Users enter a model name or arXiv ID and receive a yes/no result plus links to download locations when available.
It pulls from Hugging Face, GitHub releases, and academic repositories. No account is required for basic checks.
How to Try It
Visit the homepage and type a model identifier in the search field. Results return in under two seconds for most queries.
The interface supports batch checks via a simple CSV upload for up to 50 models at once. Export options include JSON and Markdown.
Benchmarks and Coverage
Early data shared in the thread shows coverage of roughly 12,000 model entries. Response time averages 1.8 seconds on desktop connections.
The dataset updates daily from public sources. Coverage is strongest for transformer-based LLMs released after 2022.
Pros and Cons
- Fast single-query checks without login
- Direct links to verified weight files
Free and open for non-commercial use
Limited to publicly posted weights only
No support for private or gated repositories
Sparse coverage of vision and audio models
Alternatives and Comparisons
Existing options include direct searches on Hugging Face and Papers with Code.
| Feature | Are You in the Weights? | Hugging Face Hub | Papers with Code |
|---|---|---|---|
| Single search speed | 1.8s | 3-5s | 4-6s |
| Batch upload | Yes (50 items) | No | No |
| Weight verification | Yes | Partial | No |
| License filter | No | Yes | Yes |
Who Should Use This
Researchers verifying reproducibility of recent papers benefit most. Teams maintaining internal model registries can use it for quick public availability checks.
Skip it if your work involves closed-source or enterprise-gated weights. The tool adds little value for users already fluent with Hugging Face advanced search.
Bottom line: A lightweight lookup layer that reduces time spent hunting for public model weights across scattered sources.
The project fills a narrow but recurring gap for practitioners who need fast confirmation before starting downloads or fine-tuning runs. Continued growth depends on how quickly the maintainers expand beyond current LLM-heavy coverage.
Top comments (0)