ai model report
Is meta-llama/Llama-Guard-3-1B safe to load?
CLEAN risk. 3 of 36 tracked dependencies carry advisories. Driven by No version-matched advisories and no malicious packages..
not scoredno advisory applies to a version this model pinsCLEAN
LOWShips safetensors plus 1 pickle checkpoint(s) for other tooling (original/consolidated.00.pth); from_pretrained never loads these, even with use_safetensors=False, but code that torch.load()s them directly (a reference implementation, a training script) executes pickle.
2 more findings on this model.
- Author
- meta-llama
- Task
- text-generation
- Loader
- transformers
- License
- llama3.2
- Downloads
- 34,784
- Remote code
- not required
- Access
- gated on HuggingFace
7 direct, 4 with findings, 30 reachable
| Package | Evidence | Advisories | Worst |
|---|---|---|---|
| transformers | direct, runtime | 24 | none pinned |
| torch | direct, inferred | 12 | none pinned |
Create a free accountto see all 36 dependencies, which versions are pinned, and what to change.
Risk computed 2026-10-11 at 00:54 UTC. Dependencies come from the model's requirements, its declared loader and the code it ships; advisories from NVD, GHSA and OSV.