torchforge-rl-training
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references several official repositories and domains belonging to the PyTorch/Meta ecosystem, including
github.com/meta-pytorch,github.com/pytorch, andmeta-pytorch.org. These are well-known, trusted sources for deep learning infrastructure. - [COMMAND_EXECUTION]: The skill instructions include standard development commands for setting up a conda environment, running installation scripts (
./scripts/install.sh), and launching training applications (python -m apps.grpo.main). These are typical for RL research workflows. - [DYNAMIC_EXECUTION]: The skill uses Python's
python -c "import ..."pattern to verify installation. This is a standard practice for verifying the presence of dependencies liketorch,forge, andvllmafter installation. - [DATA_EXPOSURE]: The skill mentions weight synchronization and monitoring via Weights & Biases (W&B). It correctly advises users to monitor their own dashboards and does not attempt to exfiltrate data to unauthorized third parties.
Audit Metadata