eagle3-new-model
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides configuration guidance for the EAGLE3 offline pipeline, authored by NVIDIA. It describes how to create YAML configurations for model data synthesis and hidden-state dumping.\n- [COMMAND_EXECUTION]: The documentation references internal scripts (e.g.,
common/eagle3/dump_offline_data_vllm.sh) and standard machine learning deployment flags such as--trust-remote-code. These are legitimate and expected components of the EAGLE3 workflow for handling custom model architectures.
Audit Metadata