Skills
skills/togethercomputer/skills/together-fine-tuning/Gen Agent Trust Hub

together-fine-tuning

Pass

Audited by Gen Agent Trust Hub on May 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill contains several Python scripts in the scripts/ directory that utilize the Together AI Python SDK to perform administrative and training tasks.
  • Evidence: scripts/finetune_workflow.py, scripts/dpo_workflow.py, scripts/reasoning_finetune.py, and scripts/vlm_finetune.py all execute commands via the together library to upload files, create fine-tuning jobs, and manage dedicated model endpoints.
  • [EXTERNAL_DOWNLOADS]: The skill includes functionality to fetch data from external URLs for dataset preparation.
  • Evidence: scripts/vlm_finetune.py implements a url_to_base64 function using the requests library to download images from URLs (provided via the --sample-image-url argument) and convert them to base64 for multimodal training. This is a standard utility for vision-language model (VLM) preparation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 2, 2026, 01:17 PM
Security Audit — agent-trust-hub — together-fine-tuning