domains / families / science / tao-finetune-cosmos-reason

tao-finetune-cosmos-reason

separate

Shared Cosmos3 frontend that explicitly routes Cosmos Framework and Cosmos-RL, validates runtime model/video-dataset/SLURM inputs, consumes an SQSH or packaged backend image, optionally plans explicit clean source builds, prepares checkpoints, validates the first update in-process, and returns token-weighted losses and task-aware accuracy.

When to use

• Cosmos Framework training
• Cosmos-RL training
• Cosmos3-Nano fine-tuning
• Cosmos3-Edge fine-tuning
• SLURM job planning
• Docker container execution
• Checkpoint preparation
• Source build planning

When not to use

• Inferring paths from history
• Inferring model type from config.json
• Comparing SQSH filename with image tag
• Requesting source provenance/SHA
• Inferring build from runtime selection

Skill page

Source