WRM_adaptive-all125

Complete checkpoint series for WRM_adaptive trained on the local WRM_grasp_cylinder_different_sizes_lero_tightness LeRobot dataset.

• GPULab job: 920957a7-7b2a-4e3f-89c6-ab0e5f151af8
• Training steps: 10000 through 100000, every 10000 steps
• Checkpoints: ten numbered milestones plus final EMA-enabled last.pt
• SHA-256: faed996bdc4efc1f9a432e59177c9d5dc5229939ea15bd7faaf376ef5ef91214
• Source configuration: resolved_config.yaml
• Metrics: metrics.jsonl

The repository is private and intended for AIRO-Doffy real-robot evaluation.