pi05_assembly_bench_2_full_vlm_6k_cont6k

π0.5 (full VLM) continued SFT on lukasskellijs/assembly_bench_2.

• Base / prior: lukasskellijs/pi05_assembly_bench_2_full_vlm_6k
• This run: +6k steps continued training (final checkpoint)

Load with LeRobot PI05Policy.from_pretrained.