Myungkyu/rldx_1_robodojo_preset_b64_60k

rldx1robodojopresetb64_60k

Low-level policy for RoboDojo long-horizon (8 real-robot bimanual tabletop tasks, 100 demonstrations each), fine-tuned from RLWRLD/RLDX-1-PT on `Myungkyu/RoboDojo-preset-gemini` — demonstrations with dense subtask labels from the subtask preset.

• —Architecture: RLDX-1-PT (video length 4, three live camera views, no keyframe slot)
• —Optimizer batch 64, 60000 steps, final checkpoint (the run was launched as batch 128 = 64 x gradient accumulation 2; the DeepSpeed ZeRO-2 build applied the accumulation without the intended scaling, so the effective optimizer batch was 64 — verified from the gradient norms on 2026-09-11; renamed from rldx_1_robodojo_preset_b128_60k on 2026-09-14)
• —Inputs: head + left/right wrist images, proprioception, the current subtask text (no keyframe input)

Configs reference the base backbone / tokenizer by hub id or by the training site's local path — point them at your local copies before loading.