arkojit1/pi05_mj_stack_red_yellow_20k

π₀.₅ fine-tuned on arkojit1/mj_stack_red_yellow — final, step 20,000

lerobot/pi05_base fine-tuned with LeRobot (--policy.type=pi05) on arkojit1/mj_stack_red_yellow at revision d64e37c (the 2026-10-16 upload): 100 MuJoCo Franka block-stacking episodes, 8 language tasks (red-on-yellow and yellow-on-red, four phrasings each), two 224×224 cameras (front, wrist; the third π₀.₅ image slot is padded and masked, empty_cameras=1), 4-D state [x, y, z, gripper] (world-frame end-effector position + gripper opening), 4-D action [dx, dy, dz, gripper] (action.ee.delta_xyz + action.gripper.target).

Only data/chunk-000/file-000.parquet of that revision belongs to it. The other parquet files in the repo are left over from an earlier upload and were removed before training.

This is the final checkpoint, step 20,000 (~66 epochs). It is not the run's lowest-eval checkpoint: eval reached 0.0347 at step 12,400 (~41 epochs), published as arkojit1/pi05_mj_stack_red_yellow_12.4k. Eval was flat at ~0.038–0.040 from ~16 epochs and drifted up slightly over the last quarter as train loss kept falling. That may be mild overfitting, but the drift is within the point-to-point eval noise.

Trainable	action expert only (train_expert_only=true; SigLIP + Gemma-2B frozen)
Hardware / batch	8 × MI300X, global batch 128 (16 per GPU), torch.compile (mode default)
Steps	20,000 of 20,000 (304 steps per epoch)
Optimiser	AdamW, LR 2.5e-5 peak, betas (0.9, 0.95), wd 0.01, grad clip 1.0
Schedule	cosine with warmup, pinned to 20,000 steps (warmup 666), floor 2.5e-6; bf16
Normalisation	quantile (q01–q99) for state and action
Augmentation	LeRobot default image transforms on train frames
chunk_size / n_action_steps	50 / 50
Eval split	last 10% of each task's episodes held out (38,932 training frames)
Eval loss	0.0418; run best 0.0347 @12,400 at this step (flow-matching loss on held-out episodes)
Train loss	0.014

Eval loss is a training-objective number on held-out simulator episodes, not a task success rate.

from lerobot.policies.pi05.modeling_pi05 import PI05Policy
policy = PI05Policy.from_pretrained("arkojit1/pi05_mj_stack_red_yellow_20k")