pi05 LoRA finetune โ€” Fruit Swap

LoRA-finetuned lerobot/pi05_base on the FruitSwapEasy task. Run: 19.55.32_train_pi05_fruit_swap_lora_70_20_10_fruit_swap_pi05 (split 70/20/10).

Checkpoint groups

Top 5 (mixed sort: phase1 desc, val asc):

Rank Epoch phase1 val filename
1 44 0.30 0.0434 epoch=0044-succ=0.00-phase1=0.30-val=0.0434.ckpt
2 28 0.20 0.0386 epoch=0028-succ=0.00-phase1=0.20-val=0.0386.ckpt
3 40 0.20 0.0388 epoch=0040-succ=0.00-phase1=0.20-val=0.0388.ckpt
4 38 0.20 0.0390 epoch=0038-succ=0.00-phase1=0.20-val=0.0390.ckpt
5 48 0.20 0.0423 epoch=0048-succ=0.00-phase1=0.20-val=0.0423.ckpt

Last 3 (by epoch among available ckpts):

Epoch phase1 val
48 0.20 0.0423
46 0.10 0.0441
44 0.30 0.0434

succ (eval rollout success) was 0.00 for all available ckpts at upload time; phase1 is the phase-1 sub-score, val is validation loss.

Note: epoch 22 was present locally but not in top5 or last3, so not included.

Downloads last month

-

Downloads are not tracked for this model. How to track
Video Preview
loading