Downloads · 30 days
0
myx160/sft_eyeglasses250_pi05_ppu8c_mb48_gbs384_s20000_from100_s4000_v2_step4000
sft_eyeglasses250_pi05_ppu8c_mb48_gbs384_s20000_from100_s4000_v2_step4000 is a robotics model from myx160. Use it for the robotics task on the model card, and read the license before you ship it in a product. It is set up for openpi. The card lists the license as apache-2.0.
Complete, resumable PyTorch/RLinf OpenPI SFT checkpoint for the Piper eyeglasses task.
Downloads · 30 days
0
Access
Public
Updated Sep 16, 2026
Repo size
34.1 GB
Likes
0
Public
Click a slice to open those files.
.distcp25.6 GB · 75%
From the Hugging Face model README
Complete, resumable PyTorch/RLinf OpenPI SFT checkpoint for the Piper eyeglasses task.
Task instruction:
Pick up the eyeglasses, fold the temples, and place them on the box on the right.
myx160/eyeglasses_grasp_250episode at revision fd7ce02d49c4b150becdf657ec010faaa0e63d73 (250 episodes)full_shard; 20000-step schedule; warmup 1000; peak LR 1e-5The complete checkpoint is under global_step_4000/actor/. It includes all 8 DCP
shards and optimizer state, .metadata, model_state_dict/full_weights.pt, data.pt,
rng.pt, and SFT_CHECKPOINT_COMPLETE.json. The actor directory also contains the
exact norm stats and transform contract required by the real-robot inference pipeline.
This is a training artifact. Train loss alone is not evidence of real-robot success; select checkpoints using the frozen real-world evaluation protocol.