Downloads · 30 days
0
pi05_obs_only_firstandlast3_injection_allenv_2ep
rshift8/pi05_obs_only_firstandlast3_injection_allenv_2ep
pi05_obs_only_firstandlast3_injection_allenv_2ep is a machine learning model from rshift8. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
2-epoch JAX/orbax fine-tune of RoboPRO obstacle-attention π₀.₅ (pi05obsonlyfirstandlast3injectionallenv), obstacle-only attention (target and destination attention off).
Access
Public
Updated Sep 2, 2026
Repo size
499 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
Other499 GB · 100%
At a glance
- Access
- Public
- Created
- Aug 31, 2026
- Updated
- Sep 2, 2026
- SHA
- fb51c524
Model card
From the Hugging Face model README
pi05_obs_only_firstandlast3_injection_allenv_2ep
2-epoch JAX/orbax fine-tune of RoboPRO obstacle-attention π₀.₅ (pi05_obs_only_firstandlast3_injection_allenv), obstacle-only attention (target and destination attention off).
- 230934 steps (2 × 115467), batch 32, 2-GPU data parallel
- Saves every 25k steps plus epoch ends (115467, 230934)
- Init: mzxuan/robopro_jax_30000
Each step directory contains everything needed to resume training or run eval:
params/— model weightstrain_state/— optimizer / train stateassets/—norm_stats.json_CHECKPOINT_METADATA— orbax item handlers (params, train_state, assets)train_config.py—TrainConfigsnippet to paste intosrc/openpi/training/config.py(same pattern as mzxuan/robopro_jax_30000)
- Created
- Aug 31, 2026
- Updated
- Sep 2, 2026