pi05_full_runpod

A pi0.5 checkpoint fine-tuned on the lerobot/libero_plus dataset.

This is the second stage of a two-stage run. It starts from siruku6/pi05_stage1_24000 (itself fine-tuned from lerobot/pi05_libero_base, revision a217bfd3b14673cf2ce597e69997ab21866438dd) and unfreezes every parameter, so all 4.14B parameters — vision encoder and VLM included — were updated here.

Training

Steps	3,000 (on top of stage 1's 24,000)
Batch size	64
Seed	42
Trainable parameters	all 4.14B (train_expert_only=false, freeze_vision_encoder=false)
Optimizer	AdamW, weight decay 0.01, betas (0.9, 0.95), grad clip norm 1.0
LR schedule	cosine decay with warmup — 1,000 warmup steps, peak 5e-6, decay to 2.5e-6 over 30,000
Gradient checkpointing	enabled
Action chunk	chunk_size=50, n_action_steps=10
Dataset	lerobot/libero_plus

The full configuration as actually used is in train_config.json in this repository.

Modifications to the base model

model.safetensors is modified relative to the base checkpoint: it is the result of gradient updates performed by the author on the LIBERO-plus dataset. No other file carried over from the base model is modified.

License and redistribution

This is a Model Derivative of Gemma (via PaliGemma inside pi0.5) and is therefore distributed under, and subject to, the Gemma Terms of Use.

Gemma is provided under and subject to the Gemma Terms of Use found at ai.google.dev/gemma/terms

By using or further distributing these weights (or any derivative of them) you agree to:

• the Gemma Terms of Use, a copy of which governs this repository and must be passed on to any recipient you redistribute to; and
• the Gemma Prohibited Use Policy, whose use restrictions apply to this derivative and to every derivative made from it.

Reference implementation of pi0.5: openpi (Apache License 2.0). Training data: LIBERO (Lifelong-Robot-Learning/LIBERO) — code MIT, datasets CC BY 4.0.

Provided as-is, without warranty of any kind.