EO-1 fine-tuned checkpoints (intermediate, research use only)
Intermediate checkpoints of runs that were stopped early. They are not good policies.
They need a LeRobot plugin policy type (eo1_phys) and simulated tasks that are not public.
Base model: lerobot/eo1-base, built on Qwen2.5-VL-3B-Instruct.
License: Qwen RESEARCH LICENSE AGREEMENT (non-commercial; see LICENSE and NOTICE). The weights were modified by fine-tuning.
Each checkpoint folder holds pretrained_model/ (weights, processors, train_config.json) and training_state/ (optimizer, scheduler, RNG).