Pretraining effects

We use libero_90 as the data source for pretraining, and then test the pretrained agent’s LLDM performance on libero_10.

Pretrain on LIBERO-90

Replace GPU_ID with the cuda device ID. The trained agent will be saved into the ./experiments_pretrained/ folder.

export CUDA_VISIBLE_DEVICES=GPU_ID && export MUJOCO_EGL_DEVICE_ID=GPU_ID && python lifelong/main.py seed=SEED benchmark_name=BENCHMARK policy=POLICY lifelong=multitask pretrain=true train.num_workers=8

Finetune on LIBERO-10

Replace the CKPT_PATH with the path to the saved checkpoint of the pretrained agent.

export CUDA_VISIBLE_DEVICES=GPU_ID && export MUJOCO_EGL_DEVICE_ID=GPU_ID && python lifelong/main.py seed=SEED benchmark_name=BENCHMARK policy=POLICY lifelong=ALGO pretrain_model_path=CKPT_PATH