File size: 2,827 Bytes
c99d198 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 | #!/bin/bash
# Run 4 separate RL experiments (one per pretrained visual encoder).
# Each run trains a single task to 2M steps with demo subgoals only (vgen disabled).
#
# Experiments:
# 1. Lift + mixed 75/25 encoder
# 2. Lift + simonly 35/15 encoder
# 3. Stack + mixed 75/25 encoder
# 4. Stack + simonly 35/15 encoder
#
# Usage:
# ./train_all_encoder_ablations.sh
# START_FROM=2 ./train_all_encoder_ablations.sh # skip first 2 runs
# RUN_ONLY=3 ./train_all_encoder_ablations.sh # run only the 3rd experiment
set -euo pipefail
cd "$(dirname "$0")"
SEED="${SEED:-0}"
DEVICE="${DEVICE:-cuda:0}"
START_FROM="${START_FROM:-1}"
RUN_ONLY="${RUN_ONLY:-}"
PRETRAIN_ROOT="${PRETRAIN_ROOT:-/home/lei/Documents/tong/irl4idm/multi-task-tcc-robosuite/experiments/pretrain_runs}"
run_single_experiment() {
local idx="$1"
local env_name="$2"
local exp_name="$3"
local encoder_path="$4"
local resume_flag=()
local ckpt_dir="rl_runs/${exp_name}/${SEED}/checkpoints"
if [[ -n "${RUN_ONLY}" && "${idx}" != "${RUN_ONLY}" ]]; then
return 0
fi
if [[ -z "${RUN_ONLY}" && "${idx}" -lt "${START_FROM}" ]]; then
echo "Skipping experiment ${idx} (${exp_name}) due to START_FROM=${START_FROM}"
return 0
fi
if compgen -G "${ckpt_dir}/*.ckpt" > /dev/null; then
resume_flag=(--resume)
echo "Found checkpoint in ${ckpt_dir}; resuming."
fi
echo ""
echo "=========================================="
echo " Experiment ${idx}/4: ${env_name}"
echo " RL run: rl_runs/${exp_name}/${SEED}"
echo " Encoder: ${encoder_path}"
echo " TensorBoard: tensorboard --logdir rl_runs/${exp_name}/${SEED}/tb"
echo "=========================================="
echo ""
python train_policy.py \
--experiment_name="${exp_name}" \
--env_name="${env_name}" \
--config="configs/robosuite/rl/env_reward.py:${env_name}" \
--config.reward_wrapper.pretrained_path="${encoder_path}" \
--config.reward_wrapper.type=distance_to_goal \
--config.threshold_for_vgen=1.1 \
--seed="${SEED}" \
--device="${DEVICE}" \
"${resume_flag[@]}"
}
run_single_experiment 1 Lift lift_rl_mixed_75 \
"${PRETRAIN_ROOT}/dataset=mimicgen_algo=xirl_task=lift"
run_single_experiment 2 Lift lift_rl_simonly_35_15 \
"${PRETRAIN_ROOT}/dataset=mimicgen_algo=xirl_task=lift_simonly_35_15"
run_single_experiment 3 Stack stack_rl_mixed_75 \
"${PRETRAIN_ROOT}/dataset=mimicgen_algo=xirl_task=stack"
run_single_experiment 4 Stack stack_rl_simonly_35_15 \
"${PRETRAIN_ROOT}/dataset=mimicgen_algo=xirl_task=stack_simonly_35_15"
echo ""
echo "Done: all requested encoder ablation runs complete."
echo "Results:"
echo " rl_runs/lift_rl_mixed_75/${SEED}"
echo " rl_runs/lift_rl_simonly_35_15/${SEED}"
echo " rl_runs/stack_rl_mixed_75/${SEED}"
echo " rl_runs/stack_rl_simonly_35_15/${SEED}"
|