#!/bin/bash
set -e

PROJECT_ROOT="$(cd "$(dirname "$0")/.." && pwd)"
cd "$PROJECT_ROOT"
ENV_ROOT="${PROJECT_ROOT}/envs/streampetr"
DEEPSPEED_BIN="${ENV_ROOT}/bin/deepspeed"

if [ ! -x "$DEEPSPEED_BIN" ]; then
    echo "ERROR: deepspeed not found at ${DEEPSPEED_BIN}" >&2
    exit 1
fi

export PATH="${ENV_ROOT}/bin:${PATH}"
export LD_LIBRARY_PATH="${ENV_ROOT}/lib:${LD_LIBRARY_PATH}"

OUTPUT_DIR="work_dirs/atlas_4task_coordfix"
mkdir -p "$OUTPUT_DIR"

EXTRA_ARGS=""
if [ -n "$RESUME_CKPT" ]; then
    EXTRA_ARGS="--resume $RESUME_CKPT"
fi

setsid nohup "$DEEPSPEED_BIN" --num_gpus 4 train_atlas.py \
    --llm_model pretrained/vicuna-7b-v1.5 \
    --data_json data/atlas_nuscenes_train.json,data/openlane_subsetB_lane_train_4pt_aligned.json,data/atlas_planning_train_uniad_command.json,data/atlas_caption_train_canonical.json \
    --data_root /home/guoyuanbo/autodl-tmp/data/nuscenes \
    --visual_token_mode online \
    --task_balance_mode none \
    --planning_table3_mode atlas_high_level \
    --streampetr_config configs/streampetr_atlas_aligned.py \
    --streampetr_ckpt pretrained/streampetr/streampetr_eva02_ep24.pth \
    --topomlp_config configs/topomlp_atlas_aligned.py \
    --topomlp_ckpt work_dirs/topomlp_atlas_aligned/epoch_24.pth \
    --deepspeed configs/ds_zero2.json \
    --output_dir "$OUTPUT_DIR" \
    --epochs 10 \
    --lr 2e-5 \
    --weight_decay 1e-4 \
    --batch_size 1 \
    --gradient_accumulation_steps 2 \
    --warmup_ratio 0.03 \
    --max_grad_norm 1.0 \
    --log_steps 100 \
    --save_epochs 1 \
    --keep_last_n_ckpts 0 \
    --seed 42 \
    --num_workers 4 \
    $EXTRA_ARGS > "$OUTPUT_DIR/nohup.out" 2>&1 &

echo "Training launched in background (setsid + nohup)."
echo "PID: $!"
echo "Log: $OUTPUT_DIR/nohup.out"
echo ""
echo "Monitor: tail -f $OUTPUT_DIR/nohup.out"
echo "Check GPU: nvidia-smi"