dist_train.sh

#!/bin/bash
now=$(date +"%Y%m%d_%H%M%S")

epoch=120
bs=1
gpus=1
lr=0.000005
encoder=vitl
dataset=pbr  # Changed default to pbr
img_size=512
min_depth=0.0
max_depth=1.0  # Changed to 1.0 for normalized depth
pretrained_from=../checkpoints/depth_anything_v2_${encoder}.pth
save_path=exp/pbr  # Changed to pbr

mkdir -p $save_path

python3 -m torch.distributed.launch \
    --nproc_per_node=$gpus \
    --nnodes 1 \
    --node_rank=0 \
    --master_addr=localhost \
    --master_port=20596 \
    train.py --epoch $epoch --encoder $encoder --bs $bs --lr $lr --save-path $save_path --dataset $dataset \
    --img-size $img_size --min-depth $min_depth --max-depth $max_depth --pretrained-from $pretrained_from \
    --port 20596 2>&1 | tee -a $save_path/$now.log