scripts/train_drseg.sh

#!/usr/bin/env bash
set -euo pipefail

GPUS="0"
BASE_MODEL="checkpoints/llava-v1.6-vicuna-7b"
DATA="data/DRSeg"
CLIP="checkpoints/clip-vit-large-patch14"
EXP="pixdlm_train"
PORT="${PORT:-29511}"
EPOCHS="${EPOCHS:-10}"
STEPS_PER_EPOCH="${STEPS_PER_EPOCH:-200}"
PRECISION="${PRECISION:-bf16}"

while [[ $# -gt 0 ]]; do
  case "$1" in
    --gpus) GPUS="$2"; shift 2 ;;
    --base-model) BASE_MODEL="$2"; shift 2 ;;
    --data) DATA="$2"; shift 2 ;;
    --clip) CLIP="$2"; shift 2 ;;
    --exp) EXP="$2"; shift 2 ;;
    --epochs) EPOCHS="$2"; shift 2 ;;
    --steps-per-epoch) STEPS_PER_EPOCH="$2"; shift 2 ;;
    --port) PORT="$2"; shift 2 ;;
    --precision) PRECISION="$2"; shift 2 ;;
    *) echo "Unknown argument: $1" >&2; exit 2 ;;
  esac
done

ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
cd "$ROOT"

export PYTHONPATH="$ROOT:${PYTHONPATH:-}"
export TOKENIZERS_PARALLELISM=false
export TRANSFORMERS_VERBOSITY=error

mkdir -p "logs/$EXP"

deepspeed --master_port="$PORT" --include="localhost:$GPUS" train_ds.py \
  --epochs="$EPOCHS" \
  --steps_per_epoch="$STEPS_PER_EPOCH" \
  --version="$BASE_MODEL" \
  --dataset_dir="$DATA" \
  --dataset="custom_seg" \
  --sample_rates="1" \
  --exp_name="$EXP" \
  --log_base_dir="$ROOT/logs" \
  --val_dataset="custom_seg|val" \
  --train_mask_decoder \
  --Three_Level_Multi_Scale_Decoder \
  --vision-tower="$CLIP" \
  --seg_token_num=3 \
  --num_classes_per_question=3 \
  --batch_size=1 \
  --grad_accumulation_steps=1 \
  --val_batch_size=1 \
  --preprocessor_config="$ROOT/configs/preprocessor_448.json" \
  --resize_vision_tower \
  --resize_vision_tower_size=448 \
  --vision_tower_for_mask \
  --use_expand_question_list \
  --image_feature_scale_num=3 \
  --conv_type="llava_v1" \
  --is_multipath_encoder \
  --precision="$PRECISION" 2>&1 | tee "logs/$EXP/train.log"