eval_graph.sh

export 

# --- 1. Get model and dataset name from command line ---
BASE_MODEL=$1 # qwen_3b
DATASET=$2 # cora
MP=$3 # mp layer
C_RATIO=0.1 # compression ratio

###### setting for model ######
echo "Eval on model: $BASE_MODEL with MP layer = $MP"

echo "Eval on dataset: $DATASET"

OUTPUT_DIR="./${BASE_MODEL}/"

###### setting for model ######
BASE_MODEL_DIR="./output/finetune/${BASE_MODEL}/${DATASET}/model/"
###### setting for dataset ######

case "$DATASET" in
  cora)
    DATASET="academic_cora_v1"
    ;;
  *)
    echo "Error: Unknown dataset '$DATASET'!"
    ;;
esac
###### setting for dataset ######

python -m eval_graph_model \
--data_root /eval_data \
--output_dir $OUTPUT_DIR \
--model_name_or_path $BASE_MODEL_DIR \
--max_length 15500 \
--enable_beacon True \
--local_window 3500 \
--memory_stride 1024 \
--mp_layer_num $MP \
--c_ratio $C_RATIO \
--beacon_param q k v o \
--eval_data $DATASET