File size: 1,630 Bytes
21ee636 |
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 |
#DIR="/data/wet-data/output/toxic_filtered_without_bloom_new/2024-10/"
#TRAIN_DIR=$DIR
#OUTPUT_TRAIN_DIR="/data/hineng/data_multi/"
#DATASET_NAME="multi"
#FINAL_DIR="/data/datasets/$DATASET_NAME/"
#mkdir -p $FINAL_DIR $OUTPUT_TRAIN_DIR
#langs=($(ls -1 /data/wet-data/output/toxic_filtered_without_bloom_new/2024-10/ | awk '{split($0, a,"_"); print a[1]}' | uniq))
#for lang in ${langs[@]}; do
# files=($(ls $TRAIN_DIR/$lang*))
# OUTPUT_LANG_DIR=$OUTPUT_TRAIN_DIR/$lang
# FINAL_LANG_DIR=$FINAL_DIR/$lang
# mkdir -p $OUTPUT_LANG_DIR $FINAL_LANG_DIR
# for f in ${files[@]}; do
# new_name=$(echo $f | sed "s/\//_/g")
# cmd="unzstd $f --stdout > $OUTPUT_LANG_DIR/$(basename $f).jsonl"
# echo $cmd
# eval $cmd
# done;
#final_cmd="cat $OUTPUT_LANG_DIR/*.jsonl > $FINAL_LANG_DIR/final_en.jsonl"
#echo $final_cmd
#eval $final_cmd
#done;
FINAL_DIR=/data/datasets/hindi_english_arxiv_bengali/
TOKENIZER=google/gemma-7b
TOKENIZER_TYPE=HuggingFaceTokenizer
#TOKENIZER=GPT2BPETokenizer
VOCAB_FILE=
MERGES_FILE=
ip_address=$(ifconfig | grep "inet " | grep -Fv "127.0.0.1" | grep -Fv "172.17.0.1" | awk '{print $2}')
mkdir -p $FINAL_DIR/tokenizer/
filename=$(cat $FINAL_DIR/splitfile | grep $ip_address | awk '{print $1}')
cd /Model-References/PyTorch/nlp/DeepSpeedExamples/Megatron-DeepSpeed/
python3 tools/preprocess_data.py \
--input $FINAL_DIR/$filename \
--output-prefix $FINAL_DIR/tokenizer/output_$ip_adress \
--tokenizer_model_file $TOKENIZER \
--dataset-impl mmap --tokenizer-type $TOKENIZER_TYPE\
--append-eod --workers 8 #--chunk-size 50
#--vocab-file $VOCAB_FILE \
# --merge-file $MERGES_FILE \
|