github/ColossalAI
1
0
Fork 0
mirror of https://github.com/hpcaitech/ColossalAI.git synced 2025-09-03 18:19:58 +00:00
Code Issues Packages Projects Releases Wiki Activity

upgrade colossal-chat support tp_group>1, add sp for sft

Browse Source
This commit is contained in:
YeAnbang
2024-05-27 05:55:57 +00:00
parent 73e88a5553
commit 7a7e86987d
33 changed files with 7574 additions and 105 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
applications/ColossalChat/examples/data_preparation_scripts/prepare_sft_dataset.sh
View File

@@ -1,13 +1,22 @@
SAVE_DIR=""
SAVE_DIR="/home/yeanbang/data/experiment/dataset/alpaca/test/Llama-2-7b-chat-hf"
rm -rf $SAVE_DIR/cache
rm -rf $SAVE_DIR/jsonl
rm -rf $SAVE_DIR/arrow
# python prepare_dataset.py --type sft \
# --data_input_dirs /home/yeanbang/data/experiment/dataset/sft_data/test/sft-data \
# --conversation_template_config /home/yeanbang/data/ColossalAI/applications/ColossalChat/config/conversation_template/THUDM_chatglm3-6b.json \
# --tokenizer_dir "/mnt/jfs-hdd/home/data/models/ChatGlm-6B" \
# --data_cache_dir $SAVE_DIR/cache \
# --data_jsonl_output_dir $SAVE_DIR/jsonl \
# --data_arrow_output_dir $SAVE_DIR/arrow \
python prepare_dataset.py --type sft \
--data_input_dirs /PATH/TO/SFT/DATASET \
--conversation_template_config /PATH/TO/CHAT/TEMPLATE/CONFIG.json \
--tokenizer_dir "" \
--data_input_dirs /mnt/jfs-hdd/home/yeanbang/data/dataset/sft_data/alpaca/data_preprocessed/train \
--conversation_template_config /home/yeanbang/data/ColossalAI/applications/ColossalChat/config/conversation_template/llama2.json \
--tokenizer_dir "/mnt/jfs-hdd/share/models/Llama-2-7b-chat-hf" \
--data_cache_dir $SAVE_DIR/cache \
--data_jsonl_output_dir $SAVE_DIR/jsonl \
--data_arrow_output_dir $SAVE_DIR/arrow \