Logo
Explore Help
Register Sign In
Mirrors/modelscope
1
0
Fork 0
You've already forked modelscope
mirror of https://github.com/modelscope/modelscope.git synced 2025-12-23 19:49:24 +01:00
Code Issues Packages Projects Releases Wiki Activity
Files
f856c04785fe95a1eee1708b51bc09cc667bdcd3
modelscope/examples/pytorch/llama/run_train_llama.sh

10 lines
282 B
Bash
Raw Normal View History

[to #48855014] llama finetune + deepspeed 1. llama base finetune:使用trainer从llama finetune至alpaca,效果确认并提供example 2. deepspeed通用性完善:mpu解耦;deepspeed的训练信息支持从ms log透出(目前ms打印的log有误);支持从modelscope configuration.json 进行 deepspeed config 配置;deepspeed optimizer和lr_scheduler 初始化支持;解决deepspeed和ddp同时使用报错;解决保存ckpt时报错 Link: https://code.alibaba-inc.com/Ali-MaaS/MaaS-lib/codereview/12651323
2023-06-14 10:00:23 +08:00
DATA_PARALLEL_SIZE=4
export PYTHONPATH=$PYTHONPATH:./
torchrun --nproc_per_node $DATA_PARALLEL_SIZE examples/pytorch/llama/finetune_llama.py \
--work_dir './tmp' \
--model 'skyline2006/llama-7b' \
--deepspeed 'default_offload_opt_param.json' \
--eval_interval 100
Reference in New Issue Copy Permalink
Powered by Gitea Version: 1.25.2 Page: 523ms Template: 14ms
English
Bahasa Indonesia Deutsch English Español Français Gaeilge Italiano Latviešu Magyar nyelv Nederlands Polski Português de Portugal Português do Brasil Suomi Svenska Türkçe Čeština Ελληνικά Български Русский Українська فارسی മലയാളം 日本語 简体中文 繁體中文(台灣) 繁體中文(香港) 한국어
Licenses API