hailin
/
jd_train
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

.

This commit is contained in:
hailin 2025-08-31 12:40:17 +08:00
parent 7184cb67f3
commit 56a250bc66
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
mm-zero3.sh
View File

@ -7,9 +7,9 @@ export PYTORCH_CUDA_ALLOC_CONF="max_split_size_mb:128,expandable_segments:True,g
deepspeed --hostfile hostfile \
--num_nodes 6 --num_gpus 4 \
/home/test/jd_train/train_sft_ds.py \
--model_name_or_path /home/test/Qwen3-14B \
--model_name_or_path /home/test/Qwen3-8B \
--data_glob "/home/test/datasets/my_corpus/train.jsonl" \
--output_dir /home/test/checkpoints/q3-14b-ds4 \
--output_dir /home/test/checkpoints/q3-8b-ds4 \
--seq_len 512 \
--per_device_train_batch_size 1 \
--gradient_accumulation_steps 1 \