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

.

This commit is contained in:
hailin 2025-09-01 20:41:39 +08:00
parent b8fb783871
commit 123a49db1b
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
mm-zero3.sh
View File

@ -4,7 +4,9 @@ export TORCH_EXTENSIONS_DIR=/tmp/$USER/torch_ext
export PYTORCH_CUDA_ALLOC_CONF="max_split_size_mb:128,expandable_segments:True,garbage_collection_threshold:0.9" export PYTORCH_CUDA_ALLOC_CONF="max_split_size_mb:128,expandable_segments:True,garbage_collection_threshold:0.9"
export PYTORCH_CUDA_ALLOC_CONF="max_split_size_mb:64" export OMP_NUM_THREADS=8
export MKL_NUM_THREADS=8
export OPENBLAS_NUM_THREADS=8
deepspeed --hostfile hostfile \ deepspeed --hostfile hostfile \
--num_nodes 6 --num_gpus 4 \ --num_nodes 6 --num_gpus 4 \