Archives/petals
2
0
Fork 0
mirror of https://github.com/bigscience-workshop/petals synced 2024-10-31 09:20:41 +00:00
Code Issues Packages Projects Releases Wiki Activity

set throughput=1 to fix 0 throughput problem

Browse Source
This commit is contained in:
dbaranchuk 2022-07-31 06:13:59 +03:00
parent 123e57a5fc
commit 8b3d66167f
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
cli/demo_deploy_server.sh
View File

@ -28,4 +28,4 @@ NUM_BLOCKS="3" # one converted block consumes ~3.5Gb
export OMP_NUM_THREADS="16" # just in case
CUDA_VISIBLE_DEVICES=${GPU_ID} python -m cli.run_server --converted_model_name_or_path ${MODEL_NAME} --torch_dtype float16 --initial_peer ${INITIAL_PEER} \
--compression BLOCKWISE_8BIT --identity_path ${SERVER_ID_PATH} --host_maddrs ${HOST_MADDR} \
--num_blocks ${NUM_BLOCKS} --load_in_8bit
--num_blocks ${NUM_BLOCKS} --load_in_8bit --throughput 1