Skip to content

GitLab

Commit fe2d1581 authored by Sergey Edunov's avatar Sergey Edunov
Browse files

Fix docs

parent 5d944b06
Hide whitespace changes
Inline Side-by-side
Showing with 2 additions and 0 deletions
docs/getting_started.rst
View file @ fe2d1581
...@@ -193,10 +193,12 @@ Alternatively you can manually start one process per GPU: ...@@ -193,10 +193,12 @@ Alternatively you can manually start one process per GPU:
> DATA=... # path to the preprocessed dataset, must be visible from all nodes > DATA=... # path to the preprocessed dataset, must be visible from all nodes
> HOST_PORT=master.example.com:9218 # one of the hosts used by the job > HOST_PORT=master.example.com:9218 # one of the hosts used by the job
> RANK=... # the rank of this process, from 0 to 127 in case of 128 GPUs > RANK=... # the rank of this process, from 0 to 127 in case of 128 GPUs
> LOCAL_RANK=... # the local rank of this process, from 0 to 7 in case of 8 GPUs per machine
> python train.py $DATA \ > python train.py $DATA \
--distributed-world-size 128 \ --distributed-world-size 128 \
--distributed-init-method 'tcp://$HOST_PORT' \ --distributed-init-method 'tcp://$HOST_PORT' \
--distributed-rank $RANK \ --distributed-rank $RANK \
--device-id $LOCAL_RANK \
--force-anneal 50 --lr-scheduler fixed --max-epoch 55 \ --force-anneal 50 --lr-scheduler fixed --max-epoch 55 \
--arch fconv_wmt_en_fr --optimizer nag --lr 0.1,4 --max-tokens 3000 \ --arch fconv_wmt_en_fr --optimizer nag --lr 0.1,4 --max-tokens 3000 \
--clip-norm 0.1 --dropout 0.1 --criterion label_smoothed_cross_entropy \ --clip-norm 0.1 --dropout 0.1 --criterion label_smoothed_cross_entropy \
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment