slurm
Field |
Default |
Description |
Type |
---|---|---|---|
name |
??? |
The referenced name of the object to construct |
str |
conf_key |
${oc.env:SLURM_DEFAULT_KEY,missing} |
Slurm config key to use |
str |
time_limit |
${oc.env:SLURM_TIME_LIMIT,3-00:00:00} |
Time limit string |
str |
num_jobs |
1 |
Number of redundant jobs to launch |
int |
comment |
None |
An optional comment to add to the experiment |
str | None |
master_port |
${ml.get_random_slurm_port:1337} |
The master port to use |
int |
model_parallelism |
1 |
The number of model parallel processes |
int |
pipeline_parallelism |
1 |
The number of pipeline parallel processes |
int |
backend |
None |
The distributed backend |
str | None |
model_parallel_backend |
None |
The model parallel backend |
str | None |
pipeline_parallel_backend |
None |
The pipeline parallel backend |
str | None |
data_parallel_backend |
None |
The data parallel backend |
str | None |
account |
None |
The account to use, if required |
str | None |