fixed typo in multi-gpu implementation

psteinb · Apr 23, 2018 · 1356f87 · 1356f87
1 parent 16fe67f
commit 1356f87
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 3 deletions.
diff --git a/models/tf_details/resnet_run_loop.py b/models/tf_details/resnet_run_loop.py
@@ -327,7 +327,7 @@ def resnet_main(flags, model_function, input_function, opts = None):
     logging.warning("batch sizes differ in model %i %s", flags.batch_size, opts["batch_size"])
 
   if ngpus > 1:
-    validate_batch_size_for_multi_gpu(bs)
+    validate_batch_size_for_multi_gpu(flags.batch_size)
     # There are two steps required if using multi-GPU: (1) wrap the model_fn,
     # and (2) wrap the optimizer. The first happens here, and (2) happens
     # in the model_fn itself when the optimizer is defined.

diff --git a/scripts/tf-full/resnet50-singularity-bs128.slurm b/scripts/tf-full/resnet50-singularity-bs128.slurm
@@ -10,4 +10,8 @@ cd /home/steinba/development/deeprace/
 pwd
 module load singularity/2.4.2
 
-singularity exec -B $PWD:/home/steinba/deeprace --nv /scratch/steinba/tf1.7-plus.simg python3 /home/steinba/development/deeprace/deeprace.py train -b tf -O batch_size=128 -c "k80:1,fs:nfs,singularity:lustre" -t /home/steinba/development/deeprace/scripts/tf-full/tf-full-resnet56v1-bs128-singularity.tsv resnet56v1
+TDIR=`mktemp -d`
+
+singularity exec -B $PWD:/home/steinba/deeprace --nv /scratch/steinba/tf1.7-plus.simg python3 /home/steinba/development/deeprace/deeprace.py train -R ${TDIR} -b tf -O batch_size=128 -c "k80:1,fs:nfs,singularity:lustre" -t /home/steinba/development/deeprace/scripts/tf-full/tf-full-resnet56v1-bs128-singularity.tsv resnet56v1
+
+rm -rf ${TDIR}
diff --git a/scripts/tf-full/resnet50-singularity.slurm b/scripts/tf-full/resnet50-singularity.slurm
@@ -10,4 +10,8 @@ cd /home/steinba/development/deeprace/
 pwd
 module load singularity/2.4.2
 
-singularity exec -B $PWD:/home/steinba/deeprace --nv /scratch/steinba/tf1.7-plus.simg python3 /home/steinba/development/deeprace/deeprace.py train -b tf -c "k80:1,fs:nfs,singularity:lustre" -t /home/steinba/development/deeprace/scripts/tf-full/tf-full-resnet56v1-singularity.tsv resnet56v1
+TDIR=`mktemp -d`
+
+singularity exec -B $PWD:/home/steinba/deeprace --nv /scratch/steinba/tf1.7-plus.simg python3 /home/steinba/development/deeprace/deeprace.py train -R ${TDIR} -b tf -c "k80:1,fs:nfs,singularity:lustre" -t /home/steinba/development/deeprace/scripts/tf-full/tf-full-resnet56v1-singularity.tsv resnet56v1
+
+rm -rf ${TDIR}