Reduce epoches in pytorch2.0 model (#1542)

* reducing epoches for pytorch2

* adding comment

* chaning to 36 epochs

* renaming

* renaming

perfmetrics/scripts/ml_tests/pytorch/run_model.sh

@@ -14,6 +14,7 @@
 # limitations under the License.
 
 PYTORCH_VESRION=$1
+NUM_EPOCHS=80
 
 # Install golang
 wget -O go_tar.tar.gz https://go.dev/dl/go1.21.4.linux-amd64.tar.gz -q
@@ -76,8 +77,13 @@ python -c 'import torch;torch.hub.list("facebookresearch/xcit:main")'
 # (TulsiShah) TODO: Pytorch 2.0 compile mode has issues (https://github.com/pytorch/pytorch/issues/94599),
 # which is fixed in pytorch version 2.1.0 (https://github.com/pytorch/pytorch/pull/100071).
 # We'll remove this workaround once we update our Docker image to use Pytorch 2.1.0 or greater version.
+# Reducing the epochs as pytorch2 long haul tests are running on NVIDIA L4 machines, which lack the powerful GPU of
+# the NVIDIA A100. So it is taking longer time to complete the training. We will set it back to 80 when the NVIDIA A100 GPU machine
+# will be available.
 if [ ${PYTORCH_VESRION} == "v2" ];
 then
+  NUM_EPOCHS=36
+
   allowed_functions_file="/opt/conda/lib/python3.10/site-packages/torch/_dynamo/allowed_functions.py"
   # Update the pytorch library code to bypass the kernel-cache
   echo "Updating the pytorch library code to Disallow_in_graph distributed API.."
@@ -197,7 +203,7 @@ gsutil cp start_time.txt $ARTIFACTS_BUCKET_PATH/
     --norm_last_layer False \
     --use_fp16 False \
     --clip_grad 0 \
-    --epochs 80 \
+    --epochs $NUM_EPOCHS \
     --global_crops_scale 0.25 1.0 \
     --local_crops_number 10 \
     --local_crops_scale 0.05 0.25 \