System monitoring

config/augmentations_train/osl.yaml

@@ -8,7 +8,7 @@ defaults:
       - random_horizontal_flip
     - /augmentations/color:
     - /augmentations/postprocessing:
-      - to_tensor
+      - to_tensor_torchvision
       - normalize
 # task:
 #   - one-shot-learning

config/augmentations_val/osl.yaml

@@ -8,7 +8,7 @@ defaults:
       - random_horizontal_flip
     - /augmentations/color:
     - /augmentations/postprocessing:
-      - to_tensor
+      - to_tensor_torchvision
       - normalize
 # task:
 #   - one-shot-learning

config/datasets/detection/detection_lungs.yaml

@@ -7,7 +7,7 @@ description: Набор данных содержит РГ снимки легк
 markup_info: Набор данных содержит разметку bounding box легких и патологий.
 date_time: 18.07.2022
 
-_target_: innofw.core.datamodules.lightning_datamodules.detection.YOLOv5DataModule
+_target_: innofw.core.integrations.ultralytics.datamodule.UltralyticsDataModuleAdapter
 
 train:
   source: https://api.blackhole.ai.innopolis.university/public-datasets/lungs_detection/train.zip

config/experiments/classification/KG_090323_wpng92_mnist.yaml

@@ -2,25 +2,19 @@
 defaults:
   - override /models: classification/resnet.yaml
   - override /datasets: classification/classification_mnist.yaml
-  - override /augmentations_train: none
+  - override /augmentations_train: none #classification.yaml
   - override /augmentations_val: none
   - override /augmentations_test: none
   - override /losses: log_loss.yaml
-  - override /optimizers: lion
-#  - override /callbacks: classification.yaml
-
+  - override /optimizers: adam
+  - override /schedulers:
 
 project: "mnist_classification"
 task: "image-classification"
 random_seed: 42
-original_work_dir: ${hydra:runtime.cwd}
-weights_freq: 1
-batch_size: 32
-stop_param: 1
-epochs: 10
-
-#wandb:
-#  enable: True
-#  project: mnist
-#  group: 160323_just_testing_wandb
-#  job_type: training
+batch_size: 128
+epochs: 5
+accelerator: "gpu"
+devices: 2
+num_workers: 1
+weights_freq: 7

config/experiments/one-shot-learning/IM_190722_vwer3f23_oneshotlearning.yaml

@@ -12,4 +12,7 @@ project: "faces"
 task: "one-shot-learning"
 random_seed: 42
 weights_freq: 1
-epochs: 20
+epochs: 50
+accelerator: "gpu"
+devices: 1
+batch_size: 64

config/experiments/semantic-segmentation/KG_041223_ingieq921_unet.yaml

@@ -0,0 +1,21 @@
+# @package _global_
+defaults:
+  - override /models: semantic-segmentation/unet_smp.yaml
+  - override /datasets: semantic-segmentation/segmentation_arable.yaml
+  - override /optimizers: adam.yaml
+  - override /losses: segmentation_losses
+
+trainers:
+  overfit_batches: 20  # fit only 20 val batches and 40 train batches
+
+accelerator: gpu
+
+models:
+  in_channels: 4
+
+
+project: "segmentation"
+task: "image-segmentation"
+random_seed: 42
+weights_freq: 1
+batch_size: 10

examples/measure_anomaly_detection.sh

@@ -0,0 +1,39 @@
+#!/bin/bash
+# run like this: sudo -E env "PATH=$PATH" bash examples/measure_image_classification.sh
+sudo apt-get install -y iotop > /dev/null
+PWD=$(pwd)
+
+current_dir=${PWD##*/}
+
+var2="examples"
+if [ "$current_dir" = "$var2" ]; then
+  cd ..
+fi
+echo "" > nohup.out
+echo "" > nvidiasmi_log
+echo "" > mem_log
+echo "" > cpu_log
+
+export NO_CLI=True
+
+INITIAL_PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $INITIAL_PROCESSES
+
+
+nohup time sudo -E env "PATH=$PATH" python train.py experiments=anomaly-detection/IM_040822_rbe23lls_anomaly_detection_timeseries_lstm.yaml epochs=100 optimizers=adam accelerator=gpu &
+PID=$!
+echo "Saving cpu+ram info and nvidia-smi to cpu_log + mem_log and nvidiasmi_log"
+echo $PID
+
+PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $PROCESSES
+while [ $PROCESSES -gt $INITIAL_PROCESSES ]
+do
+  top -b n1 | grep -E 'Cpu' >> cpu_log
+  top -b n1 | grep -E 'MiB Mem' >> mem_log
+  sudo nvidia-smi --query-gpu=timestamp,name,pci.bus_id,driver_version,pstate,pcie.link.gen.max,pcie.link.gen.current,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used --format=csv | grep "00000000:25:00.0" >> nvidiasmi_log
+  sleep 1
+  PROCESSES=$(sudo lsof nohup.out | wc -l) # && echo $PROCESSES
+done
+
+sudo -E env "PATH=$PATH" python examples/measurements_compaction.py
+
+sudo kill -9 $PID

examples/measure_complexing.sh

@@ -0,0 +1,42 @@
+#!/bin/bash
+# run like this: sudo -E env "PATH=$PATH" bash examples/measure_image_classification.sh
+sudo apt-get install -y iotop > /dev/null
+PWD=$(pwd)
+
+current_dir=${PWD##*/}
+
+var2="examples"
+if [ "$current_dir" = "$var2" ]; then
+  cd ..
+fi
+echo "" > nohup.out
+echo "" > nvidiasmi_log
+echo "" > mem_log
+echo "" > cpu_log
+
+export NO_CLI=True
+
+INITIAL_PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $INITIAL_PROCESSES
+
+
+nohup time sudo -E env "PATH=$PATH" python innofw/utils/data_utils/preprocessing/band_composer.py\
+         --src_type sentinel2\
+         --src_path tests/data/images/other/satellite_cropped/sentinel2/one\
+         --channels "[\"RED\", \"GRN\", \"BLU\", \"NIR\"]" &
+PID=$!
+echo "Saving cpu+ram info and nvidia-smi to cpu_log + mem_log and nvidiasmi_log"
+echo $PID
+
+PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $PROCESSES
+while [ $PROCESSES -gt $INITIAL_PROCESSES ]
+do
+  top -b n1 | grep -E 'Cpu' >> cpu_log
+  top -b n1 | grep -E 'MiB Mem' >> mem_log
+  sudo nvidia-smi --query-gpu=timestamp,name,pci.bus_id,driver_version,pstate,pcie.link.gen.max,pcie.link.gen.current,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used --format=csv | grep "00000000:25:00.0" >> nvidiasmi_log
+  sleep 1
+  PROCESSES=$(sudo lsof nohup.out | wc -l) # && echo $PROCESSES
+done
+
+sudo -E env "PATH=$PATH" python examples/measurements_compaction.py
+
+sudo kill -9 $PID

examples/measure_detection_yolov5.sh

@@ -0,0 +1,32 @@
+#!/bin/bash
+# run like this: sudo -E env "PATH=$PATH" bash examples/measure_image_classification.sh
+sudo apt-get install -y iotop > /dev/null
+PWD=$(pwd)
+
+current_dir=${PWD##*/}
+
+var2="examples"
+if [ "$current_dir" = "$var2" ]; then
+  cd ..
+fi
+echo "" > nohup.out
+echo "" > nvidiasmi_log
+echo "" > mem_log
+echo "" > cpu_log
+
+export NO_CLI=True
+
+nohup time sudo -E env "PATH=$PATH" python train.py experiments=detection/KA_120722_8adfcdaa_yolov5.yaml epochs=100 optimizers=adam accelerator=gpu &
+PID=$!
+echo "Saving cpu+ram info and nvidia-smi to cpu_log+mem_log and nvidiasmi_log"
+echo $PID
+
+while true
+do 
+  top -b n1 | grep -E 'Cpu' >> cpu_log
+  top -b n1 | grep -E 'MiB Mem' >> mem_log
+  sudo nvidia-smi --query-gpu=timestamp,name,pci.bus_id,driver_version,pstate,pcie.link.gen.max,pcie.link.gen.current,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used --format=csv >> nvidiasmi_log
+  sleep 1
+done 
+
+kill $PID

examples/measure_image_classification.sh

@@ -0,0 +1,39 @@
+#!/bin/bash
+# run like this: sudo -E env "PATH=$PATH" bash examples/measure_image_classification.sh
+sudo apt-get install -y iotop > /dev/null
+PWD=$(pwd)
+
+current_dir=${PWD##*/}
+
+var2="examples"
+if [ "$current_dir" = "$var2" ]; then
+  cd ..
+fi
+echo "" > nohup.out
+echo "" > nvidiasmi_log
+echo "" > mem_log
+echo "" > cpu_log
+
+export NO_CLI=True
+
+INITIAL_PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $INITIAL_PROCESSES
+
+
+nohup time sudo -E env "PATH=$PATH" python train.py experiments=classification/KG_090323_wpng92_mnist.yaml epochs=100 &
+PID=$!
+echo "Saving cpu+ram info and nvidia-smi to cpu_log + mem_log and nvidiasmi_log"
+echo $PID
+
+PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $PROCESSES
+while [ $PROCESSES -gt $INITIAL_PROCESSES ]
+do
+  top -b n1 | grep -E 'Cpu' >> cpu_log
+  top -b n1 | grep -E 'MiB Mem' >> mem_log
+  sudo nvidia-smi --query-gpu=timestamp,name,pci.bus_id,driver_version,pstate,pcie.link.gen.max,pcie.link.gen.current,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used --format=csv | grep "00000000:25:00.0" >> nvidiasmi_log
+  sleep 1
+  PROCESSES=$(sudo lsof nohup.out | wc -l) # && echo $PROCESSES
+done
+
+sudo -E env "PATH=$PATH" python examples/measurements_compaction.py
+
+sudo kill -9 $PID

examples/measure_lin_regression.sh

@@ -0,0 +1,39 @@
+#!/bin/bash
+# run like this: sudo -E env "PATH=$PATH" bash examples/measure_image_classification.sh
+sudo apt-get install -y iotop > /dev/null
+PWD=$(pwd)
+
+current_dir=${PWD##*/}
+
+var2="examples"
+if [ "$current_dir" = "$var2" ]; then
+  cd ..
+fi
+echo "" > nohup.out
+echo "" > nvidiasmi_log
+echo "" > mem_log
+echo "" > cpu_log
+
+export NO_CLI=True
+
+INITIAL_PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $INITIAL_PROCESSES
+
+
+nohup time sudo -E env "PATH=$PATH" python train.py experiments=regression/KA_130722_9f7134db_linear_regression.yaml &
+PID=$!
+echo "Saving cpu+ram info and nvidia-smi to cpu_log + mem_log and nvidiasmi_log"
+echo $PID
+
+PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $PROCESSES
+while [ $PROCESSES -gt $INITIAL_PROCESSES ]
+do
+  top -b n1 | grep -E 'Cpu' >> cpu_log
+  top -b n1 | grep -E 'MiB Mem' >> mem_log
+  sudo nvidia-smi --query-gpu=timestamp,name,pci.bus_id,driver_version,pstate,pcie.link.gen.max,pcie.link.gen.current,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used --format=csv | grep "00000000:25:00.0" >> nvidiasmi_log
+  sleep 1
+  PROCESSES=$(sudo lsof nohup.out | wc -l) # && echo $PROCESSES
+done
+
+sudo -E env "PATH=$PATH" python examples/measurements_compaction.py
+
+sudo kill -9 $PID

examples/measure_oneshot_learning.sh

@@ -0,0 +1,39 @@
+#!/bin/bash
+# run like this: sudo -E env "PATH=$PATH" bash examples/measure_image_classification.sh
+sudo apt-get install -y iotop > /dev/null
+PWD=$(pwd)
+
+current_dir=${PWD##*/}
+
+var2="examples"
+if [ "$current_dir" = "$var2" ]; then
+  cd ..
+fi
+echo "" > nohup.out
+echo "" > nvidiasmi_log
+echo "" > mem_log
+echo "" > cpu_log
+
+export NO_CLI=True
+
+INITIAL_PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $INITIAL_PROCESSES
+
+
+nohup time sudo -E env "PATH=$PATH" python train.py experiments=one-shot-learning/IM_190722_vwer3f23_oneshotlearning.yaml &
+PID=$!
+echo "Saving cpu+ram info and nvidia-smi to cpu_log + mem_log and nvidiasmi_log"
+echo $PID
+
+PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $PROCESSES
+while [ $PROCESSES -gt $INITIAL_PROCESSES ]
+do
+  top -b n1 | grep -E 'Cpu' >> cpu_log
+  top -b n1 | grep -E 'MiB Mem' >> mem_log
+  sudo nvidia-smi --query-gpu=timestamp,name,pci.bus_id,driver_version,pstate,pcie.link.gen.max,pcie.link.gen.current,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used --format=csv | grep "00000000:25:00.0" >> nvidiasmi_log
+  sleep 1
+  PROCESSES=$(sudo lsof nohup.out | wc -l) # && echo $PROCESSES
+done
+
+sudo -E env "PATH=$PATH" python examples/measurements_compaction.py
+
+sudo kill -9 $PID

examples/measure_segmentation.sh

@@ -0,0 +1,39 @@
+#!/bin/bash
+# run like this: sudo -E env "PATH=$PATH" bash examples/measure_image_classification.sh
+sudo apt-get install -y iotop > /dev/null
+PWD=$(pwd)
+
+current_dir=${PWD##*/}
+
+var2="examples"
+if [ "$current_dir" = "$var2" ]; then
+  cd ..
+fi
+echo "" > nohup.out
+echo "" > nvidiasmi_log
+echo "" > mem_log
+echo "" > cpu_log
+
+export NO_CLI=True
+
+INITIAL_PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $INITIAL_PROCESSES
+
+
+nohup time sudo -E env "PATH=$PATH" python train.py experiments=semantic-segmentation/KG_041223_ingieq921_unet.yaml epochs=100 optimizers=adam accelerator=gpu &
+PID=$!
+echo "Saving cpu+ram info and nvidia-smi to cpu_log + mem_log and nvidiasmi_log"
+echo $PID
+
+PROCESSES=$(sudo lsof nohup.out | wc -l) && echo $PROCESSES
+while [ $PROCESSES -gt $INITIAL_PROCESSES ]
+do
+  top -b n1 | grep -E 'Cpu' >> cpu_log
+  top -b n1 | grep -E 'MiB Mem' >> mem_log
+  sudo nvidia-smi --query-gpu=timestamp,name,pci.bus_id,driver_version,pstate,pcie.link.gen.max,pcie.link.gen.current,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used --format=csv | grep "00000000:25:00.0" >> nvidiasmi_log
+  sleep 1
+  PROCESSES=$(sudo lsof nohup.out | wc -l) # && echo $PROCESSES
+done
+
+sudo -E env "PATH=$PATH" python examples/measurements_compaction.py
+
+sudo kill -9 $PID