Tensorflow Sudoku Solver Benchmark

Benchmark tool evaluating training and inference from sudoku-ml.

Install

pip install https://github.com/cloudmercato/sudoku-game/archive/refs/heads/master.zip
pip install https://github.com/cloudmercato/sudoku-ml/archive/refs/heads/master.zip
pip install https://github.com/cloudmercato/sudoku-ml-benchmark/archive/refs/heads/master.zip

Usage

usage: sudoku-ml-bench [-h] [--batch-size BATCH_SIZE] [--epochs EPOCHS]
                       [--train-dataset-size TRAIN_DATASET_SIZE]
                       [--train-removed TRAIN_REMOVED]
                       [--infer-dataset-size INFER_DATASET_SIZE]
                       [--infer-removed INFER_REMOVED]
                       [--generator-processes GENERATOR_PROCESSES]
                       [--model-path MODEL_PATH]
                       [--model-load-file MODEL_LOAD_FILE]
                       [--model-save-file MODEL_SAVE_FILE] [--log-dir LOG_DIR]
                       [--tf-log-device] [--tf-dump-debug-info]
                       [--tf-profiler-port TF_PROFILER_PORT]
                       [--verbose VERBOSE] [--tf-verbose TF_VERBOSE]

optional arguments:
  -h, --help            show this help message and exit
  --batch-size BATCH_SIZE
  --epochs EPOCHS
  --train-dataset-size TRAIN_DATASET_SIZE
  --train-removed TRAIN_REMOVED
  --infer-dataset-size INFER_DATASET_SIZE
  --infer-removed INFER_REMOVED
  --generator-processes GENERATOR_PROCESSES
  --model-path MODEL_PATH
                        Python path to the model to compile
  --model-load-file MODEL_LOAD_FILE
                        Model load file path (h5)
  --model-save-file MODEL_SAVE_FILE
                        Model save file path (h5)
  --log-dir LOG_DIR     Tensorboard log directory
  --tf-log-device       Determines whether TF compute device info is
                        displayed.
  --tf-dump-debug-info
  --tf-profiler-port TF_PROFILER_PORT
  --verbose VERBOSE, -v VERBOSE
  --tf-verbose TF_VERBOSE, -tfv TF_VERBOSE

Docker support

Dockerfile for classic Tensorflow and the GPU version are available:

# For CPU
docker build -f Dockerfile -t sudoku-ml-bench .
docker run -it sudoku-ml-bench

# For GPU
docker build -f Dockerfile-gpu -t sudoku-ml-bench-gpu .
docker run --gpus all --ipc=host -it sudoku-ml-bench-gpu
# Add -e TF_CPP_MIN_LOG_LEVEL=3 to catch only the JSON output

The commands above will run a training, then save an inference. You can mount a volume on /models/ to keep it. In the same idea you can mount a volume on /log_dir/, to retrive the Tensorboard data.