run_script_4.py

import tensorflow as tf
import numpy as np
import Functions as fn
import importlib

importlib.reload(fn)

# testdatanorm = np.random.rand(100, 16, 100)
# testlabelsnorm = np.random.rand(100, 100)

checkpointing_config = tf.estimator.RunConfig(
    save_checkpoints_secs=20 * 60,  # Save checkpoints every 20 minutes.
    keep_checkpoint_max=2,  # Retain the 10 most recent checkpoints.
    save_summary_steps=1000,
    log_step_count_steps=1000
)

classifier = tf.estimator.Estimator(

    model_fn=fn.FC_sym_model_ri,
    model_dir='large_ri_model',
    config=checkpointing_config,
    params={

        # 'feature_columns': the_feature_column,

        # Layers.
        'NUM_COOKIES': 16,
        'COOKIE_DENSE': [100, 80, 40, 20],
        'CNN': [[32, 10], [32, 10]],  # Convolutional layers
        'POOL': 80,  # Global Pooling Label
        'DENSE': [320, 200, 200, 200, 200],  # Dense layers
        'OUT': 200  # output dimensions
    }
)