Hyperdash | Machine Learning Monitoring

Machine Learning Monitoring for Data Scientists

Real-Time Alerts

Hyperdash automatically detects the status of your training. Get notified when your training completes, fails, crashes or gets disconnected.

Track Logs Remotely

Stream console logs from your machine to your phone in real-time. No need to SSH to see errors or model performance.

Private and Secure

Since we have no visibility into your Dataset or Model, your IP stays yours. State-of-the-art security keeps your logs and metrics safe.

Get setup in seconds

Install the command-line tool and signup for an account.

$

$ pip install hyperdash
$ hyperdash signup

Command Line

Monitor any command by prefixing it with Hyperdash.

$ hyperdash run -n "MNIST" \
    python mnist.py
Iter 1280, Loss= 1.576423, Acc= 0.51562
Iter 2560, Loss= 1.450179, Acc= 0.53906
Iter 3840, Loss= 1.160066, Acc= 0.64844
Iter 5120, Loss= 0.898589, Acc= 0.73438
Iter 6400, Loss= 0.685712, Acc= 0.75781
Iter 7680, Loss= 1.085666, Acc= 0.64844

$ hyperdash run -n "MNIST RNN" python mnist.py
Iter 1280, Loss= 1.576423, Accuracy= 0.51562
Iter 2560, Loss= 1.450179, Accuracy= 0.53906
Iter 3840, Loss= 1.160066, Accuracy= 0.64844
Iter 5120, Loss= 0.898589, Accuracy= 0.73438
Iter 6400, Loss= 0.685712, Accuracy= 0.75781
Iter 7680, Loss= 1.085666, Accuracy= 0.64844

Jupyter Notebook

Watch the status of specific notebook cells.

In [13]:from hyperdash import monitor_cell

In [14]:%%monitor_cell "MNIST RNN"
with tf.Session() as sess:
  sess.run(init)
  step = 1
  # ...

Docs for Getting Started