TinyMind

Build better models,
easier, together.

Just-push-button model training

It's a familiar experience - we thought we've finally finished the work, but the real challenge has just started. Setting up the environment, putting things together, it takes forever. With TinyMind, call it done when you finish coding.

from keras.models import Sequential
from keras.layers import Dense, Activation

model = Sequential([
    Dense(32, input_shape=(784,)),
    Activation('relu'),
    Dense(10),
    Activation('softmax'),
])

model.compile(
    optimizer='rmsprop',
    loss='binary_crossentropy',
    metrics=['accuracy'
])

model.fit(data, labels)
Resource per execution:

No more “thought experiments”. Try them.

Machine learning is a lot of trial-and-error. Don't struggle again to find the best parameters and features. Simply specify your ideas and TinyMind will run through them for you. No guesswork. No endless waiting.

Range Parameter
Specify parameter values with a range
Start
End
Step
dropout
float
iterations
int
Execution Batch #5
mnist #5.1
mnist #5.2
mnist #5.3

Pricing

All the features. No hidden layers.

$0.09/ CPU per hour

High-performance Intel Xeon processor.
6.5GB memory and 20GB SSD per CPU core.
Choice of 1, 2, or 4 cores.
That's all you pay - everything (networking, IO) is included.

$0.99/ GPU per hour

Nvidia Tesla K80 GPU with 12GB GDDR5 memory.
Each GPU comes with 4 CPU cores and the associated memory and storage.
That's all you pay - everything (CPU, networking, IO) is included.
All paid users automatically receive:
Up to 5 concurrent executions.
Up to 10 scheduled executions.
At least 250GB of free storage.
Per-minute billing.
Unlimited private models and datasets.
7-day single-job run time.

Ready to get started?