Montreal Data License (MDL)

The following licensing language is made available under CC-BY4. Attribution
should be made to Montreal Data License (MDL), or License language based on
Montreal Data License.
The authors are not legal advisors to the individuals and entities making use
of these licensing terms. The licensing terms can be combined as needed to match
the rights conferred by the licensor.
The language below assumes that all rights are granted, however each right should
be conferred or not based on the users intent.
Data License for use in AI and ML:

This license covers the Data made available by Licensor to you (Licensee) under
the following terms. Licensees use of the data consists acceptance of the terms
of this license agreement (License).

1.  Definitions
a. Data means the informational content (individually or as a whole) made available
by Licensor.
b. Model means machine-learning or artificial-intelligence based algorithms,
or assemblies thereof that, in combination with different techniques, may be
used to obtain certain results. Without limitation, such results can be insights
on past data patterns, predictions on future trends or more abstract results.
c. Output means the results of operating a Trained Model as embodied in informational
content resulting therefrom.
d. Representation is a transformation of a piece of data into a different form.
Good representations can be used as input to perform useful tasks.
e. Labelled Data means the associated metadata and informational content derived
from Data which identify, comment or otherwise derive information from Data,
such as tags and labels.
  f.	Licensor means the individual or entity making the Data available to you.
g. Third Parties means individuals or entities that are not under common control
with Licensee.
h. Train means to expose an Untrained Model to the Data in order to adjust the
weights, hyperparameters and/or structure thereof.
i. Trained Model means a Model that is exposed to Data such that its weights,
parameters and architecture embody insights from the Data.
j. Untrained Model means Model that is conceived and reduced to practice as
to its structure, components and architecture but that has not been trained
on Data such that its weights, parameters and architecture do not embody insights
from the Data.

2.  General Clauses
a. Unless otherwise agreed in writing by the parties, the data is licensed as
is and as available. Licensor excludes all representations, warranties, obligations,
and liabilities, whether express or implied, to the maximum extent permitted
by law.
b. Nothing in this License permits Licensee to make use of Licensors trademarks,
tradenames, logos or to otherwise suggest endorsement or misrepresent the relationship
between the parties.
c. The rights granted under this license are deemed to be non-exclusive, worldwide,
perpetual and irrevocable, unless otherwise specified in writing by Licensor.
d. Without limiting Licensees rights available under applicable law, all rights
not expressly granted hereunder are hereby reserved by Licensor. The Data and
the database under which it is made available remain the property of Licensor
(and/or its affiliates or licensors).
e. This license shall be terminated upon any breach by Licensee of the terms
of this License.

3.  Licensed Rights to the Data
a. Licensor hereby grants the following rights to Licensee with respect to making
use of the Data itself.
i. Access the Data, where access means to access, view and/or download the Data
to view it and evaluate it (evaluation algorithms may be exposed to it, but
no Untrained Models).
    ii. Creation of Tagged Data.
iii. Distribute the Data, i.e. to make all or part of the Data available to
Third Parties under the same terms as those of this License.
    iv.	Creation of a Representation of the Data.

4.  Licensed Rights in Conjunction with Models.
a. Licensor hereby grants the following rights to Licensee with respect to making
use of the Data in conjunction with Models.
i. Benchmark: To access the Data, use the Data as training data to evaluate
the efficiency of different Untrained Models, algorithms and structures, but
excludes reuse of the Trained Model, except to show the results of the Training.
This includes the right to use the dataset to measure performance of a Trained
or Un-trained Model, without however having the right to carry-over weights,
code or architecture or implement any modifications resulting from the Evaluation.
ii. Research: To access the Data, use the Data to create or improve Models,
but with-out the right to use the Output or resulting Trained Model for any
purpose other than evaluating the Model Research under the same terms.
iii. Publish: To make available to Third Parties the Models resulting from Research,
provided however that third parties accessing such Trained Models have the right
to use them for Research or Publication only.
iv. Internal Use: To access the Data, use the Data to create or improve Models
and resulting Output, but without the right to Output Commercialization or Model
Commercialization. The Output can be used internally for any purpose, but not
made available to Third Parties or for their benefit.
b. The rights granted in (a) above exclude the following rights with respect
to making use of the Data in conjunction with Models:
i. Output Commercialization: To access the Data, use the Data to create or improve
Models and resulting Output, with the right to make the Output available to
Third Parties or to use it for their benefit, without the right to Model Commercialization.
ii. Model Commercialization: Make a Trained Model itself available to a Third
Party, or embodying the Trained Model in a product or service, with or without
direct access to the Output for such Third Party.

5. Attribution and Notice Attribution and Notice. The origin of the Data and
notices included with the Data shall be made available to Third Parties to whom
the Data, Output and/Model have been made available. Any distribution of all
or part of the Data shall be done under the same terms as those of this License.
Licensee shall make commercially reasonable efforts to link to the source of
the Data. If so indicated by the Licensor in writing alongside the Data that
the use shall be deemed confidential, then Licensee shall not publicly refer
to Licensor and/or the source of the Data.