We will be making a number of breaking changes with 0.7, because we are prepping for a 1.0 release this year, and we have code that needs major refactoring, which will affect serialization, API usage, and parameter names.
This is a necessary growing pain NuPIC needs to move forward.
How to upgrade:
Find / replace in your codebase:
-
tpParams
==>tmParams
-
tpEnable
==>tmEnable
How to upgrade:
Find / replace in your codebase:
-
CLAModel
==>HTMPredictionModel
-
clamodel
==>htmpredictionmodel
-
claModel
==>htmPredictionModel
In any model params, change the model
from CLA
to HTMPrediction
, like this:
-
'model': 'CLA',
==>'model': 'HTMPrediction',
How to upgrade:
- See PR linked above for details on the module & file name changes.
How to upgrade:
Find/replace:
-
nupic.research
==>nupic.algorithms
Module name changes;
-
BacktrackingTM
==>backtracking_tm
-
BacktrackingTMCPP
==>backtracking_tm_cpp
How to upgrade:
You probably don’t need to do anything unless you have custom encoder calls the formatBits
function on an encoder. In that case, sorry we removed this function from the API.
Python module name changes
The following PRs are refactorings to change python module names to snake_case.
- https://github.com/numenta/nupic/pull/3550
- https://github.com/numenta/nupic/pull/3561
- https://github.com/numenta/nupic/pull/3611
- https://github.com/numenta/nupic/pull/3614
- https://github.com/numenta/nupic/pull/3618
- https://github.com/numenta/nupic/pull/3627
- https://github.com/numenta/nupic/pull/3640