rel-1.0.11: Merge pull request #161 from brandonwillard/fix-fusion-c-code-condition

This release contains:

• The removal of theano.dot (use theano.tensor.dot, instead)
• A fixed Elemwise fusion for Ops without C implementations
• New modules theano.scan and theano.compile.function that replace theano.scan_module and theano.compile.function_module, respectively
• A theano.tensor.extra_ops.broadcast_to Op for NumPy's numpy.broadcast_to
• A JAX conversion for the Eye Op

rel-1.0.10: Change coverage to codecov (#124)

This release contains:

• matrix_power is now implemented via exponentiation by squaring
• Test value errors (e.g. missing test values) now raise theano.gof.utils.TestValueError
• Improvements to test value validation and evaluation, plus a new Variable.get_test_value method

rel-1.0.9: Prevent test value computation from complaining about subtensor creation

This release contains:

• A fix for test value issues arising when theano.config.compute_test_value = "raise" during Subtensor creation

rel-1.0.8: Remove redundant AdvancedBoolean*Subtensor classes

This release includes:

• Major boolean indexing fixes

rel-1.0.7

This release includes:

• Subtensor fixes
• Updates to theano.tensor.at_tensor_variable so that it processes constants more consistently
• A new broadcast_shape function
• Updates to get_scalar_constant_value making it capable of handling NoneTypes
• Moved equal_computations and is_same_graph to theano.gof.toolbox

rel-1.0.6

Adding action to publish to PyPI