Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
* Add autodiff save/load functionality * - * Update autodiffcomposition.py * Update test_autodiffcomposition.py * Merge branch 'devel' of https://github.com/PrincetonUniversity/PsyNeuLink into devel * • Autodiff: - add save and load methods (from Samyak) - test_autodiffcomposition.py: add test_autodiff_saveload, but commented out for now, as it may be causing hanging on PR * • Autodiff: - add save and load methods (from Samyak) - test_autodiffcomposition.py: add test_autodiff_saveload, but commented out for now, as it may be causing hanging on PR * - * - * • pytorchcomponents.py: - pytorch_function_creator: add SoftMax • transferfunctions.py: - disable changes to ReLU.derivative for now * • utilities.py: - iscompatible: attempt to replace try and except, commented out for now * - * - * • autodiffcomposition.py: - save and load: augment file and directory handling - exclude processing of any ModulatoryProjections * - * - * - * • autodiffcomposition.py save(): add projection.matrix.base = matrix (fixes test_autodiff_saveload) * - * • autodiffcomposition.py: - save: return path • test_autodiffcomposition.py: - test_autodiff_saveload: modify to use current working directory rather than tmp * • autodiffcomposition.py: - save() and load(): ignore CIM, learning, and other modulation-related projections * • autodiffcomposition.py: - load(): change test for path (failing on Windows) from PosixPath to Path * • autodiffcomposition.py: - add _runtime_learning_rate attribute - _build_pytorch_representation(): use _runtime_learning_rate attribute for optimizer if provided in call to learn else use learning_rate specified at construction • compositionrunner.py: - assign learning_rate to _runtime_learning_rate attribute if specified in call to learn * - * [skip ci] * [skip ci] * [skip ci] • autodiffcomposition.py: load(): add testing for match of matrix shape * [skip ci] • N-back: - reset em after each run - save and load weights - torch epochs = batch size (number training stimuli) * num_epochs * [skip ci] * [skip ci] * Feat/add pathway default matrix (#2518) * • compositioninterfacemechanism.py: - _get_source_node_for_input_CIM: restore (modeled on _get_source_of_modulation_for_parameter_CIM) but NEEDS TESTS - _get_source_of_modulation_for_parameter_CIM: clean up comments, NEEDS TESTS * - * - * - * - * - * - * • Nback - EM uses ContentAddressableMemory (instead of DictionaryMemory) - Implements FFN for comparison of current and retrieved stimulus and context • Project: replace all instances of "RETREIVE" with "RETRIEVE" * • objectivefunctions.py - add cosine_similarity (needs compiled version) * • Project: make COSINE_SIMILARITY a synonym of COSINE • nback_CAM_FFN: - refactor to implement FFN and task input - assign termination condition for execution that is dependent on control - ContentAddressableMemory: selection_function=SoftMax(output=MAX_INDICATOR, gain=SOFT_MAX_TEMP) • DriftOnASphereIntegrator: - add dimension as dependency for initializer parameter * - * - * - * - * - * - * - * - * - * - * - * - * - * - * • test_integrator.py: Added identicalness test for DriftOnASphereIntegrator agains nback-paper implementation. * - * - * Parameters: allow _validate_ methods to reference other parameters (#2512) * • Scripts: - Updated N-back to use objective_mechanism, with commented out code for version that doesn't use it once bug is fixed - Deleted N-back_WITH_OBJECTIVE_MECH.py * • Scripts: - Updated N-back to use objective_mechanism, with commented out code for version that doesn't use it once bug is fixed - Deleted N-back_WITH_OBJECTIVE_MECH.py * • N-back.py: - added stimulus generation per nback-paper protocol * - N-back.py tstep(s) -> trial(s) * - * - * • N-back.py - comp -> nback_model - implement stim_set() method * - * • N-back.py: - added training set generation * - * - * • N-back.py - modularized script * - * - * - * - * • showgraph.py: - _assign_processing_components(): fix bug in which nested graphs not highlighted in animation. * • showgraph.py * composition.py - add further description of animation, including note that animation of nested Compostions is limited. * • showgraph.py * composition.py - add animation to N-back doc * • autodiffcomposition.py - __init__(): move pathways arg to beginning, to capture positional assignment (i.e. w/o kw) * - * • N-back.py - ffn: implement as autodiff; still needs small random initial weight assignment * • pathway.py - implement default_projection attribute * • pathway.py - implement default_projection attribute * • utilities.py: random_matrxi: refactored to allow negative values and use keyword ZERO_CENTER * • projection.py RandomMatrix: added class that can be used to pass a function as matrix spec * • utilities.py - RandomMatrix moved here from projection.py • function.py - get_matrix(): added support for RandomMatrix spec * • port.py - _parse_port_spec(): added support for RandomMatrix * • port.py - _parse_port_spec(): added support for RandomMatrix * • utilities.py - is_matrix(): modified to support random_matrix and RandomMatrix * • composition.py - add_linear_processing_pathway: add support for default_matrix argument (replaces default for MappingProjection for any otherwise unspecified projections) though still not used. * - * - RandomMatrix: moved from Utilities to Function * - * [skip ci] * [skip ci] * [skip ci] • N-back.py - clean up script * [skip ci] • N-back.py - further script clean-up * [skip ci] * [skip ci] * [skip ci] * [skip ci] • BeukersNBackModel.rst: - Overview written - Needs other sections completed * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] • N-back.py: - replace functions of TransferMechanisms with ReLU - replace function of Decision Mechanisms with SoftMax - more doc cleanup * [skip ci] • N-back.py: - replace functions of TransferMechanisms with ReLU - replace function of Decision Mechanisms with SoftMax - more doc cleanup * [skip ci] * - * - * [skip ci] * [skip ci] • composition.py: implement default_projection_matrix in add_XXX_pathway() methods * [skip ci] • composition.py: implement default_projection_matrix in add_XXX_pathway() methods * [skip ci] • test_composition.py: - add test_pathway_tuple_specs() * - * - * [skip ci] * [skip ci] * [skip ci] * - Co-authored-by: jdcpni <pniintel55> Co-authored-by: Katherine Mantel <[email protected]> * Feat/add pathway default matrix (#2519) * • compositioninterfacemechanism.py: - _get_source_node_for_input_CIM: restore (modeled on _get_source_of_modulation_for_parameter_CIM) but NEEDS TESTS - _get_source_of_modulation_for_parameter_CIM: clean up comments, NEEDS TESTS * - * - * - * - * - * - * • Nback - EM uses ContentAddressableMemory (instead of DictionaryMemory) - Implements FFN for comparison of current and retrieved stimulus and context • Project: replace all instances of "RETREIVE" with "RETRIEVE" * • objectivefunctions.py - add cosine_similarity (needs compiled version) * • Project: make COSINE_SIMILARITY a synonym of COSINE • nback_CAM_FFN: - refactor to implement FFN and task input - assign termination condition for execution that is dependent on control - ContentAddressableMemory: selection_function=SoftMax(output=MAX_INDICATOR, gain=SOFT_MAX_TEMP) • DriftOnASphereIntegrator: - add dimension as dependency for initializer parameter * - * - * - * - * - * - * - * - * - * - * - * - * - * - * • test_integrator.py: Added identicalness test for DriftOnASphereIntegrator agains nback-paper implementation. * - * - * Parameters: allow _validate_ methods to reference other parameters (#2512) * • Scripts: - Updated N-back to use objective_mechanism, with commented out code for version that doesn't use it once bug is fixed - Deleted N-back_WITH_OBJECTIVE_MECH.py * • Scripts: - Updated N-back to use objective_mechanism, with commented out code for version that doesn't use it once bug is fixed - Deleted N-back_WITH_OBJECTIVE_MECH.py * • N-back.py: - added stimulus generation per nback-paper protocol * - N-back.py tstep(s) -> trial(s) * - * - * • N-back.py - comp -> nback_model - implement stim_set() method * - * • N-back.py: - added training set generation * - * - * • N-back.py - modularized script * - * - * - * - * • showgraph.py: - _assign_processing_components(): fix bug in which nested graphs not highlighted in animation. * • showgraph.py * composition.py - add further description of animation, including note that animation of nested Compostions is limited. * • showgraph.py * composition.py - add animation to N-back doc * • autodiffcomposition.py - __init__(): move pathways arg to beginning, to capture positional assignment (i.e. w/o kw) * - * • N-back.py - ffn: implement as autodiff; still needs small random initial weight assignment * • pathway.py - implement default_projection attribute * • pathway.py - implement default_projection attribute * • utilities.py: random_matrxi: refactored to allow negative values and use keyword ZERO_CENTER * • projection.py RandomMatrix: added class that can be used to pass a function as matrix spec * • utilities.py - RandomMatrix moved here from projection.py • function.py - get_matrix(): added support for RandomMatrix spec * • port.py - _parse_port_spec(): added support for RandomMatrix * • port.py - _parse_port_spec(): added support for RandomMatrix * • utilities.py - is_matrix(): modified to support random_matrix and RandomMatrix * • composition.py - add_linear_processing_pathway: add support for default_matrix argument (replaces default for MappingProjection for any otherwise unspecified projections) though still not used. * - * - RandomMatrix: moved from Utilities to Function * - * [skip ci] * [skip ci] * [skip ci] • N-back.py - clean up script * [skip ci] • N-back.py - further script clean-up * [skip ci] * [skip ci] * [skip ci] * [skip ci] • BeukersNBackModel.rst: - Overview written - Needs other sections completed * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] • N-back.py: - replace functions of TransferMechanisms with ReLU - replace function of Decision Mechanisms with SoftMax - more doc cleanup * [skip ci] • N-back.py: - replace functions of TransferMechanisms with ReLU - replace function of Decision Mechanisms with SoftMax - more doc cleanup * [skip ci] * - * - * [skip ci] * [skip ci] • composition.py: implement default_projection_matrix in add_XXX_pathway() methods * [skip ci] • composition.py: implement default_projection_matrix in add_XXX_pathway() methods * [skip ci] • test_composition.py: - add test_pathway_tuple_specs() * - * - * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] • composition.py: - add_linear_processing_pathway: fixed bug when Reinforcement or TDLearning are specified • test_composition.py: - test_pathway_tuple_specs: add tests for Reinforcement and TDLearning * • composition.py: - add_linear_processing_pathway: fixed bug when Reinforcement or TDLearning are specified • test_composition.py: - test_pathway_tuple_specs: add tests for Reinforcement and TDLearning Co-authored-by: jdcpni <pniintel55> Co-authored-by: Katherine Mantel <[email protected]> * autodiff: Use most recent context while save/load * tests/autodiff: Use portable path join * autodiff: Add assertions for save/load * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * • autodiffcomposition, test_autodiff_saveload: - merged from feat/autodiff_save * - * - * - * • autodiffcomposition.py - fix path assignment bug * - * • N-back mods * • N-back: reimplementing get_run_inputs * - * - * - * - * • N-back.py - refactoring of generate_stim_seq: continuous presentation of stimuli balancing of trial_types (with fill-in) return trial_type_seq * [skip ci] * Merge branch 'devel' of https://github.com/PrincetonUniversity/PsyNeuLink into devel � Conflicts: � psyneulink/library/compositions/autodiffcomposition.py * Merge branch 'devel' of https://github.com/PrincetonUniversity/PsyNeuLink into devel � Conflicts: � psyneulink/library/compositions/autodiffcomposition.py * [skip ci] * [skip ci] • N-back.py - docstring mods * [skip ci] * [skip ci] • N-back.py: add Kane stimuli (2back) * [skip ci] * [skip ci] * [skip ci] * • N-back.py - add analyze_results() * [skip ci] • N-back.py - add analyze_results() * [skip ci] • N-back.py: - analyze_results: fully implemented * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] * [skip ci] • composition.py, pytorchmodelcreator.py - modify loss_spec to use keywords for loss • Nback.py: - Autodiff loss_spec = CROSS_ENTROPY * [skip ci] • pytorchmodelcreator.py: - _gen_llvm_training_function_body(): - add support for loss_type = CROSS_ENTROPY • compiledloss.py: - _gen_loss_function: add support for CROSS_ENTROPYLoss - needs to be debugged - need differential implemented * [skip ci] * [skip ci] * [skip ci] • composition.py: - _create_terminal_backprop_learning_components: - support loss_function = CROSS_ENTROPY • combinationfunctions.py: - LinearCombination: add CROSS_ENTROPY as operation * [skip ci] * [skip ci] * [skip ci] * [skip ci] • transferfunctions.py: - ReLU: modified derivative to use infer input from output if provided (needed for BackPropagation) * [skip ci] * [skip ci] * [skip ci] * [skip ci] * - * • transferfunctions.py: - SoftMax.derivative: fixes NOTE: LLVM needs to be modified accordingly • test_transfer.py: - test_transfer_derivative: modify tests to match changes to SoftMax NOTE: LLVM tests don't pass * [skip ci] * [skip ci] * [skip ci] • composition.py - docstring mods re: Autodiff * [skip ci] * Merge branch 'nback' of https://github.com/PrincetonUniversity/PsyNeuLink into nback � Conflicts: � Scripts/Models (Under Development)/Nback/nback.py • composition.py: - run(): try addining Report context for LLVM execution * [skip ci] • composition.py - add Report for compiled mode • compiledloss.py: - CROSS_ENTROPYLoss: _gen_loss_function(): fixed bug, now runs _gen_inject_loss_differential(): dummy copied from MSELoss -- NEEDS TO BE FIXED • transferfunctions.py: - ReLU: added compiled support for derivative using output • test_transfer.py: - test_transfer_derivative_out: test derivatives with output instead of input as arg * Merge branch 'nback' of https://github.com/PrincetonUniversity/PsyNeuLink into nback � Conflicts: � Scripts/Models (Under Development)/Nback/nback.py • composition.py: - run(): try addining Report context for LLVM execution * [skip ci] * [skip ci] * [skip ci] * [skip ci] • Merge branch 'nback' of https://github.com/PrincetonUniversity/PsyNeuLink into nback • composition.py: - docstring mods regarding Autodiff learning * [skip ci] * • composition.py - more docstrings re: Autodiff * [skip ci] • composition.py - table for learning execution modes * [skip ci] • llvm/__init__.py - ExecuteMode: add PyTorch as synonym for Python • autodiffcomposition.py - docstrring refactoring * [skip ci] * [skip ci] • composition.py, autodiffcomposition.py - docstring mods * [skip ci] • composition.py, autodiffcomposition.py - docstring mods * [skip ci] • composition.py, autodiffcomposition.py - docstring mods * [skip ci] * [skip ci] • test_transfer.py: - get rid of duplicative test * Merge branch 'devel' of https://github.com/PrincetonUniversity/PsyNeuLink into devel Conflicts: .github/actions/install-pnl/action.yml .github/actions/on-branch/action.yml .github/workflows/pnl-ci-docs.yml .github/workflows/pnl-ci.yml .github/workflows/test-release.yml Scripts/Models (Under Development)/N-back.py * Merge branch 'devel' of https://github.com/PrincetonUniversity/PsyNeuLink into devel Conflicts: .github/actions/install-pnl/action.yml .github/actions/on-branch/action.yml .github/workflows/pnl-ci-docs.yml .github/workflows/pnl-ci.yml .github/workflows/test-release.yml Scripts/Models (Under Development)/N-back.py * • test_learning.py: - test_xor_training_identicalness_standard_composition_vs_PyTorch_vs_LLVM: replaces test_xor_training_identicalness_standard_composition_vs_Autodiff * • learningfunctions.py - BackPropagation: - fix bug in which derivative for default loss (MSE) was computed using L0 - add explicit specification for L0 loss • composition.py: - _create_terminal_backprop_learning_components: - add explicit assignment of output_port[SUM] for L0 loss • test_learning.py: - test_multilayer: - fix bug in which SSE was assigned as loss, but oputput_port[MSE] was used for objective_mechanism - replace with explicit L0 loss and ouput_port[SUM] for objective_mechanism * [skip ci] • learningfunctions.py - _create_non_terminal_backprop_learning_components: - fixed bug in which loss function for hidden layers was set to MSE rather than simple L0 * [skip ci] • All tests pass * [skip ci] • test_learning.py: - test_multilayer_truth(): parameterize test with expected results * [skip ci] • test_learning.py: - test_multilayer_truth(): test for L0, SSE and MSE * [skip ci] • All tests pass * [skip ci] • All tests pass * [skip ci] * [skip ci] • keywords.py - add Loss enum • llvm.rst - add ExecutionMode • Project - replace MSE, SSE, L0, L1, CROSS_ENTROPY, KL_DIV, NLL and POISSON_NLL with Loss enum members * [skip ci] • composition.py: - run(): add warning for use of PyTorch with Composition * [skip ci] * [skip ci] • composition.py: - run(): commented out warning, as can't distinguish ExecutionMode.PyTorch from ExecutionMode.Python * - * • test_learning.py - clean up test_xor_training_identicalness_standard_composition_vs_PyTorch_and_LLVM * • test_learning.py - clean up test_xor_training_identicalness_standard_composition_vs_PyTorch_and_LLVM Co-authored-by: SamKG <[email protected]> Co-authored-by: Katherine Mantel <[email protected]> Co-authored-by: jdcpni <pniintel55>
- Loading branch information
3 people
authored
Nov 19, 2022
1 parent
89d2561
commit fc13a63