Update pytest requirement from <9.0,>=8.0 to >=8.0,<10.0#3
Closed
dependabot[bot] wants to merge 26 commits into
Closed
Update pytest requirement from <9.0,>=8.0 to >=8.0,<10.0#3dependabot[bot] wants to merge 26 commits into
dependabot[bot] wants to merge 26 commits into
Conversation
Establishes a layered structure for comparing predictive coding and backpropagation on a synthetic binary task, with reproducible experiments, config management, and testing. Documents architecture, module boundaries, engineering standards, and extension points for future neuron adaptation policies. Lays groundwork for research and educational use while prioritizing clarity and comparability.
Introduces a new predictive-coding variant modeling circadian plasticity, including chemical buildup per neuron, adaptive suppression of learning, and sleep-driven structural changes (splitting high-use and pruning low-use neurons). Updates documentation and tests to support and illustrate the new model.
Introduces in-depth multi-run experiment support, enabling aggregate statistics across multiple seeds and noise levels. Extends experiment runner and CLI to include circadian predictive coding model with sleep dynamics, split/prune events, and end-state hidden dimension metrics. Updates documentation and tests to reflect the expanded comparison and reporting capabilities.
Introduces a synthetic, deterministic ResNet-50 speed and accuracy benchmark workflow comparing backpropagation, predictive coding, and circadian predictive coding heads. Provides new CLI, dataloaders, and torch/torchvision optional dependencies for fast, reproducible metrics without external datasets. Enables measurement of training and inference speed, convergence to target accuracy, and circadian split/prune dynamics, supporting research on local learning and network plasticity. Updates documentation and adds targeted tests. Relates to enabling rapid, reproducible performance evaluation for biologically inspired learning rules.
Enables fine-tuning of circadian predictive coding model by exposing new head parameters through CLI and benchmark config. Improves flexibility for research and experimentation with split/prune thresholds, plasticity, and sleep behavior. Documents recommended CUDA setup and tuned example for GPU runs.
Introduces adjustable difficulty and noise parameters for synthetic image tasks to strengthen ResNet-50 benchmarking. Enforces comparison across traditional backprop, predictive coding, and circadian predictive coding models. Improves result reporting with per-model comparison metrics and expands unit test coverage for new settings.
Provides detailed JSON records of hyperparameter tuning trials on the hardest dataset variant, including comparative metrics for backpropagation, predictive coding, and circadian models. Enables future analysis of model performance, parameter efficiency, and optimization strategies.
Introduces automated parameter sweeps for three model variants on the hardest synthetic vision benchmark, saving top-10 and Pareto front results. Enables comparison of accuracy, training speed, inference speed, and balanced efficiency across model families. Provides global winners and exports results for downstream analysis.
Introduces adaptive split/prune thresholds, sleep triggers based on energy plateau and chemical variance, weight-norm-aware rankings, gradual pruning with decay, replay consolidation, and homeostatic downscaling. Enables external neuron adaptation policy proposals for structural changes, enhances configurability and robustness, and expands unit test coverage for key adaptive behaviors.
Enables fine-grained control over circadian predictive coding parameters via CLI, including adaptive thresholds, sleep triggers, and replay configuration. Passes full circadian config into experiment runner for more flexible experimentation and testing. Expands test coverage for adaptive circadian configurations.
Updates neuron split logic to ensure function preservation by making outgoing weights sum to the original and duplicates incoming connections. Adopts binary cross-entropy-compatible gradients for binary predictive coding, improving consistency with BCE loss. Adds targeted unit tests to verify function-preserving splits for both NumPy and Torch implementations, supporting stable and reliable network adaptation during sleep events.
Introduces a command-line option and config parameter to select random or ImageNet-pretrained weights for ResNet-50 backbone initialization, improving reproducibility and benchmarking flexibility. Updates documentation and validation to reflect the new feature.
Introduces dual-timescale chemical dynamics, gradient importance tracking, and per-neuron cooldown for split/prune operations to improve adaptivity and stability. Adds split/prune hysteresis, prioritized and class-balanced replay consolidation, and targeted homeostatic norm matching for more robust plasticity. Updates configuration, CLI, and unit tests for full coverage of new mechanisms.
Enables systematic evaluation of ResNet-50 circadian policy hyperparameters by introducing a focused sweep script and storing resulting metrics. Supports easier comparison of accuracy, speed, and balanced scores to inform future model tuning and optimization directions.
Switches candidate parameter representation from tuples to dictionaries, enabling flexible and explicit configuration of circadian hyperparameters. Adds new adaptive and homeostatic options to support more granular experimentation and future parameter expansion. Improves readability and maintainability of sweep logic.
Introduces optional saturating chemical buildup to keep neuron chemical values bounded and adds adaptive per-neuron plasticity sensitivity based on age and importance. Enhances model flexibility for stability-plasticity trade-offs and improves configurability, including robust input validation and CLI support. Includes tests to verify bounded chemical accumulation and importance-driven plasticity adjustment.
Introduces configurable sleep-phase scheduling to control structural changes during training, including warm-up, split-first, and prune-late phases with topology caps. Implements minimum-age gating for pruning to reduce early instability and improve robustness. Enhances configuration and validation for new controls.
Introduces configurable adaptive sleep triggering for circadian benchmarks, allowing sleep events based on plateau detection and chemical variance. Implements state snapshot/restore and sleep rollback to protect against accuracy drops. Updates configs, validation, reporting, and tests for new sleep parameters and rollback tracking.
Sets pruning threshold per sleep to zero to ensure snapshot restoration logic is tested without interference from pruning. Helps isolate the effects of structural changes in the head.
Introduces support for running benchmark sweeps across multiple random seeds, aggregating results for improved reliability and reproducibility. Adds flexible configuration for evaluation batch counts and allows sleep rollback to use either accuracy or cross-entropy as the metric, addressing robustness in circadian sleep logic. Extends reporting to include final cross-entropy and energy, and updates CLI, validation, and tests to cover new options.
Updates configuration defaults for circadian and predictive parameters to reflect current best-performing settings, enabling adaptive triggers and rollback by default. Enhances validation logic to ensure hidden dimension bounds are consistent and adds related unit tests. Clarifies output formatting and argument parsing for better usability.
Enables benchmarking on CIFAR-10 and CIFAR-100 using torchvision, in addition to synthetic data. Introduces dataset selection CLI arguments, configurable data augmentation, subset sampling, and validation for dataset-specific constraints. Refactors dataloader construction for extensibility and improves dataset summary reporting. Includes tests for new validation logic.
Introduces a script to automate running ResNet benchmarks across multiple seeds, aggregating results, and exporting both per-seed and summary statistics to JSON and CSV files. Updates documentation with usage instructions and configures .gitignore to exclude benchmark result files. Enables more robust and reproducible benchmark comparisons for different training algorithms.
Introduces MIT license, governance, code of conduct, security, support, and citation files to support open-source collaboration. Adds GitHub issue and PR templates, CI workflow, and dependabot config for reproducibility and maintenance. Repositions repository messaging, documentation, and architecture to emphasize Circadian Predictive Coding as the primary algorithm, maintaining backprop and predictive coding for rigorous comparison. Improves contribution workflow, quality gates, and benchmark reproducibility.
Updates the requirements on [pytest](https://github.com/pytest-dev/pytest) to permit the latest version. - [Release notes](https://github.com/pytest-dev/pytest/releases) - [Changelog](https://github.com/pytest-dev/pytest/blob/main/CHANGELOG.rst) - [Commits](pytest-dev/pytest@8.0.0...9.0.2) --- updated-dependencies: - dependency-name: pytest dependency-version: 9.0.2 dependency-type: direct:production ... Signed-off-by: dependabot[bot] <support@github.com>
dependabot
Bot
added
dependencies
Author
|
OK, I won't notify you again about this release, but will get in touch when a new version is available. If you'd rather skip all updates until the next major or minor version, let me know by commenting If you change your mind, just re-open this PR and I'll resolve any conflicts on it. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Updates the requirements on pytest to permit the latest version.
Release notes
Sourced from pytest's releases.
Commits
3d10b51Prepare release version 9.0.2188750bMerge pull request #14030 from pytest-dev/patchback/backports/9.0.x/1e4b01d1f...b7d7befMerge pull request #14014 from bluetech/compat-notebd08e85Merge pull request #14013 from pytest-dev/patchback/backports/9.0.x/922b60377...bc78386Add CLI options reference documentation (#13930)5a4e398Fix docs typo (#14005) (#14008)d7ae6dfMerge pull request #14006 from pytest-dev/maintenance/update-plugin-list-tmpl...556f6a2pre-commit: fix rst-lint after new release (#13999) (#14001)c60fbe6Fix quadratic-time behavior when handlingunittestsubtests in Python 3.10 ...73d9b01Merge pull request #13995 from nicoddemus/patchback/backports/9.0.x/1b5200c0f...Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting
@dependabot rebase.Dependabot commands and options
You can trigger Dependabot actions by commenting on this PR:
@dependabot rebasewill rebase this PR@dependabot recreatewill recreate this PR, overwriting any edits that have been made to it@dependabot show <dependency name> ignore conditionswill show all of the ignore conditions of the specified dependency@dependabot ignore this major versionwill close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)@dependabot ignore this minor versionwill close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)@dependabot ignore this dependencywill close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)