Draft
Conversation
…are now clearly defined and typed
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
…nly-models Added `is_choice_only` field for `Config` class and updated tests
* initial poisson race implementation * add to public api * add basic test suite for poisson race * fix formatting * minimal tutorial notebook * Update src/hssm/likelihoods/analytical.py fix shape bounds Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * Update tests/test_likelihoods_poisson_race.py add pythonic syntax Co-authored-by: Carlos Paniagua <cpaniaguam@gmail.com> * fix formatting for test_likelihoods_poisson_race * build bound dicts with comprehension Co-authored-by: Carlos Paniagua <cpaniaguam@gmail.com> * fix logp_poisson_race docstring * fix docstring, add consitent safe usage * fix formatting * Update src/hssm/likelihoods/analytical.py fix doctring typo Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * fix export in init * fix docstring * remove smooth uniform for poisson race * fix parameter bounds * update tutorial notebook * address remaining comments --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Carlos Paniagua <cpaniaguam@gmail.com> Co-authored-by: Alexander <alexanderfengler@gmx.de>
…ve-load 898 use temp folder in test save load
* small bugfix en route, and add tutorial on how to use Bayesflow lre * add tutorial to docs correctly * Update pyproject.toml Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
…e necessary options
…ce-only Update HSSM class for choice-only models
* fix: flaky test in test_data_validator due to valid choice in random range The test sets choices=[0, 1, 2] but picks a supposedly invalid response from range(2, 100). When random.choice returns 2, the response is actually valid and no ValueError is raised, causing a flaky failure. Changed to range(3, 100) to guarantee an invalid choice. Made-with: Cursor * fix: make data validator test deterministic and robust Use a deterministic invalid response value and explicit response-column indexing to remove flaky randomness and schema-order brittleness in test_data_validator. Made-with: Cursor
* fix: safe negative-RT handling and JAX backend for RDM3 - Replace element-wise jax_check_parameters for rt > 0 with jnp.where + LOGP_LB floor in logp_rdm3. This clamps negative decision times to a small epsilon before passing to the PDF/CDF, preventing NaN in the likelihood and providing a finite gradient floor per trial. - Set backend to "jax" and add default HalfNormal priors for A and t in both racing_diffusion_3_config and rdm3_config. Made-with: Cursor * chore: remove unused rdm3_config.py This file is dead code — the dynamic config loader in modelconfig/__init__.py derives module names from the model name "racing_diffusion_3", so only racing_diffusion_3_config.py is ever loaded. rdm3_config.py was never imported or called. Made-with: Cursor * fix: plumb rdm3 epsilon and harden negative-RT floor Pass logp_rdm3 epsilon through JAX RDM internals to keep clamping behavior consistent and preserve LOGP_LB dtype under JAX. Add a regression test ensuring rt<=t returns LOGP_LB and stays finite for RDM3. Made-with: Cursor
* first hmm example * Address PR #923 review comments on HMM-DDM notebook - Update title to be more informative for broader audience - Remove broken link to regime-switching-bayesian repo - Add Forward Algorithm likelihood explanation blurb Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * some improvements to the notebook * add new tutorials to mkdocs --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* Update version for release * Add prepare-release skill and GitHub workflow - Claude Code skill (.claude/skills/prepare-release/SKILL.md) that automates the pre-release checklist: changelog update, announcement banner update, local docs build verification, and draft GitHub release creation - GitHub Actions workflow (.github/workflows/prepare-release.yml) as an alternative agentic approach using claude-code-action with workflow_dispatch Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Prepare v0.3.0 release: changelog, banner, docs fixes, and skill update - Add v0.3.0 changelog entry to docs/changelog.md - Update announcement banner to v0.3.0 - Fix mkdocs.yml: rename `import` to `inventories` for mkdocstrings-python 2.0 - Add poisson_race.ipynb to mkdocs execute_ignore list - Add notebook check workflow step to prepare-release skill Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Fix notebooks for v0.3.0 release verification - bayesflow_lre_integration: rename classifier_network → inference_network for bayesflow dev branch API compatibility - initial_values: add cores=1, chains=1 to fix multiprocessing EOFError - plotting: fix missing comma after sampler="numpyro" - poisson_race: fix kernel spec hssm-dev → python3 - CI: add workshop_1 and workshop_2 notebooks to skip list - pyproject.toml: move bayesflow+keras to notebook dep group, remove bayesflow optional extra (not yet released on PyPI) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Add CLAUDE.md with project context for Claude Code Provides project-scoped context: structure, tooling, dependency management, common commands, key API patterns, CI workflows, and known issues for better Claude Code assistance. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * slight improvements to tests * bump lanfactory and ssm-simulators versions * bump bayesflow version --------- Co-authored-by: Alexander <alexanderfengler@gmx.de> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* small improvements to claude * some mypy formatting in the process of creating pre-commit harmony across projects * Track uv.lock in version control Remove uv.lock from .gitignore and commit the lockfile for reproducible dev/CI environments. Raise check-added-large-files limit to 1000 KB. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * Raise check-added-large-files limit to ~99.9 MB Match GitHub's own 100 MB hard limit rather than imposing an artificial lower threshold. Notebooks and lock files should not be blocked. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
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
5 participants
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.
Create
HSSMBase. Notably, starting from a copy of HSSM, we turn it into an abstract class and specifies the_make_model_distributionas abstract (needs to be specified in subclasses).