Skip to content

CHANGELOG.md

Latest commit

 

History

History
55 lines (41 loc) · 2.75 KB

CHANGELOG.md

File metadata and controls

55 lines (41 loc) · 2.75 KB

CHANGELOG

All notable development changes for T000002-bart are documented here.

[1.1.3] - 2026-03-10

Changed

  • Updated src/run_trial.py so every participant-visible outcome and feedback phase emits set_trial_context(...), not just the response window.
  • Regenerated task-plot artifacts under references/task_plot_* and task_flow.png using the updated task-plot skill.
  • Updated README.md to embed the generated task-flow figure.
  • Normalized config/config.yaml text encoding header.

Fixed

  • Restored audit visibility for BART pop, cash, timeout, and feedback screens in downstream task-plot inference.
  • Updated the generated workflow figure to use participant-visible sample stimuli rather than abstract placeholders.

Added

  • Added comparison figure exports task_flow_gemini3.png and task_flow_opus4.6.png.

[1.1.2] - 2026-02-23

Changed

  • Removed redundant _TRIAL_COUNTER and _next_trial_id() manually defined boilerplate in favor of native psyflow.next_trial_id synchronization.
  • Removed redundant duration resolver (_deadline_s()) locally, since duration array parsing is now safely handled natively inside set_trial_context(...).
  • Refactored src/run_trial.py to maintain decoupled generalized tracking from psyflow.

[1.1.1] - 2026-02-18

Changed

  • Refactored responder context phase names in src/run_trial.py to task-specific labels (removed generic MID-style phase naming).
  • Updated stage comments in src/run_trial.py to phase-aligned labels for cleaner auditability.
  • Updated README.md to keep runtime phase documentation aligned with the implemented trial context phases.

Fixed

  • Removed legacy stage comment patterns (cue/anticipation/target/feedback) from trial runtime code.

[1.1.0] - 2026-02-17

Added

  • Added mode-aware main.py flow for human, qa, and sim modes.
  • Added split runtime configs: config/config.yaml, config/config_qa.yaml, config/config_scripted_sim.yaml, and config/config_sampler_sim.yaml.
  • Added task-local responder scaffold in esponders/task_sampler.py.
  • Added outputs/.gitkeep and standardized output folders for human/qa/sim runs.

Changed

  • Aligned trigger config to structured schema ( riggers.map, riggers.driver, riggers.policy, riggers.timing).
  • Aligned src/run_trial.py to set responder trial context via set_trial_context(...) at response windows.
  • Added/updated askbeacon.yaml with contracts.psyflow_taps: v0.1.0.
  • Updated .gitignore to match standardized task artifacts and output handling.

Verified

  • psyflow-validate passes all contract checks (including artifacts).
  • psyflow-qa --config config/config_qa.yaml --no-maturity-update passes.
  • python main.py sim --config config/config_scripted_sim.yaml runs successfully and writes sim outputs.