[wenh06]

Summary

• tutorial-visualization: merge a two-seed experiment into one mean curve with error bounds (find_log_files → aggregate_results_from_json_log_with_custom_axis → plot_mean_curve_with_error_bounds), export to PDF, and introduce the interactive Panel dashboard for notebook use. The x-axis guidance follows the cumulative-communications metric dissected in the first-experiment tutorial.
• tutorial-parallel-reproducible: switch to the threads executor with one config line, then demonstrate the reproducibility contract end to end — with rng_mode="per_client" the sequential and threads backends produce bitwise identical final weights (torch.equal), while the global mode keeps participation and aggregation backend-independent but may interleave local trajectories. The script prints the contrast honestly (global-mode weights matched on the development machine; the page marks that as luck, not a guarantee).
• test-files/tutorial_first_experiment.py: every run now takes a fresh model from the candidate_models property. The server trains the model it is given in place (self.model = model, no copy), so sharing one module across runs would warm-start every run after the first and contaminate the comparison. The documented table is unchanged (all algorithms saturate at this task's accuracy ceiling), and the run() docstring teaching the rule is shown on the page via literalinclude.
• test/test_tutorial_scripts.py: two new behavioral smoke tests — curve merge with non-empty curves and PDF export inside a temporary working directory; bitwise sequential-vs-threads equality with per-client RNG plus global-mode communication equality.
• zh_CN translations for the new and updated pages.

Validation

• pytest test/test_tutorial_scripts.py: 7 passed (~22 s, real runs included).
• English and zh_CN docs builds: zero warnings.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.79%. Comparing base (b31a5db) to head (d155122).

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #59      +/-   ##
==========================================
+ Coverage   83.63%   83.79%   +0.15%     
==========================================
  Files          55       55              
  Lines        5640     5640              
==========================================
+ Hits         4717     4726       +9     
+ Misses        923      914       -9     

see 4 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.