feat(multitask): default tensorboard_clean=True when TensorBoard logging is enabled

docs/model_selection/spotoptim_intro.qmd

@@ -167,7 +167,10 @@ results, optimizer = spotoptim_search_forecaster(
 cfg = ConfigEntsoe(...)
 cfg.tensorboard_log = True
 cfg.tensorboard_path = "runs/my_tuning_run"   # optional; defaults to runs/<stamp>
-cfg.tensorboard_clean = False                 # True wipes old runs/ subdirs first
+# tensorboard_clean defaults to True when logging is enabled: the log
+# directory is wiped at search start so each run gets a fresh dashboard.
+# Set cfg.tensorboard_clean = False to keep accumulating event files —
+# do that (or use per-task paths) when several tasks share one directory.
 ```
 
 Then watch the run live:

pyproject.toml

@@ -33,7 +33,7 @@ dependencies = [
     "scikit-learn>=1.8.0",
     "shap>=0.49.1",
     "spotforecast2-safe>=18.0.0,<19",
-    "spotoptim>=0.12.8",
+    "spotoptim>=0.12.9",
     "tqdm>=4.67.2",
 ]
 

src/spotforecast2/multitask/strategies.py

@@ -222,9 +222,14 @@ def prepare_forecaster(
         SpotOptim constructor and the ongoing search can be watched with
         ``tensorboard --logdir <path>`` (defaults to ``runs/``).  Set them
         as plain attributes, e.g. ``cfg.tensorboard_log = True`` — no
-        config-class change is required.  Live per-eval scalars under
-        parallel tuning (``n_jobs_spotoptim != 1``) require
-        ``spotoptim >= 0.12.8``.
+        config-class change is required.  When logging is enabled and
+        ``tensorboard_clean`` is unset, it defaults to ``True`` so every
+        run starts with a fresh dashboard (the configured log directory
+        is wiped at search start); set ``cfg.tensorboard_clean = False``
+        to keep accumulating event files instead — do that, or give each
+        task its own ``tensorboard_path``, when several tasks share one
+        log directory.  Live per-eval scalars under parallel tuning
+        (``n_jobs_spotoptim != 1``) require ``spotoptim >= 0.12.8``.
 
         Args:
             task: A `BaseTask` (or compatible) instance that supplies ``cv_ts``,
@@ -347,6 +352,11 @@ def prepare_forecaster(
             for key in ("tensorboard_log", "tensorboard_path", "tensorboard_clean")
             if getattr(task.config, key, None) is not None
         }
+        if tb_kwargs.get("tensorboard_log") and "tensorboard_clean" not in tb_kwargs:
+            # Fresh dashboard per run: wipe the configured log directory before
+            # the search starts (spotoptim >= 0.12.9 cleans tensorboard_path
+            # itself). Opt out with ``cfg.tensorboard_clean = False``.
+            tb_kwargs["tensorboard_clean"] = True
         if tb_kwargs:
             kwargs_spotoptim.update(tb_kwargs)
             task.logger.info("  SpotOptim TensorBoard: %s", tb_kwargs)

tests/test_multitask_strategies.py

@@ -120,12 +120,69 @@ def fake_search_forecaster(*args, **kwargs):
     ks = captured["kwargs_spotoptim"]
     assert ks["tensorboard_log"] is True
     assert ks["tensorboard_path"] == "runs/test-tb"
-    # Unset attribute must be skipped so SpotOptim's default stays in charge.
-    assert "tensorboard_clean" not in ks
+    # With logging enabled and tensorboard_clean unset, the strategy
+    # defaults it to True so every run starts with a fresh dashboard.
+    assert ks["tensorboard_clean"] is True
     assert isinstance(tuned, _FakeForecaster)
     assert tuned.params == {"alpha": 1.0}
 
 
+def test_spotoptim_strategy_respects_explicit_tensorboard_clean_false(monkeypatch):
+    """An explicit tensorboard_clean=False must not be overridden."""
+    import pandas as pd
+
+    import spotforecast2.model_selection as ms
+
+    captured = {}
+
+    def fake_search_forecaster(*args, **kwargs):
+        captured["kwargs_spotoptim"] = kwargs.get("kwargs_spotoptim")
+        results = pd.DataFrame({"params": [{"alpha": 1.0}], "lags": [[1, 2]]})
+        return results, object()
+
+    monkeypatch.setattr(ms, "spotoptim_search_forecaster", fake_search_forecaster)
+
+    task = _make_fake_task(
+        tensorboard_log=True,
+        tensorboard_path="runs/test-tb",
+        tensorboard_clean=False,
+    )
+    SpotOptimStrategy(search_space={"alpha": (0.1, 1.0)}).prepare_forecaster(
+        task, "A", _FakeForecaster(), y_train=None
+    )
+
+    ks = captured["kwargs_spotoptim"]
+    assert ks["tensorboard_log"] is True
+    assert ks["tensorboard_clean"] is False
+
+
+def test_spotoptim_strategy_no_clean_default_without_logging(monkeypatch):
+    """tensorboard_clean is only defaulted when tensorboard_log is enabled."""
+    import pandas as pd
+
+    import spotforecast2.model_selection as ms
+
+    captured = {}
+
+    def fake_search_forecaster(*args, **kwargs):
+        captured["kwargs_spotoptim"] = kwargs.get("kwargs_spotoptim")
+        results = pd.DataFrame({"params": [{"alpha": 1.0}], "lags": [[1, 2]]})
+        return results, object()
+
+    monkeypatch.setattr(ms, "spotoptim_search_forecaster", fake_search_forecaster)
+
+    # tensorboard_log=False is forwarded (it is not None) but must not
+    # trigger the clean default.
+    task = _make_fake_task(tensorboard_log=False)
+    SpotOptimStrategy(search_space={"alpha": (0.1, 1.0)}).prepare_forecaster(
+        task, "A", _FakeForecaster(), y_train=None
+    )
+
+    ks = captured["kwargs_spotoptim"]
+    assert ks["tensorboard_log"] is False
+    assert "tensorboard_clean" not in ks
+
+
 def test_spotoptim_strategy_no_tensorboard_attrs_no_kwargs(monkeypatch):
     """Without tensorboard config attrs, no tensorboard keys are forwarded."""
     import pandas as pd