working on run log working

lenskit · Dec 17, 2024 · fd7445d · fd7445d
1 parent 63aa45b
commit fd7445d
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 9 deletions.
diff --git a/codex/runlog.py b/codex/runlog.py
@@ -88,8 +88,8 @@ class CodexTask(Task):
     lenskit_version: str = lenskit.__version__  # type: ignore
     tags: list[str] = Field(default_factory=list)
 
-    model: str | None = None
-    model_config: str | dict[str, JsonValue] | None = None
+    score_model: str | None = None
+    score_model_config: str | dict[str, JsonValue] | None = None
 
     cpu_power: float | None = None
     gpu_power: float | None = None

diff --git a/codex/splitting/crossfold.py b/codex/splitting/crossfold.py
@@ -29,7 +29,7 @@ def crossfold_ratings(db: DuckDBPyConnection, cross: CrossfoldSpec, hold: Holdou
         db.execute(
             """
             INSERT INTO test_alloc (partition, user_id, item_id)
-            SELECT ?, user, item FROM test_df
+            SELECT ?, user_id, item_id FROM test_df
             """,
             [i],
         )
diff --git a/codex/tools/generate.py b/codex/tools/generate.py
@@ -96,7 +96,7 @@ def generate(
         connect_cluster() as cluster,
         ResultDB(db, store_predictions=predict) as results,
         CodexTask(
-            label=f"generate-{model}", tags=["generate"], model=model, model_config=cfg
+            label=f"generate-{model}", tags=["generate"], score_model=model, score_model_config=cfg
         ) as root_task,
     ):
         log = _log.bind(task_id=root_task.task_id)
@@ -108,10 +108,10 @@ def generate(
                 label=f"generate-{model}/train",
                 tags=["train"],
                 reset_hwm=True,
-                model=model,
-                model_config=cfg,
+                score_model=model,
+                score_model_config=cfg,
             ) as task:
-                trainable.train(data.train_data(db))
+                trainable.train(data.train_data(ratings_db))
 
             plog.debug("run record: %s", task.model_dump_json(indent=2))
             plog.info(
@@ -129,7 +129,11 @@ def generate(
                     plog.error("no test data found")
 
             with CodexTask(
-                label=f"generate-{model}/recommend", tags=["recommend"], reset_hwm=True
+                label=f"generate-{model}/recommend",
+                tags=["recommend"],
+                reset_hwm=True,
+                score_model=model,
+                score_model_config=cfg,
             ) as task:
                 for result in run_recommender(
                     trainable, test, list_length, predict, cluster=cluster
@@ -157,7 +161,7 @@ def fixed_test_sets(test: Path, train: list[Path]) -> Generator[tuple[int, Train
 def crossfold_test_sets(
     assign: Path, ratings: Path, parts: str
 ) -> Generator[tuple[int, TrainTestData]]:
-    with connect(fspath(assign)) as db:
+    with connect(fspath(assign), read_only=True) as db:
         db.execute(
             "SELECT DISTINCT partition FROM test_alloc",
         )