Code clean and workflow update

shared/workflows

@@ -1 +1 @@
-Subproject commit 873b0e6bc80052921fa2061ef9fbcd4e1c4f057e
+Subproject commit ec5c0d9ab888cc5e8fcfde8985f2e040ed170b71

src/components/quizbowl/plotting.py

@@ -273,64 +273,6 @@ def update_tossup_plot(highlighted_index: int, state: str) -> pd.DataFrame:
         return pd.DataFrame()
 
 
-# %%
-
-
-def create_df_entry(run_indices: list[int], run_outputs: list[dict]) -> dict:
-    """Create a dataframe entry from a list of model outputs."""
-    chosen_idx = None
-    earliest_ok_idx = None
-    is_correct = None
-    for i, o in enumerate(run_outputs):
-        if chosen_idx is None and o["buzz"]:
-            chosen_idx = run_indices[o["position"] - 1] + 1
-            is_correct = o["score"]
-        if earliest_ok_idx is None and o["score"]:
-            earliest_ok_idx = run_indices[o["position"] - 1] + 1
-    if is_correct is None:
-        is_correct = False
-
-    # if buzz is not the last index, correct scores 10, incorrect scores -5
-    # if buzz is the final index, correct scores 5, incorrect scores 0
-
-    if chosen_idx == -1:
-        tossup_score = 0
-    elif chosen_idx == run_indices[-1] + 1:
-        tossup_score = 5 if is_correct else 0
-    else:
-        tossup_score = 10 if is_correct else -5
-
-    gap = None if (chosen_idx is None or earliest_ok_idx is None) else chosen_idx - earliest_ok_idx
-    if earliest_ok_idx is None:
-        cls = "hopeless"
-    elif chosen_idx is None:
-        cls = "never-buzzed"  # Opportunity missed to score
-    elif chosen_idx == earliest_ok_idx:
-        cls = "best-buzz"  # Perfect timing
-    elif chosen_idx > earliest_ok_idx:
-        cls = "late-buzz"  # Opportunity missed to buzz earlier
-    elif chosen_idx < earliest_ok_idx:
-        cls = "premature"  # Opportunity missed to score
-
-    return {
-        "chosen_idx": chosen_idx,
-        "earliest_ok_idx": earliest_ok_idx,
-        "gap": gap,
-        "cls": cls,
-        "tossup_score": tossup_score,
-        "is_correct": int(is_correct),
-    }
-
-
-def prepare_tossup_results_df(run_indices: list[list[int]], model_outputs: list[list[dict]]) -> pd.DataFrame:
-    """Create a dataframe from a list of model outputs."""
-    records = []
-    for indices, outputs in zip(run_indices, model_outputs):
-        entry = create_df_entry(indices, outputs)
-        records.append(entry)
-    return pd.DataFrame.from_records(records)
-
-
 def create_tossup_eval_table(df: pd.DataFrame) -> pd.DataFrame:
     """Create a table from a dataframe."""
     # Prepare a dataframe of aggregated metrics:

src/components/quizbowl/tossup.py

@@ -14,6 +14,7 @@ from components.typed_dicts import TossupInterfaceDefaults, TossupPipelineStateD
 from display.formatting import styled_error
 from shared.workflows import factory
 from shared.workflows.metrics import evaluate_prediction
+from shared.workflows.metrics.qb_metrics import prepare_tossup_results_df
 from shared.workflows.qb_agents import QuizBowlTossupAgent, TossupResult
 from submission import submit
 
@@ -23,7 +24,6 @@ from .plotting import (
     create_tossup_eval_dashboard,
     create_tossup_eval_table,
     create_tossup_html,
-    prepare_tossup_results_df,
 )
 from .utils import create_error_message
 from .validation import UserInputWorkflowValidator
@@ -336,7 +336,7 @@ class TossupInterface:
             for example in progress.tqdm(self.ds, desc="Evaluating tossup questions"):
                 run_outputs = self.get_agent_outputs(example, pipeline_state, early_stop=True)
                 model_outputs.append(run_outputs)
-            eval_df = prepare_tossup_results_df(self.ds["run_indices"], model_outputs)
+            eval_df = prepare_tossup_results_df(model_outputs, self.ds["run_indices"])
             plot_data = create_tossup_eval_dashboard(self.ds["run_indices"], eval_df)
             output_df = create_tossup_eval_table(eval_df)
             return (