Update space

data/results.json

@@ -0,0 +1,41 @@
+[
+  {
+    "model_name": "demo-model-1",
+    "agent_name": "TemporalAgent-A",
+    "agent_type": "single-LLM",
+    "base_model": "demo-base-1",
+    "T1_acc": 71.2,
+    "T2_acc": 64.5,
+    "T3_acc": 69.8,
+    "T4_acc": 62.3,
+    "T2_MAE": 0.41,
+    "T4_sMAPE": 0.22,
+    "Retail_T3_acc": 70.1
+  },
+  {
+    "model_name": "demo-model-2",
+    "agent_name": "TemporalAgent-B",
+    "agent_type": "general agent",
+    "base_model": "demo-base-2",
+    "T1_acc": 75.4,
+    "T2_acc": 66.7,
+    "T3_acc": 72.9,
+    "T4_acc": 65.8,
+    "T2_MAE": 0.38,
+    "T4_sMAPE": 0.20,
+    "MIMIC_T3_acc": 71.6
+  },
+  {
+    "model_name": "demo-model-3",
+    "agent_name": "TemporalAgent-C",
+    "agent_type": "time-series-specific agent",
+    "base_model": "demo-base-3",
+    "T1_acc": 69.9,
+    "T2_acc": 63.2,
+    "T3_acc": 68.4,
+    "T4_acc": 61.7,
+    "T2_MAE": 0.44,
+    "T4_sMAPE": 0.24,
+    "PSML_T3_acc": 67.9
+  }
+]

src/leaderboard/load_results.py

@@ -0,0 +1,118 @@
+from __future__ import annotations
+
+import json
+import math
+import os
+from typing import Iterable
+
+import pandas as pd
+
+from src.leaderboard.schema import SCHEMA
+
+
+class ResultsValidationError(ValueError):
+    pass
+
+
+def _is_number(value) -> bool:
+    if not isinstance(value, (int, float)) or isinstance(value, bool):
+        return False
+    return math.isfinite(float(value))
+
+
+def _load_json_records(path: str) -> list[dict]:
+    with open(path, "r") as fp:
+        data = json.load(fp)
+
+    if isinstance(data, list):
+        return data
+    if isinstance(data, dict) and "records" in data and isinstance(data["records"], list):
+        return data["records"]
+    raise ResultsValidationError(
+        "JSON must be a list of records or an object with a 'records' list."
+    )
+
+
+def _load_csv_records(path: str) -> list[dict]:
+    df = pd.read_csv(path)
+    return df.to_dict(orient="records")
+
+
+def load_records(path: str) -> list[dict]:
+    if not os.path.exists(path):
+        raise ResultsValidationError(f"Results file not found: {path}")
+
+    _, ext = os.path.splitext(path)
+    ext = ext.lower()
+    if ext == ".json":
+        return _load_json_records(path)
+    if ext == ".csv":
+        return _load_csv_records(path)
+
+    raise ResultsValidationError("Unsupported file type. Use .json or .csv")
+
+
+def validate_records(records: Iterable[dict]) -> None:
+    records = list(records)
+    if not records:
+        raise ResultsValidationError("Results file is empty.")
+
+    for idx, record in enumerate(records):
+        if not isinstance(record, dict):
+            raise ResultsValidationError(f"Record {idx} is not an object.")
+
+        missing = [f for f in SCHEMA.identity_fields if f not in record]
+        if missing:
+            raise ResultsValidationError(f"Record {idx} is missing fields: {missing}")
+
+        for field in SCHEMA.identity_fields:
+            if not isinstance(record[field], str):
+                raise ResultsValidationError(
+                    f"Record {idx} field '{field}' must be a string."
+                )
+
+        missing_metrics = [m for m in SCHEMA.required_metrics if m not in record]
+        if missing_metrics:
+            raise ResultsValidationError(
+                f"Record {idx} is missing required metrics: {missing_metrics}"
+            )
+
+        for key, value in record.items():
+            if key in SCHEMA.identity_fields:
+                continue
+            if not _is_number(value):
+                raise ResultsValidationError(
+                    f"Record {idx} metric '{key}' must be numeric."
+                )
+
+
+def infer_metric_columns(records: Iterable[dict]) -> list[str]:
+    records = list(records)
+    if not records:
+        return []
+
+    all_keys = set()
+    for record in records:
+        all_keys.update(record.keys())
+    metric_keys = [k for k in all_keys if k not in SCHEMA.identity_fields]
+
+    ordered = []
+    for key in SCHEMA.required_metrics:
+        if key in metric_keys:
+            ordered.append(key)
+    for key in SCHEMA.optional_metrics:
+        if key in metric_keys:
+            ordered.append(key)
+
+    remaining = sorted([k for k in metric_keys if k not in ordered])
+    ordered.extend(remaining)
+    return ordered
+
+
+def build_dataframe(records: list[dict]) -> tuple[pd.DataFrame, list[str]]:
+    validate_records(records)
+    metric_cols = infer_metric_columns(records)
+    column_order = list(SCHEMA.identity_fields) + metric_cols
+    df = pd.DataFrame.from_records(records)
+    df = df[column_order]
+    return df, column_order

src/leaderboard/schema.py

@@ -0,0 +1,26 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+
+@dataclass(frozen=True)
+class TemporalBenchSchema:
+    identity_fields: tuple[str, ...] = (
+        "model_name",
+        "agent_name",
+        "agent_type",
+        "base_model",
+    )
+    required_metrics: tuple[str, ...] = (
+        "T1_acc",
+        "T2_acc",
+        "T3_acc",
+        "T4_acc",
+    )
+    optional_metrics: tuple[str, ...] = (
+        "T2_MAE",
+        "T4_sMAPE",
+    )
+
+
+SCHEMA = TemporalBenchSchema()