File size: 3,124 Bytes
9230ae1 359afe5 9230ae1 359afe5 9230ae1 359afe5 9230ae1 359afe5 9230ae1 359afe5 9230ae1 359afe5 9230ae1 359afe5 9230ae1 359afe5 9230ae1 359afe5 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 |
from dataclasses import dataclass, make_dataclass
from enum import Enum
import pandas as pd
from src.about import Tasks # assume Tasks = [Task1, Task2, ...]
def fields(raw_class):
return [
v for k, v in raw_class.__dict__.items()
if not (k.startswith("__") and k.endswith("__"))
]
@dataclass
class ColumnContent:
name: str
type: str
displayed_by_default: bool
hidden: bool = False
never_hidden: bool = False
# -------------------------------------------------------------------
# Build leaderboard columns
# -------------------------------------------------------------------
auto_eval_column_dict = []
# Rank/Model/Badge
auto_eval_column_dict.append(["rank", ColumnContent, ColumnContent("Rank", "number", True, never_hidden=True)])
auto_eval_column_dict.append(["model", ColumnContent, ColumnContent("Model", "markdown", True, never_hidden=True)])
auto_eval_column_dict.append(["badge", ColumnContent, ColumnContent("Badge", "str", True)])
# Per-dataset metrics
# Example: "PER ⬇️ (TIMIT)", "Avg Duration (s) (TIMIT)"
for task in Tasks:
dataset_name = task.name # short name
col_base = task.value.col_name # e.g. "PER ⬇️"
# allow multiple metrics per dataset if needed
auto_eval_column_dict.append([
f"{dataset_name}_per",
ColumnContent,
ColumnContent(f"{col_base} ({dataset_name})", "number", True),
])
auto_eval_column_dict.append([
f"{dataset_name}_avg_duration",
ColumnContent,
ColumnContent(f"Avg Duration (s) ({dataset_name})", "number", True),
])
# Global average across datasets
auto_eval_column_dict.append([
"average", ColumnContent, ColumnContent("Avg PER ⬇️ (All)", "number", True)
])
# Extra model info
auto_eval_column_dict.append(["model_type", ColumnContent, ColumnContent("Type", "str", False)])
auto_eval_column_dict.append(["architecture", ColumnContent, ColumnContent("Architecture", "str", False)])
auto_eval_column_dict.append(["weight_type", ColumnContent, ColumnContent("Weight type", "str", False, True)])
auto_eval_column_dict.append(["precision", ColumnContent, ColumnContent("Precision", "str", False)])
auto_eval_column_dict.append(["license", ColumnContent, ColumnContent("Hub License", "str", False)])
auto_eval_column_dict.append(["params", ColumnContent, ColumnContent("#Params (B)", "number", False)])
auto_eval_column_dict.append(["likes", ColumnContent, ColumnContent("Hub ❤️", "number", False)])
auto_eval_column_dict.append(["still_on_hub", ColumnContent, ColumnContent("Available on the hub", "bool", False)])
auto_eval_column_dict.append(["revision", ColumnContent, ColumnContent("Model sha", "str", False, False)])
# Final dataclass
AutoEvalColumn = make_dataclass("AutoEvalColumn", auto_eval_column_dict, frozen=True)
# -------------------------------------------------------------------
# Example: Create dataframe header
# -------------------------------------------------------------------
COLS = [c.name for c in fields(AutoEvalColumn) if not c.hidden]
df = pd.DataFrame(columns=[c.name for c in fields(AutoEvalColumn)])
|