Hugging Face's logo

Spaces:
rustmizan
/
leaderboard
Sleeping

App Files Community
leaderboard
File size: 4,158 Bytes
6f30e99
 
 
 
 
 
 
7e7fbbc
6f30e99
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7e7fbbc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6f30e99
 
 
 
 
 
 
 
 
 
 
7e7fbbc
6f30e99
 
 
 
 
7e7fbbc
6f30e99
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7e7fbbc
6f30e99
 
 
 
 
7e7fbbc
6f30e99
 
 
 
 
 
 
 
 
 
 
 
7e7fbbc
6f30e99
 
 
 
 
 
 
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
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
from dataclasses import dataclass, make_dataclass
from enum import Enum

import pandas as pd

from src.about import Tasks


def fields(raw_class):
    return [v for k, v in raw_class.__dict__.items() if k[:2] != "__" and k[-2:] != "__"]


# These classes are for user facing column names,
# to avoid having to change them all around the code
# when a modif is needed
@dataclass
class ColumnContent:
    name: str
    type: str
    displayed_by_default: bool
    hidden: bool = False
    never_hidden: bool = False


## Leaderboard columns - Define manually to avoid mutable default issues
@dataclass(frozen=True)
class AutoEvalColumn:
    # Model identification
    model = ColumnContent("Model", "markdown", True, never_hidden=True)
    dataset_variant = ColumnContent("Dataset", "str", True)

    # Primary scores - always visible
    binary_accuracy = ColumnContent("Binary Acc.", "number", True)
    cwe_f1 = ColumnContent("CWE F1", "number", True)
    function_f1 = ColumnContent("Function F1", "number", True)
    line_f1 = ColumnContent("Line F1", "number", True)
    success_at_1_function = ColumnContent("Success@1-Func", "number", True)
    success_at_1_line = ColumnContent("Success@1-Line", "number", True)

    # Detailed CWE metrics - hidden by default
    cwe_precision = ColumnContent("CWE Precision", "number", False)
    cwe_recall = ColumnContent("CWE Recall", "number", False)

    # Detailed function metrics - hidden by default
    function_precision = ColumnContent("Function Precision", "number", False)
    function_recall = ColumnContent("Function Recall", "number", False)

    # Detailed line metrics - hidden by default
    line_precision = ColumnContent("Line Precision", "number", False)
    line_recall = ColumnContent("Line Recall", "number", False)

    # Sample information
    samples = ColumnContent("Samples", "number", False)
    model_type = ColumnContent("Type", "str", False)
    precision = ColumnContent("Precision", "str", False)


## For the queue columns in the submission tab
@dataclass(frozen=True)
class EvalQueueColumn:  # Queue column
    model = ColumnContent("model", "markdown", True)
    revision = ColumnContent("revision", "str", True)
    private = ColumnContent("private", "bool", True)
    precision = ColumnContent("precision", "str", True)
    weight_type = ColumnContent("weight_type", "str", "Original")
    status = ColumnContent("status", "str", True)


## All the model information that we might need
@dataclass
class ModelDetails:
    name: str
    display_name: str = ""
    symbol: str = ""  # emoji


class ModelType(Enum):
    PT = ModelDetails(name="pretrained", symbol="🟢")
    FT = ModelDetails(name="fine-tuned", symbol="🔶")
    IFT = ModelDetails(name="instruction-tuned", symbol="⭕")
    RL = ModelDetails(name="RL-tuned", symbol="🟦")
    Unknown = ModelDetails(name="", symbol="?")

    def to_str(self, separator=" "):
        return f"{self.value.symbol}{separator}{self.value.name}"

    @staticmethod
    def from_str(type):
        if "fine-tuned" in type or "🔶" in type:
            return ModelType.FT
        if "pretrained" in type or "🟢" in type:
            return ModelType.PT
        if "RL-tuned" in type or "🟦" in type:
            return ModelType.RL
        if "instruction-tuned" in type or "⭕" in type:
            return ModelType.IFT
        return ModelType.Unknown


class WeightType(Enum):
    Adapter = ModelDetails("Adapter")
    Original = ModelDetails("Original")
    Delta = ModelDetails("Delta")


class Precision(Enum):
    float16 = ModelDetails("float16")
    bfloat16 = ModelDetails("bfloat16")
    Unknown = ModelDetails("?")

    def from_str(precision):
        if precision in ["torch.float16", "float16"]:
            return Precision.float16
        if precision in ["torch.bfloat16", "bfloat16"]:
            return Precision.bfloat16
        return Precision.Unknown


# Column selection
COLS = [c.name for c in fields(AutoEvalColumn) if not c.hidden]

EVAL_COLS = [c.name for c in fields(EvalQueueColumn)]
EVAL_TYPES = [c.type for c in fields(EvalQueueColumn)]

BENCHMARK_COLS = [t.value.col_name for t in Tasks]