Skip to content

text_backend_names

Text backend name manipulation utilities.

Functions for detecting, stripping, and generating text-backend-prefixed model name variants used in the legacy Horde API format.

TEXT_LEGACY_BACKEND_PREFIXES module-attribute 

TEXT_LEGACY_BACKEND_PREFIXES: dict[TEXT_BACKENDS, str] = {
    aphrodite: "aphrodite/",
    koboldcpp: "koboldcpp/",
}

Backend prefixes on duplicate entries for backwards compatibility in the legacy format.

has_legacy_text_backend_prefix 

has_legacy_text_backend_prefix(model_name: str) -> bool

Check if a model name has a legacy text backend prefix.

Parameters:

  • model_name (str) –

    The model name to check.

Returns:

  • bool –

    True if the model name has a legacy text backend prefix, False otherwise.

Source code in src/horde_model_reference/text_backend_names.py 
18
19
20
21
22
23
24
25
26
27
28
def has_legacy_text_backend_prefix(model_name: str) -> bool:
    """Check if a model name has a legacy text backend prefix.

    Args:
        model_name: The model name to check.

    Returns:
        True if the model name has a legacy text backend prefix, False otherwise.

    """
    return any(model_name.startswith(prefix) for prefix in TEXT_LEGACY_BACKEND_PREFIXES.values())

strip_backend_prefix 

strip_backend_prefix(model_name: str) -> str

Strip backend prefix from a model name if present.

Parameters:

  • model_name (str) –

    The model name to strip.

Returns:

  • str –

    The model name without the backend prefix.

Example

strip_backend_prefix("koboldcpp/Broken-Tutu-24B") 'Broken-Tutu-24B' strip_backend_prefix("aphrodite/ReadyArt/Broken-Tutu-24B") 'ReadyArt/Broken-Tutu-24B' strip_backend_prefix("ReadyArt/Broken-Tutu-24B") 'ReadyArt/Broken-Tutu-24B'

Source code in src/horde_model_reference/text_backend_names.py 
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
def strip_backend_prefix(model_name: str) -> str:
    """Strip backend prefix from a model name if present.

    Args:
        model_name: The model name to strip.

    Returns:
        The model name without the backend prefix.

    Example:
        >>> strip_backend_prefix("koboldcpp/Broken-Tutu-24B")
        'Broken-Tutu-24B'
        >>> strip_backend_prefix("aphrodite/ReadyArt/Broken-Tutu-24B")
        'ReadyArt/Broken-Tutu-24B'
        >>> strip_backend_prefix("ReadyArt/Broken-Tutu-24B")
        'ReadyArt/Broken-Tutu-24B'

    """
    for prefix in TEXT_LEGACY_BACKEND_PREFIXES.values():
        if model_name.startswith(prefix):
            return model_name[len(prefix) :]
    return model_name

validate_not_backend_prefixed 

validate_not_backend_prefixed(model_name: str) -> None

Reject model names that start with a known or unknown backend-like prefix.

Writes should always use the canonical (unprefixed) model name. Backend-prefixed duplicates are generated automatically. This function raises ValueError if the name looks like it was submitted with a backend prefix.

Parameters:

  • model_name (str) –

    The model name to validate.

Raises:

  • ValueError –

    If the model name starts with a known backend prefix or matches the <word>/ pattern where <word> is a single lowercase ASCII token that could be mistaken for a backend prefix (e.g., vllm/SomeModel). Legitimate author-prefixed names like ReadyArt/Model are allowed because the author segment typically contains uppercase letters or digits.

Source code in src/horde_model_reference/text_backend_names.py 
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
def validate_not_backend_prefixed(model_name: str) -> None:
    """Reject model names that start with a known or unknown backend-like prefix.

    Writes should always use the canonical (unprefixed) model name. Backend-prefixed
    duplicates are generated automatically. This function raises ``ValueError`` if the
    name looks like it was submitted with a backend prefix.

    Args:
        model_name: The model name to validate.

    Raises:
        ValueError: If the model name starts with a known backend prefix or matches
            the ``<word>/`` pattern where ``<word>`` is a single lowercase ASCII token
            that could be mistaken for a backend prefix (e.g., ``vllm/SomeModel``).
            Legitimate author-prefixed names like ``ReadyArt/Model`` are allowed
            because the author segment typically contains uppercase letters or digits.

    """
    for backend, prefix in TEXT_LEGACY_BACKEND_PREFIXES.items():
        if model_name.startswith(prefix):
            raise ValueError(
                f"Model name {model_name!r} starts with backend prefix {prefix!r} "
                f"(backend: {backend.value}). Use the canonical name without the prefix; "
                f"backend-prefixed duplicates are generated automatically."
            )

get_model_name_variants 

get_model_name_variants(canonical_name: str) -> list[str]

Get all possible name variants for a canonical model name.

Given a canonical name like "ReadyArt/Broken-Tutu-24B", returns all possible variants that might appear in the Horde API stats: - Canonical: ReadyArt/Broken-Tutu-24B - Aphrodite: aphrodite/ReadyArt/Broken-Tutu-24B (prefix + full canonical name) - KoboldCpp: koboldcpp/Broken-Tutu-24B (prefix + model name only)

Parameters:

  • canonical_name (str) –

    The canonical model name from the model reference.

Returns:

  • list[str] –

    List of all possible name variants, including the canonical name.

Example

get_model_name_variants("ReadyArt/Broken-Tutu-24B") ['ReadyArt/Broken-Tutu-24B', 'aphrodite/ReadyArt/Broken-Tutu-24B', 'koboldcpp/Broken-Tutu-24B']

Source code in src/horde_model_reference/text_backend_names.py 
 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
def get_model_name_variants(canonical_name: str) -> list[str]:
    """Get all possible name variants for a canonical model name.

    Given a canonical name like "ReadyArt/Broken-Tutu-24B", returns all possible
    variants that might appear in the Horde API stats:
    - Canonical: ReadyArt/Broken-Tutu-24B
    - Aphrodite: aphrodite/ReadyArt/Broken-Tutu-24B (prefix + full canonical name)
    - KoboldCpp: koboldcpp/Broken-Tutu-24B (prefix + model name only)

    Args:
        canonical_name: The canonical model name from the model reference.

    Returns:
        List of all possible name variants, including the canonical name.

    Example:
        >>> get_model_name_variants("ReadyArt/Broken-Tutu-24B")
        ['ReadyArt/Broken-Tutu-24B', 'aphrodite/ReadyArt/Broken-Tutu-24B', 'koboldcpp/Broken-Tutu-24B']
    """
    variants = [canonical_name]

    model_name_only = canonical_name.split("/", 1)[1] if "/" in canonical_name else canonical_name

    def _append_variant(value: str) -> None:
        if value not in variants:
            variants.append(value)

    for backend, prefix in TEXT_LEGACY_BACKEND_PREFIXES.items():
        if backend == TEXT_BACKENDS.koboldcpp:
            _append_variant(f"{prefix}{model_name_only}")
        elif backend == TEXT_BACKENDS.aphrodite:
            _append_variant(f"{prefix}{canonical_name}")

    return variants