yera.models.interfaces.llms._thinking
Utility functions for parsing model reasoning output with delimiters.
Symbols
can_disable
can_disable(
capabilities: LLMCapabilities,
) → boolReturn whether reasoning can be disabled for a model.
Parameters
Model reasoning capabilities.
Returns
Whether the model accepts the off reasoning level.
effort_to_budget
effort_to_budget(
level: ReasoningLevel,
) → intMap a reasoning effort level to a token budget.
Parameters
Reasoning effort level to translate.
Returns
The corresponding reasoning-token budget.
Raises
If the level has no fixed budget mapping.
floor_level
floor_level(
capabilities: LLMCapabilities,
) → ReasoningLevel | NoneReturn the least intensive supported reasoning level.
Parameters
Model reasoning capabilities.
Returns
The lowest supported effort level, or None when none is declared.
resolve_effort
resolve_effort(
requested: ReasoningLevel,
capabilities: LLMCapabilities,
) → ReasoningLevelValidate a requested reasoning effort against model capabilities.
Parameters
Requested reasoning effort.
Model reasoning capabilities.
Returns
The validated reasoning effort.
Raises
If the requested effort is unsupported or reasoning cannot be disabled.
ThinkingParser
A parser for model output that separates thinking and response content.
Attributes
The delimiter marking the start of thinking content.
The delimiter marking the end of thinking content.
Whether the parser is currently in a thinking block.
Whether the first block should be considered thinking.
Methods
ThinkingParser.__init__
__init__(
start_delimiter: str = '',
end_delimiter: str = ' ',
starts_thinking: bool = False,
)Initialize a streaming reasoning parser.
Parameters
Marker opening reasoning content.
Marker closing reasoning content.
Whether content begins inside a reasoning block.
Raises
If either delimiter is empty.
ThinkingParser.parse_stream
parse_stream(
stream: Iterator[str],
) → Iterator[LLMToken]Convert a stream of raw text chunks into classified tokens.
Parameters
Chunks of model output, of any size.