For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Python class

LabelScoringTokenizer

LabelScoringTokenizer

class max.pipelines.modeling.types.LabelScoringTokenizer(*args, **kwargs)

source

Bases: Protocol

Tokenizer capabilities that label scoring (/v1/decisions) needs.

Declared as a protocol, checked once when a scoring request is admitted, so a tokenizer that lacks a member is refused instead of having its context-length check skipped.

apply_chat_template()

apply_chat_template(messages, tools, **chat_template_options)

source

Renders messages with the model’s chat template.

Parameters:

• messages (list[TextGenerationRequestMessage])
• tools (list[TextGenerationRequestTool] | None)
• chat_template_options (Any)

Return type:

str

encode()

async encode(prompt, add_special_tokens=True)

source

Encodes text to token ids.

Parameters:

• prompt (str)
• add_special_tokens (bool)

Return type:

ndarray[tuple[Any, …], dtype[integer[Any]]]

max_length

property max_length: int | None

source

The context length, in tokens; None when unbounded.