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 data
FUTURE_TOKEN
FUTURE_TOKEN
max.pipelines.context.FUTURE_TOKEN = -999
Placeholder written to a generated-token slot whose value is still being computed on device under overlap scheduling.
A batch’s token array carries this sentinel in positions the current GPU
step has yet to produce; the pipeline replaces each occurrence with the
real token once that step’s outputs land (see
max.pipelines.lib.pipeline_variants.overlap_text_generation).