For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).
Python function
build_synthetic_acceptance_sampler_graph
build_synthetic_acceptance_sampler_graph()
max.pipelines.sampling.build_synthetic_acceptance_sampler_graph(device, base_acceptance_rate, num_draft_steps)
Builds a graph that implements synthetic acceptance sampling.
The seed is a graph input so callers can bind a fresh int64 value per execution; without that, static-seed RNG would produce the same random draws every call.
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!