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 module
max.pipelines.speculative
Speculative decoding pipelines and configuration for MAX.
Configuration
SpeculativeConfig | Configures speculative decoding for a pipeline. |
|---|---|
SpeculativeMethod | The supported methods for speculative decoding. |
VerifyWidthRange | One inclusive decode-batch-size range and the drafts to verify in it. |
MAGIC_DRAFT_TOKEN_ID | Sentinel draft-token id for prefill / dummy-draft graph-capture steps. |
|---|
Draft weights
DraftAliases | The weights a draft inherits from its target instead of loading. |
|---|
NO_ALIASES | A draft that loads every weight itself. |
|---|
validate_draft_state_dict | Checks a draft checkpoint covers every weight the draft module needs. |
|---|