[Doc] Fix stale rejection_sample_method and synthetic_acceptance_rate (#51611)
Signed-off-by: qwerqwerqwe8688-jpg <xuuuuu2021@163.com>
QWERQWERQWE86 committed
Aug 13, 2026 at 18:56 UTC
443fa5aca0253fd6b27ae94bd4b232418612ac51
1 file changed
+3
-2
docs/features/speculative_decoding/README.md
+3
-2
@@ -86,8 +86,9 @@ only apply to model-based methods such as `draft_model`, `mtp`, `eagle3`, and
86
| `draft_tensor_parallel_size` | `integer >= 1` | `None` | Tensor parallel size for the draft model. |
87
| `max_model_len` | `integer >= 1` | `None` | Maximum context length for the draft model. |
88
| `parallel_drafting` | `boolean` | `false` | Enable parallel draft token generation. Only compatible with EAGLE and draft-model methods. |
89
-| `rejection_sample_method` | `string` | `strict` | `strict`, `probabilistic`, or `synthetic`. |
90
-| `synthetic_acceptance_rate` | `float` | `None` | Average acceptance rate to target when `rejection_sample_method` is `synthetic`. Valid range is `[0, 1]`. |
89
+| `rejection_sample_method` | `string` | `standard` | `standard`, `synthetic`, or `block`. |
90
+| `synthetic_acceptance_rates` | `list[float]` | `None` | Per-position unconditional acceptance rates for `synthetic` rejection sampling. Each entry in `[0, 1]`; length must equal `num_speculative_tokens`; must be non-increasing. |
91
+| `synthetic_acceptance_length` | `float` | `None` | Target mean acceptance length for `synthetic`; in `[1, num_speculative_tokens + 1]`. Mutually exclusive with `synthetic_acceptance_rates`. |
92
| `use_heterogeneous_vocab` | `boolean` | `false` | Allow draft and target models with different vocabularies. Builds a token-level intersection at initialisation and constrains draft logits to shared tokens only. Only compatible with `method=draft_model`. Probabilistic draft sampling (`draft_sample_method='probabilistic'`) is not yet supported when this option is enabled. |
93
94
!!! note