mirror of
https://git.datalinker.icu/vllm-project/vllm.git
synced 2025-12-16 06:55:01 +08:00
48fb076cbc
Signed-off-by: Nick Hill <nhill@redhat.com> Signed-off-by: Andrew Feldman <afeldman@neuralmagic.com> Signed-off-by: Andrew Feldman <afeldman@redhat.com> Co-authored-by: Nick Hill <nhill@redhat.com>
44 lines
1.1 KiB
Python
44 lines
1.1 KiB
Python
# SPDX-License-Identifier: Apache-2.0
|
|
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
|
|
|
from dataclasses import dataclass
|
|
from typing import Optional
|
|
|
|
import torch
|
|
|
|
from vllm.v1.sample.logits_processor import LogitsProcessorManager
|
|
|
|
|
|
@dataclass
|
|
class SamplingMetadata:
|
|
|
|
temperature: Optional[torch.Tensor]
|
|
all_greedy: bool
|
|
all_random: bool
|
|
|
|
top_p: Optional[torch.Tensor]
|
|
top_k: Optional[torch.Tensor]
|
|
|
|
generators: dict[int, torch.Generator]
|
|
|
|
# None means no logprobs, 0 means sampled token logprobs only
|
|
max_num_logprobs: Optional[int]
|
|
|
|
no_penalties: bool
|
|
prompt_token_ids: Optional[torch.Tensor]
|
|
frequency_penalties: torch.Tensor
|
|
presence_penalties: torch.Tensor
|
|
repetition_penalties: torch.Tensor
|
|
|
|
output_token_ids: list[list[int]]
|
|
|
|
# `allowed_token_ids_mask` is a 2D bool tensor of shape (max batch size,
|
|
# vocab size).
|
|
allowed_token_ids_mask: Optional[torch.Tensor]
|
|
|
|
# req_index -> bad_words_token_ids
|
|
bad_words_token_ids: dict[int, list[list[int]]]
|
|
|
|
# Loaded logits processors
|
|
logitsprocs: LogitsProcessorManager
|