biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[V0 deprecation] Guided decoding (#21347)

Browse Source 
Signed-off-by: Reza Barazesh <rezabarazesh@meta.com>
Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
Co-authored-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Reza Barazesh
2025-07-29 03:15:30 -07:00
committed by GitHub
parent a4528f0cac
commit 37efc63b64
29 changed files with 103 additions and 2809 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
vllm/engine/arg_utils.py
View File

@@ -25,14 +25,14 @@ from vllm.config import (BlockSize, CacheConfig, CacheDType, CompilationConfig,
ConfigFormat, ConfigType, ConvertOption,
DecodingConfig, DetailedTraceModules, Device,
DeviceConfig, DistributedExecutorBackend,
GuidedDecodingBackend, GuidedDecodingBackendV1,
HfOverrides, KVEventsConfig, KVTransferConfig,
LoadConfig, LogprobsMode, LoRAConfig, ModelConfig,
ModelDType, ModelImpl, MultiModalConfig,
ObservabilityConfig, ParallelConfig, PoolerConfig,
PrefixCachingHashAlgo, RunnerOption, SchedulerConfig,
SchedulerPolicy, SpeculativeConfig, TaskOption,
TokenizerMode, VllmConfig, get_attr_docs, get_field)
GuidedDecodingBackend, HfOverrides, KVEventsConfig,
KVTransferConfig, LoadConfig, LogprobsMode,
LoRAConfig, ModelConfig, ModelDType, ModelImpl,
MultiModalConfig, ObservabilityConfig, ParallelConfig,
PoolerConfig, PrefixCachingHashAlgo, RunnerOption,
SchedulerConfig, SchedulerPolicy, SpeculativeConfig,
TaskOption, TokenizerMode, VllmConfig, get_attr_docs,
get_field)
from vllm.logger import init_logger
from vllm.platforms import CpuArchEnum, current_platform
from vllm.plugins import load_general_plugins
@@ -1343,14 +1343,6 @@ class EngineArgs:
recommend_to_remove=True)
return False
if self.guided_decoding_backend not in get_args(
GuidedDecodingBackendV1):
_raise_or_fallback(
feature_name=
f"--guided-decoding-backend={self.guided_decoding_backend}",
recommend_to_remove=False)
return False
# Need at least Ampere for now (FA support required).
# Skip this check if we are running on a non-GPU platform,
# or if the device capability is not available