[V0 Deprecation] Remove Prompt Adapters (#20588)

Signed-off-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
Michael Goin
2025-07-23 16:36:48 -07:00
committed by GitHub
parent 78c13e30e1
commit 82ec66f514
60 changed files with 126 additions and 1727 deletions
+1 -6
View File
@@ -20,7 +20,6 @@ from vllm.lora.request import LoRARequest
from vllm.multimodal import MULTIMODAL_REGISTRY, MultiModalRegistry
from vllm.outputs import PoolingRequestOutput, RequestOutput
from vllm.pooling_params import PoolingParams
from vllm.prompt_adapter.request import PromptAdapterRequest
from vllm.sampling_params import SamplingParams
from vllm.transformers_utils.config import (
maybe_register_config_serialize_by_value)
@@ -221,7 +220,6 @@ class AsyncLLM(EngineClient):
lora_request: Optional[LoRARequest] = None,
tokenization_kwargs: Optional[dict[str, Any]] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
data_parallel_rank: Optional[int] = None,
) -> RequestOutputCollector:
@@ -238,8 +236,7 @@ class AsyncLLM(EngineClient):
# Convert Input --> Request.
prompt_str, request = self.processor.process_inputs(
request_id, prompt, params, arrival_time, lora_request,
tokenization_kwargs, trace_headers, prompt_adapter_request,
priority, data_parallel_rank)
tokenization_kwargs, trace_headers, priority, data_parallel_rank)
if is_pooling or params.n == 1:
await self._add_request(request, prompt_str, None, 0, queue)
@@ -283,7 +280,6 @@ class AsyncLLM(EngineClient):
request_id: str,
lora_request: Optional[LoRARequest] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
data_parallel_rank: Optional[int] = None,
) -> AsyncGenerator[RequestOutput, None]:
@@ -314,7 +310,6 @@ class AsyncLLM(EngineClient):
sampling_params,
lora_request=lora_request,
trace_headers=trace_headers,
prompt_adapter_request=prompt_adapter_request,
priority=priority,
data_parallel_rank=data_parallel_rank,
)
+1 -4
View File
@@ -17,7 +17,6 @@ from vllm.lora.request import LoRARequest
from vllm.multimodal import MULTIMODAL_REGISTRY, MultiModalRegistry
from vllm.outputs import PoolingRequestOutput, RequestOutput
from vllm.pooling_params import PoolingParams
from vllm.prompt_adapter.request import PromptAdapterRequest
from vllm.sampling_params import SamplingParams
from vllm.transformers_utils.tokenizer_group import (
TokenizerGroup, init_tokenizer_from_configs)
@@ -192,7 +191,6 @@ class LLMEngine:
lora_request: Optional[LoRARequest] = None,
tokenization_kwargs: Optional[dict[str, Any]] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
) -> None:
# Validate the request_id type.
@@ -203,8 +201,7 @@ class LLMEngine:
# Process raw inputs into the request.
prompt_str, request = self.processor.process_inputs(
request_id, prompt, params, arrival_time, lora_request,
tokenization_kwargs, trace_headers, prompt_adapter_request,
priority)
tokenization_kwargs, trace_headers, priority)
n = params.n if isinstance(params, SamplingParams) else 1
-6
View File
@@ -16,7 +16,6 @@ from vllm.multimodal.inputs import PlaceholderRange
from vllm.multimodal.processing import EncDecMultiModalProcessor
from vllm.multimodal.utils import merge_and_sort_multimodal_metadata
from vllm.pooling_params import PoolingParams
from vllm.prompt_adapter.request import PromptAdapterRequest
from vllm.sampling_params import SamplingParams
from vllm.transformers_utils.tokenizer_group import TokenizerGroup
from vllm.v1.engine import EngineCoreRequest
@@ -226,7 +225,6 @@ class Processor:
lora_request: Optional[LoRARequest] = None,
tokenization_kwargs: Optional[dict[str, Any]] = None,
trace_headers: Optional[Mapping[str, str]] = None,
prompt_adapter_request: Optional[PromptAdapterRequest] = None,
priority: int = 0,
data_parallel_rank: Optional[int] = None,
) -> tuple[Optional[str], EngineCoreRequest]:
@@ -237,8 +235,6 @@ class Processor:
self._validate_params(params, lora_request)
if trace_headers is not None:
raise ValueError("V1 does not support tracing yet.")
if prompt_adapter_request is not None:
raise ValueError("V1 does not support prompt_adapter_request.")
data_parallel_size = self.vllm_config.parallel_config.data_parallel_size
if data_parallel_rank is not None and not (0 <= data_parallel_rank <
@@ -253,12 +249,10 @@ class Processor:
# 1. Tokenize text prompt, with LoRA request if one exists.
# 2. For multimodal models with a merged preprocessor, preprocess
# multimodal data and expand prompt token ids accordingly.
# 3. Apply prompt adapter to prompt token ids if one exists.
processed_inputs: ProcessorInputs = self.input_preprocessor.preprocess(
prompt,
tokenization_kwargs=tokenization_kwargs,
lora_request=lora_request,
prompt_adapter_request=prompt_adapter_request,
return_mm_hashes=self.use_hash,
)
from vllm.platforms import current_platform
-2
View File
@@ -318,8 +318,6 @@ def report_usage_stats(
# Feature flags
"enable_lora":
bool(vllm_config.lora_config),
"enable_prompt_adapter":
bool(vllm_config.prompt_adapter_config),
"enable_prefix_caching":
vllm_config.cache_config.enable_prefix_caching,
"enforce_eager":
-1
View File
@@ -104,7 +104,6 @@ class GPUModelRunner(LoRAModelRunnerMixin):
self.parallel_config = vllm_config.parallel_config
self.scheduler_config = vllm_config.scheduler_config
self.speculative_config = vllm_config.speculative_config
self.prompt_adapter_config = vllm_config.prompt_adapter_config
self.observability_config = vllm_config.observability_config
from vllm.model_executor.models.utils import set_cpu_offload_max_bytes
-1
View File
@@ -114,7 +114,6 @@ class TPUModelRunner(LoRAModelRunnerMixin):
self.original_parallel_config = original_parallel_config
self.scheduler_config = vllm_config.scheduler_config
self.speculative_config = vllm_config.speculative_config
self.prompt_adapter_config = vllm_config.prompt_adapter_config
self.observability_config = vllm_config.observability_config
self.device_config = vllm_config.device_config
-1
View File
@@ -62,7 +62,6 @@ class TPUWorker:
self.scheduler_config = vllm_config.scheduler_config
self.device_config = vllm_config.device_config
self.speculative_config = vllm_config.speculative_config
self.prompt_adapter_config = vllm_config.prompt_adapter_config
self.observability_config = vllm_config.observability_config
self.parallel_config.rank = rank