# SPDX-License-Identifier: Apache-2.0 import ast import copy import enum import hashlib import importlib.metadata import json import sys import warnings from collections import Counter from collections.abc import Mapping from contextlib import contextmanager from dataclasses import dataclass, field, replace from importlib.util import find_spec from pathlib import Path from typing import (TYPE_CHECKING, Any, Callable, ClassVar, Final, Literal, Optional, Protocol, Union) import torch from packaging.version import Version from pydantic import BaseModel, Field, PrivateAttr from torch.distributed import ProcessGroup, ReduceOp from transformers import PretrainedConfig import vllm.envs as envs from vllm.compilation.inductor_pass import CallableInductorPass, InductorPass from vllm.logger import init_logger from vllm.model_executor.layers.quantization import (QUANTIZATION_METHODS, get_quantization_config) from vllm.model_executor.models import ModelRegistry from vllm.platforms import CpuArchEnum from vllm.sampling_params import GuidedDecodingParams from vllm.tracing import is_otel_available, otel_import_error_traceback from vllm.transformers_utils.config import ( ConfigFormat, get_config, get_hf_image_processor_config, get_hf_text_config, get_pooling_config, get_sentence_transformer_tokenizer_config, is_encoder_decoder, try_get_generation_config, uses_mrope) from vllm.transformers_utils.s3_utils import S3Model from vllm.transformers_utils.utils import is_s3, maybe_model_redirect from vllm.utils import (GiB_bytes, LayerBlockType, cuda_device_count_stateless, get_cpu_memory, get_open_port, random_uuid, resolve_obj_by_qualname) if TYPE_CHECKING: from ray.util.placement_group import PlacementGroup from vllm.executor.executor_base import ExecutorBase from vllm.model_executor.layers.quantization.base_config import ( QuantizationConfig) from vllm.model_executor.model_loader.loader import BaseModelLoader from vllm.transformers_utils.tokenizer_group.base_tokenizer_group import ( BaseTokenizerGroup) else: QuantizationConfig = None logger = init_logger(__name__) # This value is chosen to have a balance between ITL and TTFT. Note it is # not optimized for throughput. _DEFAULT_MAX_NUM_BATCHED_TOKENS = 2048 _POOLING_MODEL_MAX_NUM_BATCHED_TOKENS = 32768 _MULTIMODAL_MODEL_MAX_NUM_BATCHED_TOKENS = 5120 TaskOption = Literal["auto", "generate", "embedding", "embed", "classify", "score", "reward", "transcription"] _ResolvedTask = Literal["generate", "embed", "classify", "score", "reward", "draft", "transcription"] RunnerType = Literal["generate", "pooling", "draft", "transcription"] _RUNNER_TASKS: dict[RunnerType, list[_ResolvedTask]] = { "generate": ["generate"], "pooling": ["embed", "classify", "score", "reward"], "draft": ["draft"], "transcription": ["transcription"], } _TASK_RUNNER: dict[_ResolvedTask, RunnerType] = { task: runner for runner, tasks in _RUNNER_TASKS.items() for task in tasks } HfOverrides = Union[dict[str, Any], Callable[[PretrainedConfig], PretrainedConfig]] class SupportsHash(Protocol): def compute_hash(self) -> str: ... class SupportsMetricsInfo(Protocol): def metrics_info(self) -> dict[str, str]: ... class ModelImpl(str, enum.Enum): AUTO = "auto" VLLM = "vllm" TRANSFORMERS = "transformers" class ModelConfig: """Configuration for the model. Args: model: Name or path of the huggingface model to use. It is also used as the content for `model_name` tag in metrics output when `served_model_name` is not specified. task: The task to use the model for. Each vLLM instance only supports one task, even if the same model can be used for multiple tasks. When the model only supports one task, "auto" can be used to select it; otherwise, you must specify explicitly which task to use. tokenizer: Name or path of the huggingface tokenizer to use. tokenizer_mode: Tokenizer mode. "auto" will use the fast tokenizer if available, "slow" will always use the slow tokenizer, "mistral" will always use the tokenizer from `mistral_common`, and "custom" will use --tokenizer to select the preregistered tokenizer. trust_remote_code: Trust remote code (e.g., from HuggingFace) when downloading the model and tokenizer. allowed_local_media_path: Allowing API requests to read local images or videos from directories specified by the server file system. This is a security risk. Should only be enabled in trusted environments. dtype: Data type for model weights and activations. The "auto" option will use FP16 precision for FP32 and FP16 models, and BF16 precision for BF16 models. seed: Random seed for reproducibility. revision: The specific model version to use. It can be a branch name, a tag name, or a commit id. If unspecified, will use the default version. code_revision: The specific revision to use for the model code on Hugging Face Hub. It can be a branch name, a tag name, or a commit id. If unspecified, will use the default version. tokenizer_revision: The specific tokenizer version to use. It can be a branch name, a tag name, or a commit id. If unspecified, will use the default version. max_model_len: Maximum length of a sequence (including prompt and output). If None, will be derived from the model. spec_target_max_model_len: Specify the the maximum length for spec decoding draft models. quantization: Quantization method that was used to quantize the model weights. If None, we assume the model weights are not quantized. enforce_eager: Whether to enforce eager execution. If True, we will disable CUDA graph and always execute the model in eager mode. If False, we will use CUDA graph and eager execution in hybrid. If None, the user did not specify, so default to False. max_seq_len_to_capture: Maximum sequence len covered by CUDA graphs. When a sequence has context length larger than this, we fall back to eager mode. Additionally for encoder-decoder models, if the sequence length of the encoder input is larger than this, we fall back to the eager mode. max_logprobs: Maximum number of log probabilities. Defaults to 20. disable_sliding_window: Whether to disable sliding window. If True, we will disable the sliding window functionality of the model. If the model does not support sliding window, this argument is ignored. skip_tokenizer_init: If true, skip initialization of tokenizer and detokenizer. served_model_name: The model name used in metrics tag `model_name`, matches the model name exposed via the APIs. If multiple model names provided, the first name will be used. If not specified, the model name will be the same as `model`. limit_mm_per_prompt: Maximum number of data items per modality per prompt. Only applicable for multimodal models. use_async_output_proc: Whether to use async output processor. Defaults to True. config_format: The config format which shall be loaded. Defaults to 'auto' which defaults to 'hf'. hf_overrides: If a dictionary, contains arguments to be forwarded to the HuggingFace config. If a callable, it is called to update the HuggingFace config. mm_processor_kwargs: Arguments to be forwarded to the model's processor for multi-modal data, e.g., image processor. disable_mm_preprocessor_cache: If true, then disables caching of the multi-modal preprocessor/mapper. (not recommended) override_neuron_config: Initialize non default neuron config or override default neuron config that are specific to Neuron devices, this argument will be used to configure the neuron config that can not be gathered from the vllm arguments. override_pooler_config: Initialize non default pooling config or override default pooling config for the pooling model. logits_processor_pattern: Optional regex pattern specifying valid logits processor qualified names that can be passed with the `logits_processors` extra completion argument. Defaults to None, which allows no processors. generation_config: Configuration parameter file for generation. model_impl: Which implementation of the model to use: "auto" will try to use the vLLM implementation if it exists and fall back to the Transformers implementation if no vLLM implementation is available. "vllm" will use the vLLM model implementation. "transformers" will use the Transformers model implementation. override_generation_config: Override the generation config with the given config. """ def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ factors: list[Any] = [] factors.append(self.model) factors.append(self.dtype) factors.append(self.quantization) factors.append(self.revision) factors.append(self.code_revision) factors.append(self.trust_remote_code) factors.append(self.rope_scaling) factors.append(self.rope_theta) # rope cos/sin cache depends on the max_position_embeddings factors.append( getattr(self.hf_config, "max_position_embeddings", "None")) return hashlib.sha256(str(factors).encode()).hexdigest() def __init__( self, model: str, task: Union[TaskOption, Literal["draft"]], tokenizer: str, tokenizer_mode: str, trust_remote_code: bool, dtype: Union[str, torch.dtype], seed: int, hf_config_path: Optional[str] = None, allowed_local_media_path: str = "", revision: Optional[str] = None, code_revision: Optional[str] = None, rope_scaling: Optional[dict[str, Any]] = None, rope_theta: Optional[float] = None, tokenizer_revision: Optional[str] = None, max_model_len: Optional[int] = None, spec_target_max_model_len: Optional[int] = None, quantization: Optional[str] = None, enforce_eager: Optional[bool] = None, max_seq_len_to_capture: Optional[int] = None, max_logprobs: int = 20, disable_sliding_window: bool = False, disable_cascade_attn: bool = False, skip_tokenizer_init: bool = False, served_model_name: Optional[Union[str, list[str]]] = None, limit_mm_per_prompt: Optional[Mapping[str, int]] = None, use_async_output_proc: bool = True, config_format: ConfigFormat = ConfigFormat.AUTO, hf_overrides: Optional[HfOverrides] = None, mm_processor_kwargs: Optional[dict[str, Any]] = None, disable_mm_preprocessor_cache: bool = False, override_neuron_config: Optional[dict[str, Any]] = None, override_pooler_config: Optional["PoolerConfig"] = None, logits_processor_pattern: Optional[str] = None, generation_config: str = "auto", enable_sleep_mode: bool = False, override_generation_config: Optional[dict[str, Any]] = None, model_impl: Union[str, ModelImpl] = ModelImpl.AUTO, ) -> None: self.model = maybe_model_redirect(model) self.tokenizer = maybe_model_redirect(tokenizer) self.hf_config_path = hf_config_path if isinstance(hf_config_path, str): self.hf_config_path = maybe_model_redirect(hf_config_path) self.tokenizer_mode = tokenizer_mode self.trust_remote_code = trust_remote_code self.allowed_local_media_path = allowed_local_media_path self.seed = seed self.revision = revision self.code_revision = code_revision self.rope_scaling = rope_scaling self.rope_theta = rope_theta self.model_impl = model_impl if hf_overrides is None: hf_overrides = {} if callable(hf_overrides): hf_overrides_kw = {} hf_overrides_fn = hf_overrides else: hf_overrides_kw = hf_overrides hf_overrides_fn = None if rope_scaling is not None: hf_override: dict[str, Any] = {"rope_scaling": rope_scaling} hf_overrides_kw.update(hf_override) hf_overrides_str = json.dumps(hf_overrides) msg = ( "`--rope-scaling` will be removed in a future release. " f"'Please instead use `--hf-overrides '{hf_overrides_str}'`") warnings.warn(DeprecationWarning(msg), stacklevel=2) if rope_theta is not None: hf_override = {"rope_theta": rope_theta} hf_overrides_kw.update(hf_override) hf_overrides_str = json.dumps(hf_overrides) msg = ( "`--rope-theta` will be removed in a future release. " f"'Please instead use `--hf-overrides '{hf_overrides_str}'`") warnings.warn(DeprecationWarning(msg), stacklevel=2) self.maybe_pull_model_tokenizer_for_s3(model, tokenizer) if (backend := envs.VLLM_ATTENTION_BACKEND ) and backend == "FLASHINFER" and find_spec("flashinfer") is None: raise ValueError( "VLLM_ATTENTION_BACKEND is set to FLASHINFER, but flashinfer " "module was not found. See " "https://github.com/vllm-project/vllm/blob/main/docker/Dockerfile " # noqa: E501 "for instructions on how to install it.") # The tokenizer version is consistent with the model version by default. if tokenizer_revision is None: self.tokenizer_revision = revision else: self.tokenizer_revision = tokenizer_revision self.quantization = quantization self.enforce_eager = enforce_eager self.max_seq_len_to_capture = max_seq_len_to_capture self.max_logprobs = max_logprobs self.disable_sliding_window = disable_sliding_window self.disable_cascade_attn = disable_cascade_attn self.skip_tokenizer_init = skip_tokenizer_init self.enable_sleep_mode = enable_sleep_mode from vllm.platforms import current_platform if self.enable_sleep_mode and not current_platform.is_cuda(): raise ValueError("Sleep mode is only supported on CUDA devices.") hf_config = get_config(self.hf_config_path or self.model, trust_remote_code, revision, code_revision, config_format) if hf_overrides_kw: logger.info("Overriding HF config with %s", hf_overrides_kw) hf_config.update(hf_overrides_kw) if hf_overrides_fn: logger.info("Overriding HF config with %s", hf_overrides_fn) hf_config = hf_overrides_fn(hf_config) self.hf_config = hf_config self.hf_text_config = get_hf_text_config(self.hf_config) self.encoder_config = self._get_encoder_config() self.hf_image_processor_config = get_hf_image_processor_config( self.model, revision) self.dtype = _get_and_verify_dtype(self.hf_config, dtype) self.use_async_output_proc = use_async_output_proc self.mm_processor_kwargs = mm_processor_kwargs self.disable_mm_preprocessor_cache = disable_mm_preprocessor_cache # Set enforce_eager to False if the value is unset. if self.enforce_eager is None: self.enforce_eager = False interleaved_attn_models = ["gemma2", "gemma3_text", "cohere2"] sliding_window = getattr(self.hf_text_config, "sliding_window", None) has_interleaved_attention = (sliding_window is not None) and ( isinstance(sliding_window, list) or (self.hf_text_config.model_type in interleaved_attn_models)) if (not self.disable_sliding_window and has_interleaved_attention): if (backend := envs.VLLM_ATTENTION_BACKEND) in ("XFORMERS", "FLASHINFER"): sliding_window_len_min = get_min_sliding_window( self.hf_text_config.sliding_window) logger.warning_once( f"{self.hf_text_config.model_type} has interleaved " "attention, which is currently not supported by the " f"{backend} backend. Disabling sliding window and capping " "the max length to the sliding window size " f"({sliding_window_len_min}).") self.disable_sliding_window = True else: # for a model with interleaved attention, # the scheduler and the model treat it as full attention # (i.e., not dropping any tokens outside the window). # only the attention layer itself is aware of the sliding # window, and use the window size to compute the attention. self.hf_text_config.interleaved_sliding_window = sliding_window delattr(self.hf_text_config, "sliding_window") sliding_window = None self.max_model_len = _get_and_verify_max_len( hf_config=self.hf_text_config, max_model_len=max_model_len, disable_sliding_window=self.disable_sliding_window, sliding_window_len=self.get_hf_config_sliding_window(), spec_target_max_model_len=spec_target_max_model_len, encoder_config=self.encoder_config) self.served_model_name = get_served_model_name(model, served_model_name) self.multimodal_config = self._init_multimodal_config( limit_mm_per_prompt) if not self.skip_tokenizer_init: self._verify_tokenizer_mode() self.is_attention_free = self._init_attention_free() self.is_hybrid = self._init_is_hybrid() self.has_noops = self._init_has_noops() self.has_inner_state = self._init_has_inner_state() if current_platform.is_neuron(): self.override_neuron_config = override_neuron_config else: self.override_neuron_config = None supported_tasks, task = self._resolve_task(task) self.supported_tasks = supported_tasks self.task: Final = task if self.task in ("draft", "generate"): self.truncation_side = "left" else: self.truncation_side = "right" self.pooler_config = self._init_pooler_config(override_pooler_config) self.logits_processor_pattern = logits_processor_pattern self.generation_config = generation_config self.override_generation_config = override_generation_config or {} self._verify_quantization() self._verify_cuda_graph() self._verify_bnb_config() @property def registry(self): return ModelRegistry @property def architectures(self) -> list[str]: return getattr(self.hf_config, "architectures", []) def maybe_pull_model_tokenizer_for_s3(self, model: str, tokenizer: str) -> None: """ Pull the model config or tokenizer to a temporary directory in case of S3. Args: model: The model name or path. tokenizer: The tokenizer name or path. """ if is_s3(model) or is_s3(tokenizer): if is_s3(model): s3_model = S3Model() s3_model.pull_files( model, allow_pattern=["*.model", "*.py", "*.json"]) self.model_weights = self.model self.model = s3_model.dir if is_s3(tokenizer): s3_tokenizer = S3Model() s3_tokenizer.pull_files( model, ignore_pattern=["*.pt", "*.safetensors", "*.bin"]) self.tokenizer = s3_tokenizer.dir def _init_multimodal_config( self, limit_mm_per_prompt: Optional[Mapping[str, int]] ) -> Optional["MultiModalConfig"]: if self.registry.is_multimodal_model(self.architectures): return MultiModalConfig(limit_per_prompt=limit_mm_per_prompt or {}) if limit_mm_per_prompt: raise ValueError("`limit_mm_per_prompt` is only supported for " "multimodal models.") return None def _get_encoder_config(self): return get_sentence_transformer_tokenizer_config( self.model, self.revision) def _init_pooler_config( self, override_pooler_config: Optional["PoolerConfig"], ) -> Optional["PoolerConfig"]: if self.runner_type == "pooling": user_config = override_pooler_config or PoolerConfig() base_config = get_pooling_config(self.model, self.revision) if base_config is not None: # Only set values that are not overridden by the user for k, v in base_config.items(): if getattr(user_config, k) is None: setattr(user_config, k, v) return user_config return None def _init_attention_free(self) -> bool: return self.registry.is_attention_free_model(self.architectures) def _init_is_hybrid(self) -> bool: return self.registry.is_hybrid_model(self.architectures) def _init_has_noops(self) -> bool: architectures = getattr(self.hf_config, "architectures", []) return self.registry.is_noops_model(architectures) def _init_has_inner_state(self) -> bool: return self.registry.model_has_inner_state(self.architectures) def _verify_tokenizer_mode(self) -> None: tokenizer_mode = self.tokenizer_mode.lower() if tokenizer_mode not in ["auto", "slow", "mistral", "custom"]: raise ValueError( f"Unknown tokenizer mode: {self.tokenizer_mode}. Must be " "either 'auto', 'slow', 'mistral' or 'custom'.") self.tokenizer_mode = tokenizer_mode def _get_preferred_task( self, architectures: list[str], supported_tasks: set[_ResolvedTask], ) -> Optional[_ResolvedTask]: model_id = self.model if get_pooling_config(model_id, self.revision): return "embed" if self.registry.is_cross_encoder_model(architectures): return "score" if self.registry.is_transcription_model(architectures): return "transcription" suffix_to_preferred_task: list[tuple[str, _ResolvedTask]] = [ # Other models follow this pattern ("ForCausalLM", "generate"), ("ForConditionalGeneration", "generate"), ("ForSequenceClassification", "classify"), ("ChatModel", "generate"), ("LMHeadModel", "generate"), ("EmbeddingModel", "embed"), ("RewardModel", "reward"), ] _, arch = self.registry.inspect_model_cls(architectures) for suffix, pref_task in suffix_to_preferred_task: if arch.endswith(suffix) and pref_task in supported_tasks: return pref_task return None def _resolve_task( self, task_option: Union[TaskOption, Literal["draft"]], ) -> tuple[set[_ResolvedTask], _ResolvedTask]: if task_option == "draft": return {"draft"}, "draft" registry = self.registry architectures = self.architectures runner_support: dict[RunnerType, bool] = { # NOTE: Listed from highest to lowest priority, # in case the model supports multiple of them "transcription": registry.is_transcription_model(architectures), "generate": registry.is_text_generation_model(architectures), "pooling": registry.is_pooling_model(architectures), } supported_runner_types_lst: list[RunnerType] = [ runner_type for runner_type, is_supported in runner_support.items() if is_supported ] supported_tasks_lst: list[_ResolvedTask] = [ task for runner_type in supported_runner_types_lst for task in _RUNNER_TASKS[runner_type] ] supported_tasks = set(supported_tasks_lst) if task_option == "auto": selected_task = next(iter(supported_tasks_lst)) if len(supported_tasks_lst) > 1: preferred_task = self._get_preferred_task( architectures, supported_tasks) if preferred_task is not None: selected_task = preferred_task logger.info( "This model supports multiple tasks: %s. " "Defaulting to '%s'.", supported_tasks, selected_task) else: # Aliases if task_option == "embedding": preferred_task = self._get_preferred_task( architectures, supported_tasks) if preferred_task != "embed": msg = ("The 'embedding' task will be restricted to " "embedding models in a future release. Please " "pass `--task classify`, `--task score`, or " "`--task reward` explicitly for other pooling " "models.") warnings.warn(msg, DeprecationWarning, stacklevel=2) task_option = preferred_task or "embed" if task_option not in supported_tasks: msg = ( f"This model does not support the '{task_option}' task. " f"Supported tasks: {supported_tasks}") raise ValueError(msg) selected_task = task_option return supported_tasks, selected_task def _parse_quant_hf_config(self): quant_cfg = getattr(self.hf_config, "quantization_config", None) if quant_cfg is None: # compressed-tensors uses a "compression_config" key quant_cfg = getattr(self.hf_config, "compression_config", None) return quant_cfg def _verify_quantization(self) -> None: supported_quantization = QUANTIZATION_METHODS optimized_quantization_methods = [ "fp8", "marlin", "modelopt", "gptq_marlin_24", "gptq_marlin", "awq_marlin", "fbgemm_fp8", "compressed_tensors", "compressed-tensors", "experts_int8", "quark", "nvfp4" ] if self.quantization is not None: self.quantization = self.quantization.lower() # Parse quantization method from the HF model config, if available. quant_cfg = self._parse_quant_hf_config() if quant_cfg is not None: quant_method = quant_cfg.get("quant_method", "").lower() # Detect which checkpoint is it for name in QUANTIZATION_METHODS: method = get_quantization_config(name) quantization_override = method.override_quantization_method( quant_cfg, self.quantization) if quantization_override: quant_method = quantization_override self.quantization = quantization_override break # Verify quantization configurations. if self.quantization is None: self.quantization = quant_method elif self.quantization != quant_method: raise ValueError( "Quantization method specified in the model config " f"({quant_method}) does not match the quantization " f"method specified in the `quantization` argument " f"({self.quantization}).") if self.quantization is not None: if self.quantization not in supported_quantization: raise ValueError( f"Unknown quantization method: {self.quantization}. Must " f"be one of {supported_quantization}.") from vllm.platforms import current_platform current_platform.verify_quantization(self.quantization) if self.quantization not in optimized_quantization_methods: logger.warning( "%s quantization is not fully " "optimized yet. The speed can be slower than " "non-quantized models.", self.quantization) def _verify_cuda_graph(self) -> None: if self.max_seq_len_to_capture is None: self.max_seq_len_to_capture = self.max_model_len self.max_seq_len_to_capture = min(self.max_seq_len_to_capture, self.max_model_len) def _verify_bnb_config(self) -> None: """ The current version of bitsandbytes (0.45.3) with 8-bit models does not yet support CUDA graph. # TODO Remove this when bitsandbytes supports. """ is_bitsandbytes = self.quantization == "bitsandbytes" has_quantization_config = (getattr(self.hf_config, "quantization_config", None) is not None) is_8bit = (self.hf_config.quantization_config.get( "load_in_8bit", False) if has_quantization_config else False) if all([ is_bitsandbytes, has_quantization_config, is_8bit, not self.enforce_eager, ]): logger.warning( "CUDA graph is not supported on BitsAndBytes 8bit yet, " "fallback to the eager mode.") self.enforce_eager = True def _verify_with_expert_parallelism(self) -> None: num_expert_names = [ "moe_num_experts", # Dbrx "num_experts", # Jamba "n_routed_experts", # DeepSeek "num_local_experts", # Mixtral ] num_experts = 0 for name in num_expert_names: num_experts = getattr(self.hf_text_config, name, 0) if num_experts > 0: break if num_experts < 1: raise ValueError( "Number of experts in the model must be greater than 0 " "when expert parallelism is enabled.") def verify_async_output_proc(self, parallel_config, speculative_config, device_config) -> None: if not self.use_async_output_proc: # Nothing to check return if parallel_config.pipeline_parallel_size > 1: self.use_async_output_proc = False return # Reminder: Please update docs/source/features/compatibility_matrix.md # If the feature combo become valid from vllm.platforms import current_platform if not current_platform.is_async_output_supported(self.enforce_eager): self.use_async_output_proc = False return if envs.VLLM_USE_RAY_SPMD_WORKER: self.use_async_output_proc = False return # Async postprocessor is not necessary for pooling models # since there is no token generation if self.runner_type == "pooling": self.use_async_output_proc = False # Reminder: Please update docs/source/features/compatibility_matrix.md # If the feature combo become valid if speculative_config: self.use_async_output_proc = False def verify_with_parallel_config( self, parallel_config: "ParallelConfig", ) -> None: total_num_attention_heads = getattr(self.hf_text_config, "num_attention_heads", 0) tensor_parallel_size = parallel_config.tensor_parallel_size if total_num_attention_heads % tensor_parallel_size != 0: raise ValueError( f"Total number of attention heads ({total_num_attention_heads})" " must be divisible by tensor parallel size " f"({tensor_parallel_size}).") if parallel_config.enable_expert_parallel: self._verify_with_expert_parallelism() pipeline_parallel_size = parallel_config.pipeline_parallel_size if pipeline_parallel_size > 1: if not self.registry.is_pp_supported_model(self.architectures): raise NotImplementedError( "Pipeline parallelism is not supported for this model. " "Supported models implement the `SupportsPP` interface.") if self.use_async_output_proc: self.use_async_output_proc = False def get_hf_config_sliding_window( self) -> Union[Optional[int], list[Optional[int]]]: """Get the sliding window size, or None if disabled.""" # Some models, like Qwen2 and Qwen1.5, use `use_sliding_window` in # addition to sliding window size. We check if that field is present # and if it's False, return None. if (hasattr(self.hf_text_config, "use_sliding_window") and not self.hf_text_config.use_sliding_window): return None return getattr(self.hf_text_config, "sliding_window", None) def get_sliding_window(self) -> Optional[Union[int, list[Optional[int]]]]: """Get the sliding window size, or None if disabled. """ # If user disables sliding window, return None. if self.disable_sliding_window: return None # Otherwise get the value from the hf config. return self.get_hf_config_sliding_window() def get_vocab_size(self) -> int: return self.hf_text_config.vocab_size def get_hidden_size(self) -> int: return self.hf_text_config.hidden_size @property def is_deepseek_mla(self) -> bool: if not hasattr(self.hf_text_config, "model_type"): return False elif self.hf_text_config.model_type in \ ('deepseek_v2', 'deepseek_v3', 'deepseek_mtp'): return self.hf_text_config.kv_lora_rank is not None elif self.hf_text_config.model_type == 'eagle': # if the model is an EAGLE module, check for the # underlying architecture return self.hf_text_config.model.model_type in \ ('deepseek_v2', 'deepseek_v3') \ and self.hf_text_config.kv_lora_rank is not None return False def get_head_size(self) -> int: # TODO remove hard code if self.is_deepseek_mla: qk_rope_head_dim = getattr(self.hf_text_config, "qk_rope_head_dim", 0) if self.use_mla: return self.hf_text_config.kv_lora_rank + qk_rope_head_dim else: qk_nope_head_dim = getattr(self.hf_text_config, "qk_nope_head_dim", 0) if qk_rope_head_dim and qk_nope_head_dim: return qk_rope_head_dim + qk_nope_head_dim if hasattr(self.hf_text_config, "model_type") and (self.hf_text_config.model_type == "zamba2"): return self.hf_text_config.attention_head_dim if self.is_attention_free: return 0 if hasattr(self.hf_text_config, "head_dim"): return self.hf_text_config.head_dim # FIXME(woosuk): This may not be true for all models. return (self.hf_text_config.hidden_size // self.hf_text_config.num_attention_heads) def get_total_num_kv_heads(self) -> int: """Returns the total number of KV heads.""" # For GPTBigCode & Falcon: # NOTE: for falcon, when new_decoder_architecture is True, the # multi_query flag is ignored and we use n_head_kv for the number of # KV heads. falcon_model_types = ["falcon", "RefinedWeb", "RefinedWebModel"] new_decoder_arch_falcon = ( self.hf_config.model_type in falcon_model_types and getattr(self.hf_config, "new_decoder_architecture", False)) if not new_decoder_arch_falcon and getattr(self.hf_text_config, "multi_query", False): # Multi-query attention, only one KV head. # Currently, tensor parallelism is not supported in this case. return 1 # For DBRX and MPT if self.hf_config.model_type == "mpt": if "kv_n_heads" in self.hf_config.attn_config: return self.hf_config.attn_config["kv_n_heads"] return self.hf_config.num_attention_heads if self.hf_config.model_type == "dbrx": return getattr(self.hf_config.attn_config, "kv_n_heads", self.hf_config.num_attention_heads) if self.hf_config.model_type == "nemotron-nas": for block in self.hf_config.block_configs: if not block.attention.no_op: return self.hf_config.num_attention_heads \ // block.attention.n_heads_in_group raise RuntimeError("Couldn't determine number of kv heads") if self.is_attention_free: return 0 attributes = [ # For Falcon: "n_head_kv", "num_kv_heads", # For LLaMA-2: "num_key_value_heads", # For ChatGLM: "multi_query_group_num", ] for attr in attributes: num_kv_heads = getattr(self.hf_text_config, attr, None) if num_kv_heads is not None: return num_kv_heads # For non-grouped-query attention models, the number of KV heads is # equal to the number of attention heads. return self.hf_text_config.num_attention_heads def get_num_kv_heads(self, parallel_config: "ParallelConfig") -> int: """Returns the number of KV heads per GPU.""" if self.use_mla: # When using MLA during decode it becomes MQA return 1 total_num_kv_heads = self.get_total_num_kv_heads() # If tensor parallelism is used, we divide the number of KV heads by # the tensor parallel size. We will replicate the KV heads in the # case where the number of KV heads is smaller than the tensor # parallel size so each GPU has at least one KV head. return max(1, total_num_kv_heads // parallel_config.tensor_parallel_size) def get_num_attention_heads(self, parallel_config: "ParallelConfig") -> int: num_heads = getattr(self.hf_text_config, "num_attention_heads", 0) return num_heads // parallel_config.tensor_parallel_size def get_layers_start_end_indices( self, parallel_config: "ParallelConfig") -> tuple[int, int]: from vllm.distributed.utils import get_pp_indices if self.hf_text_config.model_type == "deepseek_mtp": total_num_hidden_layers = getattr(self.hf_text_config, "num_nextn_predict_layers", 0) else: total_num_hidden_layers = getattr(self.hf_text_config, "num_hidden_layers", 0) # the layout order is: DP x PP x TP pp_rank = (parallel_config.rank // parallel_config.tensor_parallel_size ) % parallel_config.pipeline_parallel_size pp_size = parallel_config.pipeline_parallel_size start, end = get_pp_indices(total_num_hidden_layers, pp_rank, pp_size) return start, end def get_num_layers(self, parallel_config: "ParallelConfig") -> int: start, end = self.get_layers_start_end_indices(parallel_config) return end - start def get_num_layers_by_block_type( self, parallel_config: "ParallelConfig", block_type: LayerBlockType = LayerBlockType.attention, ) -> int: # This function relies on 'layers_block_type' in hf_config, # for w/o this attribute, we will need to have workarounds like so attn_block_type = block_type == LayerBlockType.attention is_transformer = not self.is_hybrid and \ not self.has_noops and \ not self.is_attention_free start, end = self.get_layers_start_end_indices(parallel_config) if is_transformer: # Handle the basic case first return end - start if attn_block_type else 0 elif self.is_attention_free: # Attention free # Note that this code assumes there # is only one type of attention-free block type. return 0 if attn_block_type else end - start elif self.has_noops: block_configs = self.hf_config.block_configs return sum(not bc.attention.no_op for bc in block_configs[start:end]) else: # Hybrid model layers_block_type_value = getattr(self.hf_config, "layers_block_type", None) if layers_block_type_value is None: raise ValueError("The model is an hybrid without a " "layers_block_type in the hf_config, " "cannot determine the num of " f"{block_type.value} layers") if hasattr(self.hf_text_config, "model_type") and (self.hf_text_config.model_type == "zamba2"): if attn_block_type: return sum(t == "hybrid" for t in layers_block_type_value[start:end]) else: return self.get_num_layers(parallel_config) return sum(t == block_type.value for t in layers_block_type_value[start:end]) def get_multimodal_config(self) -> "MultiModalConfig": """ Get the multimodal configuration of the model. Raises: ValueError: If the model is not multimodal. """ if self.multimodal_config is None: raise ValueError("The model is not multimodal.") return self.multimodal_config def try_get_generation_config(self) -> dict[str, Any]: if self.generation_config in ("auto", "vllm"): config = try_get_generation_config( self.hf_config_path or self.model, trust_remote_code=self.trust_remote_code, revision=self.revision, ) else: config = try_get_generation_config( self.generation_config, trust_remote_code=self.trust_remote_code, ) if config is None: return {} return config.to_diff_dict() def get_diff_sampling_param(self) -> dict[str, Any]: """ This method returns a dictionary containing the parameters that differ from the default sampling parameters. If `generation_config` is `"vllm"`, an empty dictionary is returned. Returns: dict[str, Any]: A dictionary with the differing sampling parameters, if `generation_config` is `"vllm"` an empty dictionary. """ if self.generation_config == "vllm": config = {} else: config = self.try_get_generation_config() # Overriding with given generation config config.update(self.override_generation_config) available_params = [ "repetition_penalty", "temperature", "top_k", "top_p", "min_p", "max_new_tokens", ] if any(p in config for p in available_params): diff_sampling_param = { p: config.get(p) for p in available_params if config.get(p) is not None } # Huggingface definition of max_new_tokens is equivalent # to vLLM's max_tokens if "max_new_tokens" in diff_sampling_param: diff_sampling_param["max_tokens"] = diff_sampling_param.pop( "max_new_tokens") else: diff_sampling_param = {} if diff_sampling_param: logger.warning_once( "Default sampling parameters have been overridden by the " "model's Hugging Face generation config recommended from the " "model creator. If this is not intended, please relaunch " "vLLM instance with `--generation-config vllm`.") return diff_sampling_param @property def is_encoder_decoder(self) -> bool: """Extract the HF encoder/decoder model flag.""" return is_encoder_decoder(self.hf_config) @property def uses_mrope(self) -> bool: return uses_mrope(self.hf_config) @property def is_multimodal_model(self) -> bool: return self.multimodal_config is not None @property def is_cross_encoder(self) -> bool: return self.registry.is_cross_encoder_model(self.architectures) @property def use_mla(self) -> bool: return self.is_deepseek_mla and not envs.VLLM_MLA_DISABLE @property def supported_runner_types(self) -> set[RunnerType]: return {_TASK_RUNNER[task] for task in self.supported_tasks} @property def runner_type(self) -> RunnerType: return _TASK_RUNNER[self.task] @property def is_v1_compatible(self) -> bool: architectures = getattr(self.hf_config, "architectures", []) return ModelRegistry.is_v1_compatible(architectures) class CacheConfig: """Configuration for the KV cache. Args: block_size: Size of a cache block in number of tokens. gpu_memory_utilization: Fraction of GPU memory to use for the vLLM execution. swap_space: Size of the CPU swap space per GPU (in GiB). cache_dtype: Data type for kv cache storage. is_attention_free: Whether the model is attention-free. num_gpu_blocks_override: Number of GPU blocks to use. This overrides the profiled num_gpu_blocks if specified. Does nothing if None. sliding_window: Sliding window size for the KV cache. Can not work with prefix caching enabled. enable_prefix_caching: Whether to enable prefix caching. cpu_offload_gb: Size of the CPU offload buffer in GiB. """ def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ factors: list[Any] = [] factors.append(self.cache_dtype) # `cpu_offload_gb` does not use `torch.compile` yet. hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __init__( self, block_size: int, gpu_memory_utilization: float, swap_space: float, cache_dtype: str, is_attention_free: bool = False, num_gpu_blocks_override: Optional[int] = None, sliding_window: Optional[int] = None, enable_prefix_caching: bool = False, prefix_caching_hash_algo: str = "builtin", cpu_offload_gb: float = 0, calculate_kv_scales: Optional[bool] = None, ) -> None: self.block_size = block_size self.gpu_memory_utilization = gpu_memory_utilization self.swap_space_bytes = swap_space * GiB_bytes self.num_gpu_blocks_override = num_gpu_blocks_override self.cache_dtype = cache_dtype self.is_attention_free = is_attention_free self.sliding_window = sliding_window self.enable_prefix_caching = enable_prefix_caching self.prefix_caching_hash_algo = prefix_caching_hash_algo self.cpu_offload_gb = cpu_offload_gb self.calculate_kv_scales = calculate_kv_scales self._verify_args() self._verify_cache_dtype() self._verify_prefix_caching() # Will be set after profiling. self.num_gpu_blocks: Optional[int] = None self.num_cpu_blocks: Optional[int] = None # Set calculate_kv_scales to False if the value is unset. if self.calculate_kv_scales is None: self.calculate_kv_scales = False def metrics_info(self): # convert cache_config to dict(key: str, value: str) for prometheus # metrics info return {key: str(value) for key, value in self.__dict__.items()} def _verify_args(self) -> None: if self.cpu_offload_gb < 0: raise ValueError("CPU offload space must be non-negative" f", but got {self.cpu_offload_gb}") if self.gpu_memory_utilization > 1.0: raise ValueError( "GPU memory utilization must be less than 1.0. Got " f"{self.gpu_memory_utilization}.") def _verify_cache_dtype(self) -> None: if self.cache_dtype == "auto": pass elif self.cache_dtype in ("fp8", "fp8_e4m3", "fp8_e5m2"): logger.info( "Using fp8 data type to store kv cache. It reduces the GPU " "memory footprint and boosts the performance. " "Meanwhile, it may cause accuracy drop without a proper " "scaling factor") else: raise ValueError(f"Unknown kv cache dtype: {self.cache_dtype}") def _verify_prefix_caching(self) -> None: if not self.enable_prefix_caching: return if self.sliding_window is not None and not envs.VLLM_USE_V1: raise NotImplementedError( "Prefix caching is not supported with sliding window. " "Run with --disable-sliding-window to use prefix caching.") if self.enable_prefix_caching and self.prefix_caching_hash_algo not in ( "builtin", "sha256"): raise ValueError( "Unknown prefix caching hash algorithm: " f"{self.prefix_caching_hash_algo}. Must be either " "'builtin' or 'sha256'.") def verify_with_parallel_config( self, parallel_config: "ParallelConfig", ) -> None: total_cpu_memory = get_cpu_memory() # FIXME(woosuk): Here, it is assumed that the GPUs in a tensor parallel # group are in the same node. However, the GPUs may span multiple nodes. num_gpus_per_node = parallel_config.tensor_parallel_size cpu_memory_usage = self.swap_space_bytes * num_gpus_per_node msg = (f"{cpu_memory_usage / GiB_bytes:.2f} GiB out of the " f"{total_cpu_memory / GiB_bytes:.2f} GiB total CPU memory " "is allocated for the swap space.") if cpu_memory_usage > 0.7 * total_cpu_memory: raise ValueError("Too large swap space. " + msg) elif cpu_memory_usage > 0.4 * total_cpu_memory: logger.warning("Possibly too large swap space. %s", msg) @dataclass class TokenizerPoolConfig: """Configuration for the tokenizer pool. Args: pool_size: Number of tokenizer workers in the pool. pool_type: Type of the pool. extra_config: Additional config for the pool. The way the config will be used depends on the pool type. """ pool_size: int pool_type: Union[str, type["BaseTokenizerGroup"]] extra_config: dict def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __post_init__(self): if self.pool_type not in ("ray", ) and not isinstance( self.pool_type, type): raise ValueError(f"Unknown pool type: {self.pool_type}") if not isinstance(self.extra_config, dict): raise ValueError("extra_config must be a dictionary.") @classmethod def create_config( cls, tokenizer_pool_size: int, tokenizer_pool_type: Union[str, type["BaseTokenizerGroup"]], tokenizer_pool_extra_config: Optional[Union[str, dict]] ) -> Optional["TokenizerPoolConfig"]: """Create a TokenizerPoolConfig from the given parameters. If tokenizer_pool_size is 0, return None. Args: tokenizer_pool_size: Number of tokenizer workers in the pool. tokenizer_pool_type: Type of the pool. tokenizer_pool_extra_config: Additional config for the pool. The way the config will be used depends on the pool type. This can be a JSON string (will be parsed). """ if tokenizer_pool_size: if isinstance(tokenizer_pool_extra_config, str): tokenizer_pool_extra_config_parsed = json.loads( tokenizer_pool_extra_config) else: tokenizer_pool_extra_config_parsed = ( tokenizer_pool_extra_config or {}) tokenizer_pool_config = cls(tokenizer_pool_size, tokenizer_pool_type, tokenizer_pool_extra_config_parsed) else: tokenizer_pool_config = None return tokenizer_pool_config class LoadFormat(str, enum.Enum): AUTO = "auto" PT = "pt" SAFETENSORS = "safetensors" NPCACHE = "npcache" DUMMY = "dummy" TENSORIZER = "tensorizer" SHARDED_STATE = "sharded_state" GGUF = "gguf" BITSANDBYTES = "bitsandbytes" MISTRAL = "mistral" RUNAI_STREAMER = "runai_streamer" FASTSAFETENSORS = "fastsafetensors" @dataclass class LoadConfig: """ download_dir: Directory to download and load the weights, default to the default cache directory of huggingface. load_format: The format of the model weights to load: "auto" will try to load the weights in the safetensors format and fall back to the pytorch bin format if safetensors format is not available. "pt" will load the weights in the pytorch bin format. "safetensors" will load the weights in the safetensors format. "npcache" will load the weights in pytorch format and store a numpy cache to speed up the loading. "dummy" will initialize the weights with random values, which is mainly for profiling. "tensorizer" will use CoreWeave's tensorizer library for fast weight loading. "bitsandbytes" will load nf4 type weights. "sharded_state" will load weights from pre-sharded checkpoint files, supporting efficient loading of tensor-parallel models. "gguf" will load weights from GGUF format files. "mistral" will load weights from consolidated safetensors files used by Mistral models. "runai_streamer" will load weights from RunAI streamer format files. model_loader_extra_config: The extra config for the model loader. ignore_patterns: The list of patterns to ignore when loading the model. Default to "original/**/*" to avoid repeated loading of llama's checkpoints. use_tqdm_on_load: Whether to enable tqdm for showing progress bar during loading. Default to True """ load_format: Union[str, LoadFormat, "BaseModelLoader"] = LoadFormat.AUTO download_dir: Optional[str] = None model_loader_extra_config: Optional[Union[str, dict]] = field( default_factory=dict) ignore_patterns: Optional[Union[list[str], str]] = None use_tqdm_on_load: bool = True def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __post_init__(self): model_loader_extra_config = self.model_loader_extra_config or {} if isinstance(model_loader_extra_config, str): self.model_loader_extra_config = json.loads( model_loader_extra_config) if isinstance(self.load_format, str): load_format = self.load_format.lower() self.load_format = LoadFormat(load_format) if self.ignore_patterns is not None and len(self.ignore_patterns) > 0: logger.info( "Ignoring the following patterns when downloading weights: %s", self.ignore_patterns) else: self.ignore_patterns = ["original/**/*"] @dataclass class ParallelConfig: """Configuration for the distributed execution.""" pipeline_parallel_size: int = 1 # Number of pipeline parallel groups. tensor_parallel_size: int = 1 # Number of tensor parallel groups. data_parallel_size: int = 1 # Number of data parallel groups. data_parallel_rank: int = 0 # Rank of the data parallel group. # Local rank of the data parallel group, defaults to global rank. data_parallel_rank_local: Optional[int] = None # IP of the data parallel master. data_parallel_master_ip: str = "127.0.0.1" data_parallel_master_port: int = 29500 # Port of the data parallel master. enable_expert_parallel: bool = False # Use EP instead of TP for MoE layers. # Maximum number of multiple batches # when load model sequentially. To avoid RAM OOM when using tensor # parallel and large models. max_parallel_loading_workers: Optional[int] = None # Disable the custom all-reduce kernel and fall back to NCCL. disable_custom_all_reduce: bool = False # Config for the tokenizer pool. If None, will use synchronous tokenization. tokenizer_pool_config: Optional[TokenizerPoolConfig] = None # Whether to profile Ray workers with nsight, see https://docs.ray.io/en/latest/ray-observability/user-guides/profiling.html#profiling-nsight-profiler. ray_workers_use_nsight: bool = False # ray distributed model workers placement group. placement_group: Optional["PlacementGroup"] = None # Backend to use for distributed model # workers, either "ray" or "mp" (multiprocessing). If the product # of pipeline_parallel_size and tensor_parallel_size is less than # or equal to the number of GPUs available, "mp" will be used to # keep processing on a single host. Otherwise, this will default # to "ray" if Ray is installed and fail otherwise. Note that tpu # and hpu only support Ray for distributed inference. distributed_executor_backend: Optional[Union[str, type["ExecutorBase"]]] = None # the full name of the worker class to use. If "auto", the worker class # will be determined based on the platform. worker_cls: str = "auto" sd_worker_cls: str = "auto" worker_extension_cls: str = "" # world_size is TPxPP, it affects the number of workers we create. world_size: int = field(init=False) # world_size_across_dp is TPxPPxDP, it is the size of the world # including data parallelism. world_size_across_dp: int = field(init=False) rank: int = 0 def get_next_dp_init_port(self) -> int: """ We might need to initialize process groups in multiple processes that is related to data parallelism, e.g. both in the worker and in the engine, which can live in different processes. To avoid port conflicts, we increment the port number each time we need to initialize a new process group related to data parallelism. """ answer = self.data_parallel_master_port self.data_parallel_master_port += 1 return answer def stateless_init_dp_group(self) -> "ProcessGroup": from vllm.distributed.utils import ( stateless_init_torch_distributed_process_group) # use gloo since the engine process might not have cuda device dp_group = stateless_init_torch_distributed_process_group( self.data_parallel_master_ip, self.get_next_dp_init_port(), self.data_parallel_rank, self.data_parallel_size, backend="gloo") return dp_group @staticmethod def has_unfinished_dp(dp_group: "ProcessGroup", has_unfinished: bool) -> bool: tensor = torch.tensor([has_unfinished], dtype=torch.int32, device="cpu") # dp rank 0: has_unfinished_seqs=True # dp rank 1: has_unfinished_seqs=False # aggregated: has_unfinished_seqs=True # so this is an OR operation, i.e. MAX in integers torch.distributed.all_reduce(tensor, op=ReduceOp.MAX, group=dp_group) aggregated_has_unfinished = bool(tensor.item()) return aggregated_has_unfinished def compute_hash(self): """ Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ factors: list[Any] = [] factors.append(self.pipeline_parallel_size) factors.append(self.tensor_parallel_size) return hashlib.sha256(str(factors).encode()).hexdigest() def __post_init__(self) -> None: self.world_size = self.pipeline_parallel_size * \ self.tensor_parallel_size if self.data_parallel_size > 1: # Data parallel was specified in the engine args. self.data_parallel_master_port = get_open_port() # TODO multi-node else: # Otherwise fall back to env vars (e.g. for offline SPMD case). self.data_parallel_size = envs.VLLM_DP_SIZE self.data_parallel_rank = envs.VLLM_DP_RANK self.data_parallel_rank_local = envs.VLLM_DP_RANK_LOCAL self.data_parallel_master_ip = envs.VLLM_DP_MASTER_IP self.data_parallel_master_port = envs.VLLM_DP_MASTER_PORT self.world_size_across_dp = self.world_size * self.data_parallel_size if self.distributed_executor_backend == "external_launcher": import os os.environ["VLLM_ENABLE_V1_MULTIPROCESSING"] = "0" logger.info("Disabling V1 multiprocessing for external launcher.") ray_only_devices: list[str] = [] from vllm.platforms import current_platform if (current_platform.device_type in ray_only_devices and self.world_size > 1): if self.distributed_executor_backend is None: self.distributed_executor_backend = "ray" if self.distributed_executor_backend != "ray": raise ValueError( f"{current_platform.device_type.upper()} backend only " "supports Ray for distributed inference.") if self.distributed_executor_backend is None and self.world_size > 1: # We use multiprocessing by default if world_size fits on the # current node and we aren't in a ray placement group. from vllm.executor import ray_utils backend = "mp" ray_found = ray_utils.ray_is_available() if current_platform.is_neuron(): # neuron uses single process to control multiple devices backend = "uni" elif (current_platform.is_cuda() and cuda_device_count_stateless() < self.world_size): if not ray_found: raise ValueError("Unable to load Ray which is " "required for multi-node inference, " "please install Ray with `pip install " "ray`.") from ray_utils.ray_import_err backend = "ray" elif ray_found: if self.placement_group: backend = "ray" else: from ray import is_initialized as ray_is_initialized if ray_is_initialized(): from ray.util import get_current_placement_group if get_current_placement_group(): backend = "ray" self.distributed_executor_backend = backend logger.info("Defaulting to use %s for distributed inference", backend) if self.distributed_executor_backend is None and self.world_size == 1: self.distributed_executor_backend = "uni" self._verify_args() @property def use_ray(self) -> bool: return self.distributed_executor_backend == "ray" or ( isinstance(self.distributed_executor_backend, type) and self.distributed_executor_backend.uses_ray) def _verify_args(self) -> None: # Lazy import to avoid circular import from vllm.executor.executor_base import ExecutorBase from vllm.platforms import current_platform if self.distributed_executor_backend not in ( "ray", "mp", "uni", "external_launcher", None) and not (isinstance( self.distributed_executor_backend, type) and issubclass( self.distributed_executor_backend, ExecutorBase)): raise ValueError( "Unrecognized distributed executor backend " f"{self.distributed_executor_backend}. Supported " "values are 'ray', 'mp' 'uni', 'external_launcher' or" " custom ExecutorBase subclass.") if self.use_ray: from vllm.executor import ray_utils ray_utils.assert_ray_available() device_capability = current_platform.get_device_capability() if (current_platform.is_rocm() and device_capability is not None and device_capability < (9, 4)): self.disable_custom_all_reduce = True logger.info( "Disabled the custom all-reduce kernel because it is not " "supported on AMD GPUs older than MI300X.") if self.ray_workers_use_nsight and not self.use_ray: raise ValueError("Unable to use nsight profiling unless workers " "run with Ray.") assert isinstance(self.worker_extension_cls, str), ( "worker_extension_cls must be a string (qualified class name).") @dataclass class SchedulerConfig: """Scheduler configuration.""" runner_type: str = "generate" # The runner type to launch for the model. # Maximum number of tokens to be processed in a single iteration. max_num_batched_tokens: int = field(default=None) # type: ignore # Maximum number of sequences to be processed in a single iteration. max_num_seqs: int = 128 # Maximum length of a sequence (including prompt and generated text). max_model_len: int = 8192 # Maximum number of sequences that can be partially prefilled concurrently max_num_partial_prefills: int = 1 # Maximum number of "very long prompt" sequences that can be prefilled # concurrently (long is defined by long_prefill_threshold) max_long_partial_prefills: int = 1 # calculate context length that determines which sequences are # considered "long" long_prefill_token_threshold: int = 0 # The number of slots to allocate per sequence per # step, beyond the known token ids. This is used in speculative # decoding to store KV activations of tokens which may or may not be # accepted. num_lookahead_slots: int = 0 # Apply a delay (of delay factor multiplied by previous # prompt latency) before scheduling next prompt. delay_factor: float = 0.0 # If True, prefill requests can be chunked based # on the remaining max_num_batched_tokens. enable_chunked_prefill: bool = False is_multimodal_model: bool = False # NOTE: The following multimodal encoder budget will be initialized to # max_num_batched_tokens and overridden in case max multimodal embedding # size is larger. # TODO (ywang96): Make these configurable. # Multimodal encoder compute budget, only used in V1 max_num_encoder_input_tokens: int = field(default=None) # type: ignore # Multimodal encoder cache size, only used in V1 encoder_cache_size: int = field(default=None) # type: ignore # Whether to perform preemption by swapping or # recomputation. If not specified, we determine the mode as follows: # We use recomputation by default since it incurs lower overhead than # swapping. However, when the sequence group has multiple sequences # (e.g., beam search), recomputation is not currently supported. In # such a case, we use swapping instead. preemption_mode: Optional[str] = None num_scheduler_steps: int = 1 multi_step_stream_outputs: bool = False # Private API. If used, scheduler sends delta data to # workers instead of an entire data. It should be enabled only # when SPMD worker architecture is enabled. I.e., # VLLM_USE_RAY_SPMD_WORKER=1 send_delta_data: bool = False # The scheduling policy to use. "fcfs" (default) or "priority". policy: str = "fcfs" chunked_prefill_enabled: bool = field(init=False) # scheduler class or path. "vllm.core.scheduler.Scheduler" (default) # or "mod.custom_class". scheduler_cls: Union[str, type[object]] = "vllm.core.scheduler.Scheduler" def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __post_init__(self) -> None: if self.max_num_batched_tokens is None: if self.enable_chunked_prefill: if self.num_scheduler_steps > 1: # Multi-step Chunked-Prefill doesn't allow prompt-chunking # for now. Have max_num_batched_tokens set to max_model_len # so we don't reject sequences on account of a short # max_num_batched_tokens. self.max_num_batched_tokens = max( self.max_model_len, _DEFAULT_MAX_NUM_BATCHED_TOKENS) else: self.max_num_batched_tokens = ( _DEFAULT_MAX_NUM_BATCHED_TOKENS) else: # If max_model_len is too short, use # _DEFAULT_MAX_NUM_BATCHED_TOKENS as the default value # for higher throughput. self.max_num_batched_tokens = max( self.max_model_len, _DEFAULT_MAX_NUM_BATCHED_TOKENS) if self.runner_type == "pooling": # Choose specific value for higher throughput self.max_num_batched_tokens = max( self.max_num_batched_tokens, _POOLING_MODEL_MAX_NUM_BATCHED_TOKENS, ) if self.is_multimodal_model: # The value needs to be at least the number of multimodal tokens self.max_num_batched_tokens = max( self.max_num_batched_tokens, _MULTIMODAL_MODEL_MAX_NUM_BATCHED_TOKENS, ) self.max_num_encoder_input_tokens = self.max_num_batched_tokens self.encoder_cache_size = self.max_num_batched_tokens if self.enable_chunked_prefill: logger.info( "Chunked prefill is enabled with max_num_batched_tokens=%d.", self.max_num_batched_tokens) self.chunked_prefill_enabled = self.enable_chunked_prefill if self.max_num_partial_prefills > 1: if self.long_prefill_token_threshold == 0: self.long_prefill_token_threshold = int(self.max_model_len * 0.04) logger.info( "Concurrent partial prefills enabled with " "max_num_partial_prefills=%d, max_long_partial_prefills=%d, " "long_prefill_token_threshold=%d", self.max_num_partial_prefills, self.max_long_partial_prefills, self.long_prefill_token_threshold) self._verify_args() def _verify_args(self) -> None: if (self.max_num_batched_tokens < self.max_model_len and not self.chunked_prefill_enabled): raise ValueError( f"max_num_batched_tokens ({self.max_num_batched_tokens}) is " f"smaller than max_model_len ({self.max_model_len}). " "This effectively limits the maximum sequence length to " "max_num_batched_tokens and makes vLLM reject longer " "sequences. Please increase max_num_batched_tokens or " "decrease max_model_len.") if self.max_num_batched_tokens < self.max_num_seqs: raise ValueError( f"max_num_batched_tokens ({self.max_num_batched_tokens}) must " "be greater than or equal to max_num_seqs " f"({self.max_num_seqs}).") if self.num_lookahead_slots < 0: raise ValueError( "num_lookahead_slots " f"({self.num_lookahead_slots}) must be greater than or " "equal to 0.") if self.num_scheduler_steps < 1: raise ValueError( "num_scheduler_steps " f"({self.num_scheduler_steps}) must be greater than or " "equal to 1.") if self.max_num_partial_prefills < 1: raise ValueError( f"max_num_partial_prefills ({self.max_num_partial_prefills}) " "must be greater than or equal to 1.") elif self.max_num_partial_prefills > 1: if not self.chunked_prefill_enabled: raise ValueError("Chunked prefill must be enabled to set " "max_num_partial_prefills > 1.") if self.long_prefill_token_threshold > self.max_model_len: raise ValueError( "long_prefill_token_threshold " f"({self.long_prefill_token_threshold}) cannot be greater " f"than the max_model_len ({self.max_model_len}).") if (self.max_long_partial_prefills < 1) or (self.max_long_partial_prefills > self.max_num_partial_prefills): raise ValueError( f"max_long_partial_prefills ({self.max_long_partial_prefills}) " "must be greater than or equal to 1 and less than or equal to " f"max_num_partial_prefills ({self.max_num_partial_prefills}).") @property def is_multi_step(self) -> bool: return self.num_scheduler_steps > 1 class DeviceConfig: device: Optional[torch.device] device_type: str def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # the device/platform information will be summarized # by torch/vllm automatically. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __init__(self, device: str = "auto") -> None: if device == "auto": # Automated device type detection from vllm.platforms import current_platform self.device_type = current_platform.device_type if not self.device_type: raise RuntimeError("Failed to infer device type") else: # Device type is assigned explicitly self.device_type = device # Some device types require processing inputs on CPU if self.device_type in ["neuron"]: self.device = torch.device("cpu") elif self.device_type in ["tpu"]: self.device = None else: # Set device with device type self.device = torch.device(self.device_type) @dataclass class SpeculativeConfig: """ Configuration for speculative decoding. Configurable parameters include: - General Speculative Decoding Control: - num_speculative_tokens (int): The number of speculative tokens, if provided. It will default to the number in the draft model config if present, otherwise, it is required. - model (Optional[str]): The name of the draft model, eagle head, or additional weights, if provided. - method (Optional[str]): The name of the speculative method to use. If users provide and set the `model` param, the speculative method type will be detected automatically if possible, if `model` param is not provided, the method name must be provided. - Possible values: - ngram Related additional configuration: - prompt_lookup_max (Optional[int]): Maximum size of ngram token window when using Ngram proposer, required when method is set to ngram. - prompt_lookup_min (Optional[int]): Minimum size of ngram token window when using Ngram proposer, if provided. Defaults to 1. - eagle - medusa - mlp_speculator - draft_model - acceptance_method (str): The method to use for accepting draft tokens. This can take two possible values: 'rejection_sampler' and 'typical_acceptance_sampler' for RejectionSampler and TypicalAcceptanceSampler respectively. If not specified, it defaults to 'rejection_sampler'. - Possible values: - rejection_sampler - typical_acceptance_sampler Related additional configuration: - posterior_threshold (Optional[float]): A threshold value that sets a lower bound on the posterior probability of a token in the target model for it to be accepted. This threshold is used only when we use the TypicalAcceptanceSampler for token acceptance. - posterior_alpha (Optional[float]): Scaling factor for entropy-based threshold, applied when using TypicalAcceptanceSampler. - draft_tensor_parallel_size (Optional[int]): The degree of the tensor parallelism for the draft model. Can only be 1 or the same as the target model's tensor parallel size. - disable_logprobs (bool): If set to True, token log probabilities are not returned during speculative decoding. If set to False, token log probabilities are returned according to the log probability settings in SamplingParams. If not specified, it defaults to True. - Draft Model Configuration: - quantization (Optional[str]): Quantization method that was used to quantize the draft model weights. If None, we assume the model weights are not quantized. Note that it only takes effect when using the draft model-based speculative method. - max_model_len (Optional[int]): The maximum model length of the draft model. Used when testing the ability to skip speculation for some sequences. - revision: The specific model version to use for the draft model. It can be a branch name, a tag name, or a commit id. If unspecified, will use the default version. - code_revision: The specific revision to use for the draft model code on Hugging Face Hub. It can be a branch name, a tag name, or a commit id. If unspecified, will use the default version. - Advanced Control: - disable_mqa_scorer (bool): Disable the MQA scorer and fall back to batch expansion for scoring proposals. If not specified, it defaults to False. - disable_by_batch_size (Optional[int]): Disable speculative decoding for new incoming requests when the number of enqueued requests is larger than this value, if provided. Although the parameters above are structured hierarchically, there is no need to nest them during configuration. Non-configurable internal parameters include: - Model Configuration: - target_model_config (ModelConfig): The configuration of the target model. - draft_model_config (ModelConfig): The configuration of the draft model initialized internal. - Parallelism Configuration: - target_parallel_config (ParallelConfig): The parallel configuration for the target model. - draft_parallel_config (ParallelConfig): The parallel configuration for the draft model initialized internal. - Execution Control: - enable_chunked_prefill (bool): Whether vLLM is configured to use chunked prefill or not. Used for raising an error since it's not yet compatible with speculative decode. - disable_log_stats (bool): Whether to disable the periodic printing of stage times in speculative decoding. """ # speculative configs from cli args num_speculative_tokens: int = field(default=None, init=True) # type: ignore method: Optional[str] = None acceptance_method: str = "rejection_sampler" draft_tensor_parallel_size: Optional[int] = None disable_logprobs: bool = True model: Optional[str] = None quantization: Optional[str] = None max_model_len: Optional[int] = None revision: Optional[str] = None code_revision: Optional[str] = None disable_mqa_scorer: bool = False disable_by_batch_size: Optional[int] = None prompt_lookup_max: Optional[int] = None prompt_lookup_min: Optional[int] = None posterior_threshold: Optional[float] = None posterior_alpha: Optional[float] = None # required configuration params passed from engine target_model_config: ModelConfig = field(default=None, init=True) # type: ignore target_parallel_config: ParallelConfig = field(default=None, init=True) # type: ignore enable_chunked_prefill: bool = field(default=None, init=True) # type: ignore disable_log_stats: bool = field(default=None, init=True) # type: ignore # params generated in the post-init stage draft_model_config: ModelConfig = field(default=None, init=True) # type: ignore draft_parallel_config: ParallelConfig = field(default=None, init=True) # type: ignore def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # spec decode does not use `torch.compile` yet. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str @classmethod def from_dict(cls, dict_value: dict) -> "SpeculativeConfig": """Parse the CLI value for the speculative config.""" return cls(**dict_value) @staticmethod def hf_config_override(hf_config: PretrainedConfig) -> PretrainedConfig: if hf_config.model_type == "deepseek_v3": hf_config.model_type = "deepseek_mtp" if hf_config.model_type == "deepseek_mtp": n_predict = getattr(hf_config, "num_nextn_predict_layers", None) hf_config.update({ "n_predict": n_predict, "architectures": ["DeepSeekMTPModel"] }) return hf_config def __post_init__(self): # Note: "method" is a new parameter that helps to extend the # configuration of non-model-based proposers, and the "model" parameter # will be used to set the draft model, eagle head, or additional weight # when needed. If users do not specify "method", the speculative method # will be detected automatically if possible. If the speculative method # can not be detected, it will be considered as the "draft_model" by # default. if self.model is None and self.num_speculative_tokens is not None: # TODO(Shangming): Refactor mtp configuration logic when supporting # mtp acceleration for more models besides deepseek_v3 if self.target_model_config.hf_text_config.model_type \ == "deepseek_v3": # use the draft model from the same model: self.model = self.target_model_config.model elif self.method in ("ngram", "[ngram]"): self.model = "ngram" else: raise ValueError("num_speculative_tokens was provided without " "speculative model.") # Automatically configure the method for ngram when "model" is used # instead of "method" if self.method is None and (self.model is not None and self.model in ("ngram", "[ngram]")): self.method = "ngram" if self.method in ("ngram", "[ngram]"): # Unified to "ngram" internally self.method = "ngram" # Set default values if not provided if (self.prompt_lookup_min is None and self.prompt_lookup_max is None): # TODO(woosuk): Tune these values. They are arbitrarily chosen. self.prompt_lookup_min = 5 self.prompt_lookup_max = 5 elif self.prompt_lookup_min is None: assert self.prompt_lookup_max is not None self.prompt_lookup_min = self.prompt_lookup_max elif self.prompt_lookup_max is None: assert self.prompt_lookup_min is not None self.prompt_lookup_max = self.prompt_lookup_min # Validate values if self.prompt_lookup_min < 1: raise ValueError( f"prompt_lookup_min={self.prompt_lookup_min} must be > 0") if self.prompt_lookup_max < 1: raise ValueError( f"prompt_lookup_max={self.prompt_lookup_max} must be > 0") if self.prompt_lookup_min > self.prompt_lookup_max: raise ValueError( f"prompt_lookup_min={self.prompt_lookup_min} must " f"be <= prompt_lookup_max={self.prompt_lookup_max}") # TODO: current we still need extract vocab_size from target model # config, in future, we may try refactor it out, and set # draft related config as None here. self.draft_model_config = self.target_model_config self.draft_parallel_config = self.target_parallel_config else: self.prompt_lookup_max = 0 self.prompt_lookup_min = 0 if self.model is not None: self.draft_model_config = ModelConfig( model=self.model, task="draft", tokenizer=self.target_model_config.tokenizer, tokenizer_mode=self.target_model_config.tokenizer_mode, trust_remote_code=self.target_model_config. trust_remote_code, allowed_local_media_path=self.target_model_config. allowed_local_media_path, dtype=self.target_model_config.dtype, seed=self.target_model_config.seed, revision=self.revision, code_revision=self.code_revision, tokenizer_revision=self.target_model_config. tokenizer_revision, max_model_len=None, spec_target_max_model_len=self.target_model_config. max_model_len, quantization=self.quantization, enforce_eager=self.target_model_config.enforce_eager, max_seq_len_to_capture=self.target_model_config. max_seq_len_to_capture, max_logprobs=self.target_model_config.max_logprobs, hf_overrides=SpeculativeConfig.hf_config_override, ) # Automatically detect the method if "eagle-" in self.draft_model_config.model.lower(): self.method = "eagle" elif self.draft_model_config.hf_config.model_type == "medusa": self.method = "medusa" elif (self.draft_model_config.hf_config.model_type == "mlp_speculator"): self.method = "mlp_speculator" else: self.method = "draft_model" # Replace hf_config for EAGLE draft_model if self.method == "eagle": if self.enable_chunked_prefill: raise ValueError( "Chunked prefill and EAGLE are not compatible.") from vllm.transformers_utils.configs.eagle import ( EAGLEConfig) if isinstance(self.draft_model_config.hf_config, EAGLEConfig): pass else: eagle_config = EAGLEConfig( self.draft_model_config.hf_config) self.draft_model_config.hf_config = eagle_config if (self.num_speculative_tokens is not None and hasattr(self.draft_model_config.hf_config, "num_lookahead_tokens")): self.draft_model_config.hf_config.num_lookahead_tokens = \ self.num_speculative_tokens n_predict = getattr(self.draft_model_config.hf_config, "n_predict", None) if n_predict is not None: if self.num_speculative_tokens is None: # Default to max value defined in draft model config. self.num_speculative_tokens = n_predict elif self.num_speculative_tokens > n_predict and \ self.num_speculative_tokens % n_predict != 0: # Ensure divisibility for MTP module reuse. raise ValueError( f"num_speculative_tokens:{self.num_speculative_tokens}" f" must be divisible by {n_predict=}") self.draft_tensor_parallel_size = \ SpeculativeConfig._verify_and_get_draft_tp( self.target_parallel_config, self.draft_tensor_parallel_size, self.draft_model_config.hf_config ) self.draft_model_config.max_model_len = ( SpeculativeConfig._maybe_override_draft_max_model_len( self.max_model_len, self.draft_model_config.max_model_len, self.target_model_config.max_model_len, )) self.draft_parallel_config = ( SpeculativeConfig.create_draft_parallel_config( self.target_parallel_config, self.draft_tensor_parallel_size)) if self.acceptance_method == "typical_acceptance_sampler": if self.posterior_threshold is None: self.posterior_threshold = 0.09 if self.posterior_alpha is None: self.posterior_alpha = 0.3 self._verify_args() @staticmethod def _maybe_override_draft_max_model_len( speculative_max_model_len: Optional[int], draft_max_model_len: int, target_max_model_len: int, ) -> int: """Determine the max sequence len for the draft model. This is usually the draft_max_model_len, but may be the target_max_model_len if it is less than the draft_max_model_len, or may be speculative_max_model_len if it is specified. This is necessary so that sequences do not exceed the capacity of the draft model or the target model. speculative_max_model_len is mainly used for testing that sequences can skip speculation. """ if speculative_max_model_len is not None: if speculative_max_model_len > draft_max_model_len: raise ValueError(f"{speculative_max_model_len=} cannot be " f"larger than {draft_max_model_len=}") if speculative_max_model_len > target_max_model_len: raise ValueError(f"{speculative_max_model_len=} cannot be " f"larger than {target_max_model_len=}") return speculative_max_model_len return min( draft_max_model_len, target_max_model_len, ) @staticmethod def _verify_and_get_draft_tp( target_parallel_config: ParallelConfig, speculative_draft_tensor_parallel_size: Optional[int], draft_hf_config: PretrainedConfig) -> int: """ Verifies and adjusts the tensor parallel size for a draft model specified using speculative_draft_tensor_parallel_size. """ # If speculative_draft_tensor_parallel_size is unset then set it # appropriately else verify that it is set correctly. if speculative_draft_tensor_parallel_size is None: if draft_hf_config.model_type == "mlp_speculator": speculative_draft_tensor_parallel_size = 1 if target_parallel_config.tensor_parallel_size > 1: logger.warning( "%s cannot currently be run with tp>1; " "setting speculative_draft_tensor_parallel_size=1", draft_hf_config.model_type) else: speculative_draft_tensor_parallel_size = \ target_parallel_config.tensor_parallel_size elif speculative_draft_tensor_parallel_size not in ( 1, target_parallel_config.tensor_parallel_size): raise ValueError( f"{speculative_draft_tensor_parallel_size=} cannot be " f"other value than 1 or target model tensor_parallel_size") return speculative_draft_tensor_parallel_size @staticmethod def create_draft_parallel_config( target_parallel_config: ParallelConfig, speculative_draft_tensor_parallel_size: int, ) -> ParallelConfig: """Create a parallel config for use by the draft worker. This is mostly a copy of the target parallel config, except the tp_size. """ draft_parallel_config = ParallelConfig( pipeline_parallel_size=target_parallel_config. pipeline_parallel_size, tensor_parallel_size=speculative_draft_tensor_parallel_size, distributed_executor_backend=target_parallel_config. distributed_executor_backend, max_parallel_loading_workers=target_parallel_config. max_parallel_loading_workers, disable_custom_all_reduce=target_parallel_config. disable_custom_all_reduce, tokenizer_pool_config=target_parallel_config.tokenizer_pool_config, ray_workers_use_nsight=target_parallel_config. ray_workers_use_nsight, placement_group=target_parallel_config.placement_group, ) return draft_parallel_config def _verify_args(self) -> None: if self.num_speculative_tokens is None: raise ValueError( "num_speculative_tokens must be provided with " "speculative model unless the draft model config contains an " "n_predict parameter.") if self.num_speculative_tokens <= 0: raise ValueError("Expected num_speculative_tokens to be greater " f"than zero ({self.num_speculative_tokens}).") if self.draft_model_config: self.draft_model_config.verify_with_parallel_config( self.draft_parallel_config) # Validate and set draft token acceptance related settings. if self.acceptance_method is None: raise ValueError("acceptance_method is not set. " "Expected values are rejection_sampler or " "typical_acceptance_sampler.") if (self.acceptance_method != 'rejection_sampler' and self.acceptance_method != 'typical_acceptance_sampler'): raise ValueError( "Expected acceptance_method to be either " "rejection_sampler or typical_acceptance_sampler. Instead it " f"is {self.acceptance_method}") if self.acceptance_method == "typical_acceptance_sampler" and ( (self.posterior_threshold is not None and self.posterior_threshold < 0) or (self.posterior_alpha is not None and self.posterior_alpha < 0)): raise ValueError( "Expected the posterior_threshold and posterior_alpha of " "typical_acceptance_sampler to be > 0. " "Instead found posterior_threshold = " f"{self.posterior_threshold} and posterior_alpha = " f"{self.posterior_alpha}") if (self.disable_by_batch_size is not None and self.disable_by_batch_size < 2): raise ValueError("Expect the batch size threshold of disabling " "speculative decoding is > 1, but got " f"{self.disable_by_batch_size=}") @property def num_lookahead_slots(self) -> int: """The number of additional slots the scheduler should allocate per step, in addition to the slots allocated for each known token. This is equal to the number of speculative tokens, as each speculative token must be scored. """ return self.num_speculative_tokens def __repr__(self) -> str: if self.prompt_lookup_max is not None and self.prompt_lookup_max > 0: draft_model = "ngram" else: draft_model = self.draft_model_config.model num_spec_tokens = self.num_speculative_tokens return f"SpeculativeConfig({draft_model=}, {num_spec_tokens=})" @dataclass class LoRAConfig: max_lora_rank: int max_loras: int fully_sharded_loras: bool = False max_cpu_loras: Optional[int] = None lora_dtype: Optional[Union[torch.dtype, str]] = None lora_extra_vocab_size: int = 256 # This is a constant. lora_vocab_padding_size: ClassVar[int] = 256 long_lora_scaling_factors: Optional[tuple[float]] = None bias_enabled: bool = False def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ factors: list[Any] = [] factors.append(self.max_lora_rank) factors.append(self.max_loras) factors.append(self.fully_sharded_loras) factors.append(self.lora_dtype) factors.append(self.lora_extra_vocab_size) factors.append(self.long_lora_scaling_factors) factors.append(self.bias_enabled) hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __post_init__(self): # Setting the maximum rank to 512 should be able to satisfy the vast # majority of applications. possible_max_ranks = (8, 16, 32, 64, 128, 256, 320, 512) possible_lora_extra_vocab_size = (256, 512) if self.max_lora_rank not in possible_max_ranks: raise ValueError( f"max_lora_rank ({self.max_lora_rank}) must be one of " f"{possible_max_ranks}.") if self.lora_extra_vocab_size not in possible_lora_extra_vocab_size: raise ValueError( f"lora_extra_vocab_size ({self.lora_extra_vocab_size}) " f"must be one of {possible_lora_extra_vocab_size}.") if self.max_loras < 1: raise ValueError(f"max_loras ({self.max_loras}) must be >= 1.") if self.max_cpu_loras is None: self.max_cpu_loras = self.max_loras elif self.max_cpu_loras < self.max_loras: raise ValueError( f"max_cpu_loras ({self.max_cpu_loras}) must be >= " f"max_loras ({self.max_loras})") def verify_with_cache_config(self, cache_config: CacheConfig): # TODO LoRA supports CPU offload. if cache_config.cpu_offload_gb > 0: raise ValueError("CPU offload is not supported with LoRA yet.") def verify_with_model_config(self, model_config: ModelConfig): if self.lora_dtype in (None, "auto"): self.lora_dtype = model_config.dtype elif isinstance(self.lora_dtype, str): self.lora_dtype = getattr(torch, self.lora_dtype) def verify_with_scheduler_config(self, scheduler_config: SchedulerConfig): # Reminder: Please update docs/source/features/compatibility_matrix.md # If the feature combo become valid if scheduler_config.chunked_prefill_enabled: logger.warning("LoRA with chunked prefill is still experimental " "and may be unstable.") @dataclass class PromptAdapterConfig: max_prompt_adapters: int max_prompt_adapter_token: int max_cpu_prompt_adapters: Optional[int] = None prompt_adapter_dtype: Optional[torch.dtype] = None def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __post_init__(self): if self.max_prompt_adapters < 1: raise ValueError(f"max_prompt_adapters " f"({self.max_prompt_adapters}) must be >= 1.") if self.max_prompt_adapter_token == 0: raise ValueError("max_prompt_adapter_token must be set.") if self.max_cpu_prompt_adapters is None: self.max_cpu_prompt_adapters = self.max_prompt_adapters def verify_with_model_config(self, model_config: ModelConfig): if self.prompt_adapter_dtype in (None, "auto"): self.prompt_adapter_dtype = model_config.dtype elif isinstance(self.prompt_adapter_dtype, str): self.prompt_adapter_dtype = getattr(torch, self.prompt_adapter_dtype) @dataclass class MultiModalConfig: """Controls the behavior of multimodal models.""" limit_per_prompt: Mapping[str, int] = field(default_factory=dict) """ The maximum number of input items allowed per prompt for each modality. """ def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def get_limit_per_prompt(self, modality: str) -> int: """ Get the maximum number of input items allowed per prompt for the given modality. If not set by the user, this defaults to `1`. """ return self.limit_per_prompt.get(modality, 1) # TODO: Add configs to init vision tower or not. @dataclass class PoolerConfig: """Controls the behavior of output pooling in pooling models.""" pooling_type: Optional[str] = None """ The pooling method of the pooling model. This should be a key in :class:`vllm.model_executor.layers.pooler.PoolingType`. """ normalize: Optional[bool] = None """ Whether to normalize the pooled outputs. Usually, this should be set to ``True`` for embedding outputs. """ softmax: Optional[bool] = None """ Whether to apply softmax to the pooled outputs. Usually, this should be set to ``True`` for classification outputs. """ step_tag_id: Optional[int] = None """ If set, only the score corresponding to the ``step_tag_id`` in the generated sentence should be returned. Otherwise, the scores for all tokens are returned. """ returned_token_ids: Optional[list[int]] = None """ A list of indices for the vocabulary dimensions to be extracted, such as the token IDs of ``good_token`` and ``bad_token`` in the ``math-shepherd-mistral-7b-prm`` model. """ def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str @staticmethod def from_json(json_str: str) -> "PoolerConfig": return PoolerConfig(**json.loads(json_str)) _STR_DTYPE_TO_TORCH_DTYPE = { "half": torch.float16, "float16": torch.float16, "float": torch.float32, "float32": torch.float32, "bfloat16": torch.bfloat16, } _ROCM_NOT_SUPPORTED_DTYPE: list[str] = [] # def _get_and_verify_dtype( config: PretrainedConfig, dtype: Union[str, torch.dtype], ) -> torch.dtype: # NOTE: getattr(config, "torch_dtype", torch.float32) is not correct # because config.torch_dtype can be None. config_dtype = getattr(config, "torch_dtype", None) # Fallbacks for multi-modal models if the root config # does not define torch_dtype if config_dtype is None and hasattr(config, "text_config"): config_dtype = getattr(config.text_config, "torch_dtype", None) if config_dtype is None and hasattr(config, "vision_config"): config_dtype = getattr(config.vision_config, "torch_dtype", None) if config_dtype is None: config_dtype = torch.float32 if isinstance(dtype, str): dtype = dtype.lower() if dtype == "auto": if config_dtype == torch.float32: # Following common practice, we use float16 for float32 models torch_dtype = torch.float16 else: torch_dtype = config_dtype from vllm.platforms import current_platform if (current_platform.is_cpu() and current_platform.get_cpu_architecture() == CpuArchEnum.POWERPC and (config_dtype == torch.float16 or config_dtype == torch.float32)): logger.info( "For POWERPC, we cast models to bfloat16 instead of " "using float16 by default. Float16 is not currently " "supported for POWERPC.") torch_dtype = torch.bfloat16 # TODO: change this condition to check if the platform support bf16 # instead of checking the OS. For instance M2 shall supports bf16 # already. But we need to modify `cpu_extension.cmake` to activate # the feature in the build. if (current_platform.is_cpu() and sys.platform.startswith("darwin") and current_platform.get_cpu_architecture() == CpuArchEnum.ARM and config_dtype == torch.bfloat16): logger.info("For macOS with Apple Silicon, currently bfloat16 " "is not supported. Setting dtype to float16.") torch_dtype = torch.float16 if current_platform.is_hpu() and config_dtype == torch.float16: logger.info( "For HPU, we cast models to bfloat16 instead of " "using float16 by default. Please specify `dtype` if you " "want to use float16.") torch_dtype = torch.bfloat16 else: if dtype not in _STR_DTYPE_TO_TORCH_DTYPE: raise ValueError(f"Unknown dtype: {dtype}") torch_dtype = _STR_DTYPE_TO_TORCH_DTYPE[dtype] elif isinstance(dtype, torch.dtype): torch_dtype = dtype else: raise ValueError(f"Unknown dtype: {dtype}") # Verify the dtype. if torch_dtype != config_dtype: if torch_dtype == torch.float32: # Upcasting to float32 is allowed. logger.info("Upcasting %s to %s.", config_dtype, torch_dtype) pass elif config_dtype == torch.float32: # Downcasting from float32 to float16 or bfloat16 is allowed. logger.info("Downcasting %s to %s.", config_dtype, torch_dtype) pass else: # Casting between float16 and bfloat16 is allowed with a warning. logger.warning("Casting %s to %s.", config_dtype, torch_dtype) return torch_dtype def _get_and_verify_max_len( hf_config: PretrainedConfig, max_model_len: Optional[int], disable_sliding_window: bool, sliding_window_len: Optional[Union[int, list[Optional[int]]]], spec_target_max_model_len: Optional[int] = None, encoder_config: Optional[Any] = None, ) -> int: """Get and verify the model's maximum length.""" derived_max_model_len = float("inf") possible_keys = [ # OPT "max_position_embeddings", # GPT-2 "n_positions", # MPT "max_seq_len", # ChatGLM2 "seq_length", # Command-R "model_max_length", # Whisper "max_target_positions", # Others "max_sequence_length", "max_seq_length", "seq_len", ] # Choose the smallest "max_length" from the possible keys. max_len_key = None for key in possible_keys: max_len = getattr(hf_config, key, None) if max_len is not None: max_len_key = key if max_len < derived_max_model_len \ else max_len_key derived_max_model_len = min(derived_max_model_len, max_len) # If sliding window is manually disabled, max_length should be less # than the sliding window length in the model config. if disable_sliding_window and sliding_window_len is not None: sliding_window_len_min = get_min_sliding_window(sliding_window_len) max_len_key = "sliding_window" \ if sliding_window_len_min < derived_max_model_len else max_len_key derived_max_model_len = min(derived_max_model_len, sliding_window_len_min) # If none of the keys were found in the config, use a default and # log a warning. if derived_max_model_len == float("inf"): if max_model_len is not None: # If max_model_len is specified, we use it. return max_model_len if spec_target_max_model_len is not None: # If this is a speculative draft model, we use the max model len # from the target model. return spec_target_max_model_len default_max_len = 2048 logger.warning( "The model's config.json does not contain any of the following " "keys to determine the original maximum length of the model: " "%s. Assuming the model's maximum length is %d.", possible_keys, default_max_len) derived_max_model_len = default_max_len rope_scaling = getattr(hf_config, "rope_scaling", None) # NOTE(woosuk): Gemma3's max_model_len (128K) is already scaled by RoPE # scaling, so we skip applying the scaling factor again. if rope_scaling is not None and "gemma3" not in hf_config.model_type: # No need to consider "type" key because of patch_rope_scaling when # loading HF config rope_type = rope_scaling["rope_type"] if rope_type not in ("su", "longrope", "llama3"): if disable_sliding_window: # TODO(robertgshaw): Find a model that supports rope_scaling # with sliding window to see if this case should be allowed. raise NotImplementedError( "Disabling sliding window is not supported for models " "with rope_scaling. Please raise an issue so we can " "investigate.") # NOTE: rope_type == "default" does not define factor # https://github.com/huggingface/transformers/blob/v4.45.2/src/transformers/modeling_rope_utils.py scaling_factor = rope_scaling.get("factor", 1.0) if rope_type == "yarn": derived_max_model_len = rope_scaling[ "original_max_position_embeddings"] derived_max_model_len *= scaling_factor if encoder_config and "max_seq_length" in encoder_config: derived_max_model_len = encoder_config["max_seq_length"] # If the user specified a max length, make sure it is smaller than the # derived length from the HF model config. if max_model_len is None: max_model_len = int(derived_max_model_len) elif max_model_len > derived_max_model_len: # Some models might have a separate key for specifying model_max_length # that will be bigger than derived_max_model_len. We compare user input # with model_max_length and allow this override when it's smaller. model_max_length = getattr(hf_config, "model_max_length", None) if model_max_length is not None and max_model_len <= model_max_length: if disable_sliding_window: # TODO(robertgshaw): Find a model that has model_max_length # with sliding window to see if this case should be allowed. raise NotImplementedError( "Disabling sliding window is not supported for models " "model_max_length in the config. Please raise an issue " "so we can investigate.") else: msg = ( f"User-specified max_model_len ({max_model_len}) is greater " f"than the derived max_model_len ({max_len_key}=" f"{derived_max_model_len} or model_max_length=" f"{model_max_length} in model's config.json). This may lead " "to incorrect model outputs or CUDA errors.") if envs.VLLM_ALLOW_LONG_MAX_MODEL_LEN: logger.warning( "%s Make sure the value is correct and within the " "model context size.", msg) else: raise ValueError( f"{msg} To allow overriding this maximum, set " "the env var VLLM_ALLOW_LONG_MAX_MODEL_LEN=1") return int(max_model_len) def get_min_sliding_window( sliding_window: Union[int, list[Optional[int]]]) -> int: if isinstance(sliding_window, list): return min(s for s in sliding_window if s is not None) return sliding_window def get_served_model_name(model: str, served_model_name: Optional[Union[str, list[str]]]): """ If the input is a non-empty list, the first model_name in `served_model_name` is taken. If the input is a non-empty string, it is used directly. For cases where the input is either an empty string or an empty list, the fallback is to use `self.model`. """ if not served_model_name: return model if isinstance(served_model_name, list): return served_model_name[0] return served_model_name @dataclass class DecodingConfig: """Dataclass which contains the decoding strategy of the engine""" # Which guided decoding algo to use. # 'outlines' / 'lm-format-enforcer' / 'xgrammar' guided_decoding_backend: str = 'xgrammar' reasoning_backend: Optional[str] = None def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __post_init__(self): v0_valid_guided_backends = [ 'outlines', 'lm-format-enforcer', 'xgrammar' ] v1_valid_guided_backends = ['xgrammar', 'guidance', 'auto'] backend = GuidedDecodingParams( backend=self.guided_decoding_backend).backend_name if envs.VLLM_USE_V1: valid_guided_backends = v1_valid_guided_backends else: valid_guided_backends = v0_valid_guided_backends if backend not in valid_guided_backends: raise ValueError(f"Invalid guided_decoding_backend '{backend}'," f" must be one of {valid_guided_backends}") @dataclass class ObservabilityConfig: """Configuration for observability - metrics and tracing.""" show_hidden_metrics: bool = False otlp_traces_endpoint: Optional[str] = None # Collecting detailed timing information for each request can be expensive. # If set, collects the model forward time for the request. collect_model_forward_time: bool = False # If set, collects the model execute time for the request. collect_model_execute_time: bool = False def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str def __post_init__(self): if not is_otel_available() and self.otlp_traces_endpoint is not None: raise ValueError( "OpenTelemetry is not available. Unable to configure " "'otlp_traces_endpoint'. Ensure OpenTelemetry packages are " f"installed. Original error:\n{otel_import_error_traceback}") class KVTransferConfig(BaseModel): """Configuration for distributed KV cache transfer.""" # The KV connector for vLLM to transmit KV caches between vLLM instances. kv_connector: Optional[str] = None # The device used by kv connector to buffer the KV cache. # Currently only support 'cuda'. kv_buffer_device: Optional[str] = "cuda" # The buffer size for TorchDistributedConnector. Measured in number of # bytes. Recommended value: 1e9 (about 1GB). kv_buffer_size: float = 1e9 # Whether this vLLM instance produces, consumes KV cache, or both. Choices # are 'kv_producer', 'kv_consumer', and 'both'. kv_role: Optional[str] = None # The rank of this vLLM instance in the KV cache transfer. Typical value: # 0 for prefill instance, 1 for decode instance. # Currently only 1P1D is supported. kv_rank: Optional[int] = None # The number of parallel instances for KV cache transfer. For # PyNcclConnector, this should be 2. kv_parallel_size: int = 1 # The KV connector ip, used to build distributed connection kv_ip: str = "127.0.0.1" # The KV connector port, used to build distributed connection kv_port: int = 14579 # any extra config that the connector may need kv_connector_extra_config: dict[str, Any] = {} def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ # no factors to consider. # this config will not affect the computation graph. factors: list[Any] = [] hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest() return hash_str @classmethod def from_cli(cls, cli_value: str) -> "KVTransferConfig": """Parse the CLI value for the kv cache transfer config.""" return KVTransferConfig.model_validate_json(cli_value) def model_post_init(self, __context: Any) -> None: if self.kv_role is not None and self.kv_role not in [ "kv_producer", "kv_consumer", "kv_both" ]: raise ValueError( f"Unsupported kv_role: {self.kv_role}. " f"Supported roles are `kv_producer`, `kv_consumer`, " f"and `kv_both`") if self.kv_connector is not None and self.kv_role is None: raise ValueError("Please specify kv_disagg_role when kv_connector " "is set, supported roles are `kv_producer`, " "`kv_consumer`, and `kv_both`") @property def is_kv_transfer_instance(self) -> bool: return self.kv_connector is not None and \ self.kv_role in ["kv_producer", "kv_consumer", "kv_both"] @property def is_kv_producer(self) -> bool: return self.kv_connector is not None and \ self.kv_role in ["kv_producer", "kv_both"] @property def is_kv_consumer(self) -> bool: return self.kv_connector is not None and \ self.kv_role in ["kv_consumer", "kv_both"] def get_from_extra_config(self, key, default) -> Any: return self.kv_connector_extra_config.get(key, default) class CompilationLevel: # constants for the levels of the compilation process NO_COMPILATION = 0 DYNAMO_AS_IS = 1 DYNAMO_ONCE = 2 PIECEWISE = 3 class CompilationConfig(BaseModel): """ Configuration for compilation. It has three parts: - Top-level Compilation control: - level: the level of compilation. - 0: no compilation. - 1: dynamo as is. - 2: dynamo once. - 3: piecewise compilation. - debug_dump_path: the path to dump the debug information. - cache_dir: the directory to store the compiled graph, to accelerate Inductor compilation. By default, it will use model-related information to generate a cache directory. - backend: the backend for compilation. It needs to be a string. - "" (empty string): use the default backend. - "eager"/"openxla"/...: use the specified backend registered in PyTorch. - "full.module.name": a qualified name which can be used to import the backend function. We use string to avoid serialization issues when using compilation in a distributed setting. When the compilation level is 1 or 2, the backend is used for the compilation directly (it sees the whole graph). When the compilation level is 3, the backend is used for the piecewise compilation (it sees a part of the graph). - custom_ops: fine-grained control over which custom ops to enable/disable. Use 'all' to enable all, 'none' to disable all. Also specify a list of custom op names to enable (prefixed with a '+'), or disable (prefixed with a '-'). Examples: - 'all,-op1' to enable all except op1 - 'none,+op1,+op2' to enable only op1 and op2 By default, all custom ops are enabled when running without Inductor and disabled when running with Inductor (compile_level >= Inductor). - splitting_ops: a list of ops to split the full graph into subgraphs, used in piecewise compilation. - CudaGraph capture: - use_cudagraph: whether to use cudagraph inside compilation. - False: cudagraph inside compilation is not used. - True: cudagraph inside compilation is used. It requires that all input buffers have fixed addresses, and all splitting ops write their outputs to input buffers. Note that this is orthogonal to the cudagraph capture logic outside of compilation. TODO: move outside cudagraph logic into compilation. torch.compile will handle cudagraph capture logic in the future. - cudagraph_capture_sizes: sizes to capture cudagraph. - None (default): capture sizes are inferred from vllm config. - list[int]: capture sizes are specified as given. - cudagraph_num_of_warmups: number of warmup runs for cudagraph. It means the first several runs will be treated as warmup runs. Only after that, the execution will be recorded, and the recorded cudagraph will be used for subsequent runs. - cudagraph_copy_inputs: whether to copy input tensors for cudagraph. If the caller can guarantee that the same input buffers are always used, it can set this to False. Otherwise, it should set this to True, and the compiler will copy the input to an internally managed buffer. Default is False. - Inductor compilation: - use_inductor: whether to use inductor compilation. - False: inductor compilation is not used. graph runs in eager. - True: inductor compilation is used. one graph for symbolic shape is compiled. In addition, compile for compile_sizes, using configurations in inductor_compile_config. - compile_sizes: sizes to compile for inductor. In addition to integers, it also supports "cudagraph_capture_sizes" to specify the sizes for cudagraph capture. - inductor_compile_config: additional configurations for inductor. - None: use default configurations. - inductor_passes: additional passes for inductor. It is a dictionary from pass name to pass function qualified name. We use function name because the config uses json format. If we pass the config from Python, functions can also be passed directly via Python object constructor, e.g. `CompilationConfig(inductor_passes={"a": func})` - custom inductor passes: see PassConfig for more details Why we have different sizes for cudagraph and inductor: - cudagraph: a cudagraph captured for a specific size can only be used for the same size. We need to capture all the sizes we want to use. - inductor: a graph compiled by inductor for a general shape can be used for different sizes. Inductor can also compile for specific sizes, where it can have more information to optimize the graph with fully static shapes. However, we find the general shape compilation is sufficient for most cases. It might be beneficial to compile for certain small batchsizes, where inductor is good at optimizing. """ # noqa level: int = 0 debug_dump_path: str = "" cache_dir: str = "" backend: str = "" custom_ops: list[str] = Field(default_factory=list) splitting_ops: list[str] = Field(default=None) # type: ignore use_inductor: bool = True compile_sizes: Optional[list[Union[int, str]]] = Field(default=None) inductor_compile_config: dict = Field(default_factory=dict) inductor_passes: dict[str, str] = Field(default_factory=dict) use_cudagraph: bool = False cudagraph_num_of_warmups: int = 0 cudagraph_capture_sizes: Optional[list[int]] = None cudagraph_copy_inputs: bool = False class PassConfig(BaseModel): """ Configuration for custom Inductor passes. This is separate from general CompilationConfig so that inductor passes don't all have access to full configuration - that would create a cycle as the PassManager is set as a property of config. - dump_graph_stages: list of stages for which we want to dump the graph. Each pass defines its own stages (before, after, maybe in-between). - dump_graph_dir: directory to dump the graphs. Default is . - enable_fusion: whether to enable the custom fusion pass. - enable_noop: whether to enable the custom no-op elimination pass. TODO(luka) better pass enabling system. """ dump_graph_stages: list[str] = Field(default_factory=list) dump_graph_dir: Path = Field(default=Path(".")) enable_fusion: bool = True enable_noop: bool = True def uuid(self): """ Produces a hash unique to the pass configuration. Any new fields that affect compilation should be added to the hash. Do not include dump_graph_* in the hash - they don't affect compilation. """ dict_ = self.model_dump(include={"enable_fusion", "enable_noop"}) return InductorPass.hash_dict(dict_) def model_post_init(self, __context: Any) -> None: if not self.enable_noop and self.enable_fusion: logger.warning_once( "Fusion enabled but reshape elimination disabled. " "RMSNorm + quant (fp8) fusion might not work") pass_config: PassConfig = Field(default_factory=PassConfig) # not configurable, computed after init max_capture_size: int = PrivateAttr local_cache_dir: str = PrivateAttr # local cache dir for each rank # optimization: # Intuitively, bs_to_padded_graph_size should be dict[int, int]. # since we know all keys are in a range [0, max_capture_size], # we can optimize it to list[int] for better lookup performance. bs_to_padded_graph_size: list[int] = PrivateAttr # keep track of enabled and disabled custom ops enabled_custom_ops: Counter[str] = PrivateAttr disabled_custom_ops: Counter[str] = PrivateAttr traced_files: set[str] = PrivateAttr compilation_time: float = PrivateAttr # Per-model forward context # Map from layer name to the attention cls static_forward_context: dict[str, Any] = PrivateAttr def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ factors: list[Any] = [] factors.append(self.level) factors.append(self.backend) factors.append(self.custom_ops) factors.append(self.splitting_ops) factors.append(self.use_inductor) factors.append(self.inductor_compile_config) factors.append(self.inductor_passes) factors.append(self.pass_config.uuid()) return hashlib.sha256(str(factors).encode()).hexdigest() def __repr__(self) -> str: exclude = { "static_forward_context", "enabled_custom_ops", "disabled_custom_ops", "compilation_time", "bs_to_padded_graph_size", "pass_config", "traced_files", } return self.model_dump_json(exclude=exclude, exclude_unset=True) __str__ = __repr__ @classmethod def from_cli(cls, cli_value: str) -> "CompilationConfig": """Parse the CLI value for the compilation config.""" if cli_value in ["0", "1", "2", "3"]: return cls(level=int(cli_value)) # do not use `eval`, it is dangerous and can execute arbitrary code dict_value = ast.literal_eval(cli_value) return CompilationConfig.model_validate(dict_value) def model_post_init(self, __context: Any) -> None: count_none = self.custom_ops.count("none") count_all = self.custom_ops.count("all") assert count_none + count_all <= 1, "Can only specify 'none' or 'all'" # TODO(zou3519/luka): There are 2 issues with auto-functionalization V2: # 1. A bug in PyTorch, fixed in 2.7: # https://github.com/pytorch/pytorch/issues/147924 # 2. Custom passes (fusion) rely on auto-functionalization V1 and don't # work with V2. Addressing this will take extra engineering effort # and it is not yet a priority. RFC here: # https://github.com/vllm-project/vllm/issues/14703 if Version(importlib.metadata.version('torch')) >= Version("2.6"): KEY = 'enable_auto_functionalized_v2' if KEY not in self.inductor_compile_config: self.inductor_compile_config[KEY] = False if self.splitting_ops is None: self.splitting_ops = [] for k, v in self.inductor_passes.items(): if not isinstance(v, str): assert callable(v), ( f"pass {k} should be callable or a qualified name") self.inductor_compile_config[k] = v if isinstance( v, InductorPass) else CallableInductorPass(v) continue # resolve function from qualified name names = v.split(".") module = ".".join(names[:-1]) func_name = names[-1] func = __import__(module).__dict__[func_name] self.inductor_compile_config[k] = func if isinstance( func, InductorPass) else CallableInductorPass(func) self.enabled_custom_ops = Counter() self.disabled_custom_ops = Counter() self.traced_files = set() self.static_forward_context = {} self.compilation_time = 0.0 def init_backend(self, vllm_config: "VllmConfig") -> Union[str, Callable]: if self.level == CompilationLevel.NO_COMPILATION: raise ValueError("No compilation level is set.") from torch._dynamo.backends.registry import list_backends torch_backends = list_backends(exclude_tags=tuple()) if self.level in [ CompilationLevel.DYNAMO_AS_IS, CompilationLevel.DYNAMO_ONCE ]: if self.backend == "": return "eager" if self.backend in torch_backends: return self.backend return resolve_obj_by_qualname(self.backend) # TODO: pass user-specified backend to piecewise compilation # merge with the config use_inductor assert self.level == CompilationLevel.PIECEWISE from vllm.compilation.backends import VllmBackend return VllmBackend(vllm_config) def init_with_cudagraph_sizes(self, cudagraph_capture_sizes: list[int]) -> None: """To complete the initialization of config, we need to know the cudagraph sizes.""" if self.cudagraph_capture_sizes is None: self.cudagraph_capture_sizes = cudagraph_capture_sizes else: # de-duplicate the sizes provided by the config self.cudagraph_capture_sizes = list( set(self.cudagraph_capture_sizes)) logger.info(("cudagraph sizes specified by model runner" " %s is overridden by config %s"), cudagraph_capture_sizes, self.cudagraph_capture_sizes) computed_compile_sizes = [] if self.compile_sizes is not None: # de-duplicate the sizes provided by the config self.compile_sizes = list(set(self.compile_sizes)) for x in self.compile_sizes: if isinstance(x, str): assert x == "cudagraph_capture_sizes", \ "Unrecognized size type in compile_sizes, " \ f"expect 'cudagraph_capture_sizes', got {x}" computed_compile_sizes.extend(self.cudagraph_capture_sizes) else: assert isinstance(x, int) computed_compile_sizes.append(x) self.compile_sizes = computed_compile_sizes # type: ignore # sort to make sure cudagraph capture sizes are in descending order self.cudagraph_capture_sizes.sort(reverse=True) self.max_capture_size = self.cudagraph_capture_sizes[ 0] if self.cudagraph_capture_sizes else 0 # pre-compute the mapping from batch size to padded graph size self.bs_to_padded_graph_size = [ 0 for i in range(self.max_capture_size + 1) ] for end, start in zip(self.cudagraph_capture_sizes, self.cudagraph_capture_sizes[1:] + [0]): for bs in range(start, end): if bs == start: self.bs_to_padded_graph_size[bs] = start else: self.bs_to_padded_graph_size[bs] = end self.bs_to_padded_graph_size[ self.max_capture_size] = self.max_capture_size def set_splitting_ops_for_v1(self): # If default, override splitting ops for piecewise cudagraph on V1. # NOTE: this function needs to be called if not self.splitting_ops: self.splitting_ops = [ "vllm.unified_attention", "vllm.unified_attention_with_output", ] @dataclass class VllmConfig: """Dataclass which contains all vllm-related configuration. This simplifies passing around the distinct configurations in the codebase. """ model_config: ModelConfig = field(default=None, init=True) # type: ignore cache_config: CacheConfig = field(default=None, init=True) # type: ignore parallel_config: ParallelConfig = field(default_factory=ParallelConfig, init=True) scheduler_config: SchedulerConfig = field(default_factory=SchedulerConfig, init=True) device_config: DeviceConfig = field(default=None, init=True) # type: ignore load_config: LoadConfig = field(default=None, init=True) # type: ignore lora_config: Optional[LoRAConfig] = None speculative_config: SpeculativeConfig = field(default=None, init=True) # type: ignore decoding_config: Optional[DecodingConfig] = None observability_config: Optional[ObservabilityConfig] = None prompt_adapter_config: Optional[PromptAdapterConfig] = None quant_config: Optional[QuantizationConfig] = None compilation_config: CompilationConfig = field(default=None, init=True) # type: ignore kv_transfer_config: KVTransferConfig = field(default=None, init=True) # type: ignore # some opaque config, only used to provide additional information # for the hash computation, mainly used for testing, debugging or out of # tree config registration. additional_config: SupportsHash = field(default=None, init=True) # type: ignore instance_id: str = "" def compute_hash(self) -> str: """ WARNING: Whenever a new field is added to this config, ensure that it is included in the factors list if it affects the computation graph. Provide a hash that uniquely identifies all the configs that affect the structure of the computation graph from input ids/embeddings to the final hidden states, excluding anything before input ids/embeddings and after the final hidden states. """ factors: list[Any] = [] # summarize vllm config vllm_factors: list[Any] = [] from vllm import __version__ vllm_factors.append(__version__) vllm_factors.append(envs.VLLM_USE_V1) if self.model_config: vllm_factors.append(self.model_config.compute_hash()) else: vllm_factors.append("None") if self.cache_config: vllm_factors.append(self.cache_config.compute_hash()) else: vllm_factors.append("None") if self.parallel_config: vllm_factors.append(self.parallel_config.compute_hash()) else: vllm_factors.append("None") if self.scheduler_config: vllm_factors.append(self.scheduler_config.compute_hash()) else: vllm_factors.append("None") if self.device_config: vllm_factors.append(self.device_config.compute_hash()) else: vllm_factors.append("None") if self.load_config: vllm_factors.append(self.load_config.compute_hash()) else: vllm_factors.append("None") if self.lora_config: vllm_factors.append(self.lora_config.compute_hash()) # LoRA creates static buffers based on max_num_batched_tokens. # The tensor sizes and strides get captured in the torch.compile # graph explicitly. vllm_factors.append( str(self.scheduler_config.max_num_batched_tokens)) else: vllm_factors.append("None") if self.speculative_config: vllm_factors.append(self.speculative_config.compute_hash()) else: vllm_factors.append("None") if self.decoding_config: vllm_factors.append(self.decoding_config.compute_hash()) else: vllm_factors.append("None") if self.observability_config: vllm_factors.append(self.observability_config.compute_hash()) else: vllm_factors.append("None") if self.prompt_adapter_config: vllm_factors.append(self.prompt_adapter_config.compute_hash()) else: vllm_factors.append("None") if self.quant_config: pass # should be captured by model_config.quantization if self.compilation_config: vllm_factors.append(self.compilation_config.compute_hash()) else: vllm_factors.append("None") if self.kv_transfer_config: vllm_factors.append(self.kv_transfer_config.compute_hash()) else: vllm_factors.append("None") if self.additional_config: vllm_factors.append(self.additional_config.compute_hash()) else: vllm_factors.append("None") factors.append(vllm_factors) hash_str = hashlib.md5(str(factors).encode(), usedforsecurity=False).hexdigest()[:10] return hash_str def pad_for_cudagraph(self, batch_size: int) -> int: # if batch_size > self.compilation_config.max_capture_size, # it should raise an IndexError. # the caller should make sure the batch_size is within the range, # i.e., batch_size <= self.compilation_config.max_capture_size return self.compilation_config.bs_to_padded_graph_size[batch_size] @staticmethod def _get_quantization_config( model_config: ModelConfig, load_config: LoadConfig) -> Optional[QuantizationConfig]: """Get the quantization config.""" from vllm.platforms import current_platform if model_config.quantization is not None: from vllm.model_executor.model_loader.weight_utils import ( get_quant_config) quant_config = get_quant_config(model_config, load_config) capability_tuple = current_platform.get_device_capability() if capability_tuple is not None: capability = capability_tuple.to_int() if capability < quant_config.get_min_capability(): raise ValueError( f"The quantization method {model_config.quantization} " "is not supported for the current GPU. Minimum " f"capability: {quant_config.get_min_capability()}. " f"Current capability: {capability}.") supported_dtypes = quant_config.get_supported_act_dtypes() if model_config.dtype not in supported_dtypes: raise ValueError( f"{model_config.dtype} is not supported for quantization " f"method {model_config.quantization}. Supported dtypes: " f"{supported_dtypes}") return quant_config return None def with_hf_config( self, hf_config: PretrainedConfig, architectures: Optional[list[str]] = None, ) -> "VllmConfig": if architectures is not None: hf_config = copy.deepcopy(hf_config) hf_config.architectures = architectures model_config = copy.deepcopy(self.model_config) model_config.hf_config = hf_config return replace(self, model_config=model_config) def __post_init__(self): """Verify configs are valid & consistent with each other. """ if self.model_config is not None: self.model_config.verify_async_output_proc(self.parallel_config, self.speculative_config, self.device_config) self.model_config.verify_with_parallel_config(self.parallel_config) if self.cache_config is not None: self.cache_config.verify_with_parallel_config(self.parallel_config) if self.lora_config: self.lora_config.verify_with_cache_config(self.cache_config) self.lora_config.verify_with_model_config(self.model_config) self.lora_config.verify_with_scheduler_config( self.scheduler_config) if self.prompt_adapter_config: self.prompt_adapter_config.verify_with_model_config( self.model_config) if self.quant_config is None and \ self.model_config is not None and self.load_config is not None: self.quant_config = VllmConfig._get_quantization_config( self.model_config, self.load_config) from vllm.platforms import current_platform if self.scheduler_config is not None and \ self.model_config is not None and \ self.scheduler_config.chunked_prefill_enabled and \ self.model_config.dtype == torch.float32 and \ current_platform.get_device_capability() == (7, 5): logger.warning_once( "Turing devices tensor cores do not support float32 matmul. " "To workaround this limitation, vLLM will set 'ieee' input " "precision for chunked prefill triton kernels.") if self.compilation_config is None: self.compilation_config = CompilationConfig() if envs.VLLM_USE_V1 and self.model_config is not None and \ not self.model_config.enforce_eager: # NOTE(woosuk): Currently, we use inductor because the piecewise # CUDA graphs do not work properly with the custom CUDA kernels. # FIXME(woosuk): Disable inductor to reduce the compilation time # and avoid any potential issues with the inductor. # FIXME(rob): Add function to set all of these. self.compilation_config.custom_ops = ["none"] self.compilation_config.use_cudagraph = True self.compilation_config.use_inductor = True self.compilation_config.cudagraph_num_of_warmups = 1 self.compilation_config.pass_config.enable_fusion = False self.compilation_config.pass_config.enable_noop = False self.compilation_config.level = CompilationLevel.PIECEWISE self.compilation_config.set_splitting_ops_for_v1() self._set_cudagraph_sizes() if self.cache_config is not None and \ self.cache_config.cpu_offload_gb > 0 and \ self.compilation_config.level != CompilationLevel.NO_COMPILATION \ and not envs.VLLM_USE_V1: logger.warning( "CPU offload is not supported with `torch.compile` in v0 yet." " Disabling `torch.compile`.") self.compilation_config.level = CompilationLevel.NO_COMPILATION if ((not envs.VLLM_USE_V1) and self.lora_config is not None and self.compilation_config.level != CompilationLevel.NO_COMPILATION): logger.warning( "LoRA for V0 is not supported with `torch.compile` yet. " "Disabling `torch.compile`.") self.compilation_config.level = CompilationLevel.NO_COMPILATION if self.model_config and self.model_config.use_mla and \ not (current_platform.is_cuda() or current_platform.is_rocm()): logger.info( "MLA is enabled on a non-GPU platform; forcing chunked " "prefill and prefix caching to be disabled.") self.scheduler_config.enable_chunked_prefill = False self.scheduler_config.chunked_prefill_enabled = False self.scheduler_config.max_num_batched_tokens = max( self.scheduler_config.max_model_len, _DEFAULT_MAX_NUM_BATCHED_TOKENS) if self.cache_config is not None: self.cache_config.enable_prefix_caching = False current_platform.check_and_update_config(self) if not self.instance_id: self.instance_id = random_uuid()[:5] def _set_cudagraph_sizes(self): """ cudagraph batchsize padding logic: `[1, 2, 4] + [8 * i for i in range(1, 1025)]` is a list of all possible batch sizes that cudagraph will capture. Depending on the engine's configuration of `max_num_seqs`, the candidate batch sizes to capture cudagraph will shrink to the subset which just cover the range of `[1, max_num_seqs]`. In the common case, `max_num_seqs` is 256, and the cudagraph batch sizes will be `[1, 2, 4, 8, 16, 24, 32, 40, ..., 256]`. However, if users specify the cudagraph capture sizes through compilation config, we will use the specified sizes instead. In the end, `vllm_config.compilation_config.cudagraph_capture_sizes` will be the final sizes to capture cudagraph (in descending order). During runtime, if batchsize is larger than `vllm_config.compilation_config.cudagraph_capture_sizes`, no cudagraph will be used. If the batch size is no larger than `vllm_config.compilation_config.cudagraph_capture_sizes`, we can quickly find the padded graph size for a given batch size by looking up `vllm_config.compilation_config.bs_to_padded_graph_size`. """ # calculate the default `batch_size_capture_list` if not envs.VLLM_USE_V1: batch_size_capture_list = [] max_batchsize_to_capture = 0 if self.scheduler_config is not None and \ self.model_config is not None and \ not self.model_config.enforce_eager: possible_sizes = [1, 2, 4] + [8 * i for i in range(1, 1025)] # find the minimum size that is larger than max_num_seqs, # which then becomes the max_batchsize_to_capture larger_sizes = [ x for x in possible_sizes if x >= self.scheduler_config.max_num_seqs ] if larger_sizes: max_batchsize_to_capture = larger_sizes[0] else: max_batchsize_to_capture = possible_sizes[-1] # filter out the sizes that are # larger than max_batchsize_to_capture batch_size_capture_list = [ size for size in possible_sizes if size <= max_batchsize_to_capture ] else: batch_size_capture_list = [] if self.model_config is not None and \ not self.model_config.enforce_eager: batch_size_capture_list = [1, 2, 4 ] + [i for i in range(8, 513, 8)] max_num_tokens = self.scheduler_config.max_num_batched_tokens batch_size_capture_list = [ size for size in batch_size_capture_list if size <= max_num_tokens ] self.compilation_config.init_with_cudagraph_sizes( batch_size_capture_list) def __str__(self): return ( f"model={self.model_config.model!r}," f" speculative_config={self.speculative_config!r}," f" tokenizer={self.model_config.tokenizer!r}, " f"skip_tokenizer_init={self.model_config.skip_tokenizer_init}," f" tokenizer_mode={self.model_config.tokenizer_mode}, " f"revision={self.model_config.revision}, " f"override_neuron_config={self.model_config.override_neuron_config}," f" tokenizer_revision={self.model_config.tokenizer_revision}, " f"trust_remote_code={self.model_config.trust_remote_code}, " f"dtype={self.model_config.dtype}, " f"max_seq_len={self.model_config.max_model_len}," f" download_dir={self.load_config.download_dir!r}, " f"load_format={self.load_config.load_format}, " f"tensor_parallel_size={self.parallel_config.tensor_parallel_size}," f" pipeline_parallel_size={self.parallel_config.pipeline_parallel_size}, " # noqa f"disable_custom_all_reduce={self.parallel_config.disable_custom_all_reduce}, " # noqa f"quantization={self.model_config.quantization}, " f"enforce_eager={self.model_config.enforce_eager}, " f"kv_cache_dtype={self.cache_config.cache_dtype}, " f" device_config={self.device_config.device}, " f"decoding_config={self.decoding_config!r}, " f"observability_config={self.observability_config!r}, " f"seed={self.model_config.seed}, " f"served_model_name={self.model_config.served_model_name}, " f"num_scheduler_steps={self.scheduler_config.num_scheduler_steps}, " f"multi_step_stream_outputs={self.scheduler_config.multi_step_stream_outputs}, " # noqa f"enable_prefix_caching={self.cache_config.enable_prefix_caching}, " f"chunked_prefill_enabled={self.scheduler_config.chunked_prefill_enabled}, " # noqa f"use_async_output_proc={self.model_config.use_async_output_proc}, " f"disable_mm_preprocessor_cache={self.model_config.disable_mm_preprocessor_cache!r}, " # noqa f"mm_processor_kwargs={self.model_config.mm_processor_kwargs}, " f"pooler_config={self.model_config.pooler_config!r}, " f"compilation_config={self.compilation_config!r}") _current_vllm_config: Optional[VllmConfig] = None @contextmanager def set_current_vllm_config(vllm_config: VllmConfig, check_compile=False): """ Temporarily set the current vLLM config. Used during model initialization. We save the current vLLM config in a global variable, so that all modules can access it, e.g. custom ops can access the vLLM config to determine how to dispatch. """ global _current_vllm_config old_vllm_config = _current_vllm_config from vllm.compilation.counter import compilation_counter num_models_seen = compilation_counter.num_models_seen try: _current_vllm_config = vllm_config yield finally: logger.debug("enabled custom ops: %s", vllm_config.compilation_config.enabled_custom_ops) logger.debug("disabled custom ops: %s", vllm_config.compilation_config.disabled_custom_ops) if check_compile and \ vllm_config.compilation_config.level == CompilationLevel.PIECEWISE \ and compilation_counter.num_models_seen == num_models_seen: # If the model supports compilation, # compilation_counter.num_models_seen should be increased # by at least 1. # If it is not increased, it means the model does not support # compilation (does not have @support_torch_compile decorator). logger.warning( "`torch.compile` is turned on, but the model %s" " does not support it. Please open an issue on GitHub" " if you want it to be supported.", vllm_config.model_config.model) _current_vllm_config = old_vllm_config def get_current_vllm_config() -> VllmConfig: if _current_vllm_config is None: # in ci, usually when we test custom ops/modules directly, # we don't set the vllm config. In that case, we set a default # config. logger.warning("Current vLLM config is not set.") from vllm.config import VllmConfig return VllmConfig() return _current_vllm_config