aphrodite-engine/aphrodite/common/sampling_params.py

"""Sampling parameters for text generation."""
import copy
import os
from enum import IntEnum
from functools import cached_property
from typing import Any, Callable, Dict, List, Optional, Set, Union

import msgspec
import torch
from loguru import logger
from typing_extensions import Annotated

_SAMPLING_EPS = 1e-5
_MAX_TEMP = 1e-2

APHRODITE_NO_DEPRECATION_WARNING = bool(
    int(os.environ.get("APHRODITE_NO_DEPRECATION_WARNING", "0")))


class SamplingType(IntEnum):
    GREEDY = 0
    RANDOM = 1
    RANDOM_SEED = 2
    BEAM = 3


LogitsProcessorFunc = Union[Callable[[List[int], torch.Tensor], torch.Tensor],
                            Callable[[List[int], List[int], torch.Tensor],
                                     torch.Tensor]]
"""LogitsProcessor is a function that takes a list
of previously generated tokens, the logits tensor
for the next token and, optionally, prompt tokens as a
first argument, and returns a modified tensor of logits
to sample from."""


class SamplingParams(
    msgspec.Struct,
    omit_defaults=True,
    dict=True):
    """Sampling parameters for text generation.

    Overall, we follow the sampling parameters from the OpenAI text completion
    API (https://platform.openai.com/docs/api-reference/completions/create).
    In addition, we support multiple additional samplers which are not supported
    by OpenAI.

    Args:
        n: Number of output sequences to return for the given prompt.
        best_of: Number of output sequences that are generated from the prompt.
            From these `best_of` sequences, the top `n` sequences are returned.
            `best_of` must be greater than or equal to `n`. This is treated as
            the beam width when `use_beam_search` is True. By default, `best_of`
            is set to `n`.
        presence_penalty: Float that penalizes new tokens based on whether they
            appear in the generated text so far. Values > 0 encourage the model
            to use new tokens, while values < 0 encourage the model to repeat
            tokens.
        frequency_penalty: Float that penalizes new tokens based on their
            frequency in the generated text so far. Values > 0 encourage the
            model to use new tokens, while values < 0 encourage the model to
            repeat tokens.
        repetition_penalty: Float that penalizes new tokens based on their
            frequency in the generated text so far.
            freq_pen is applied additively while
            rep_pen is applied multiplicatively.
            Must be in [1, inf). Set to 1 to disable the effect.
        temperature: Float that controls the randomness of the sampling. Lower
            values make the model more deterministic, while higher values make
            the model more random. Zero means greedy sampling.
        top_p: Float that controls the cumulative probability of the top tokens
            to consider. Must be in (0, 1]. Set to 1 to consider all tokens.
        top_k: Integer that controls the number of top tokens to consider. Set
            to -1 to consider all tokens.
        top_a: Float that controls the cutoff for Top-A sampling.
            Exact cutoff is top_a*max_prob**2. Must be in [0,inf], 0 to disable.
        min_p: Float that controls the cutoff for min-p sampling.
            Exact cutoff is min_p*max_prob. Must be in [0,1], 0 to disable.
        tfs: Float that controls the cumulative approximate curvature of the
            distribution to retain for Tail Free Sampling.
            Must be in (0, 1]. Set to 1 to disable
        eta_cutoff: Float that controls the cutoff threshold for Eta sampling
            (a form of entropy adaptive truncation sampling)
            threshold is computed as min(eta, sqrt(eta)*entropy(probs)).
            Specified in units of 1e-4. Set to 0 to disable
        epsilon_cutoff: Float that controls the cutoff threshold for
            Epsilon sampling (simple probability threshold truncation).
            Specified in units of 1e-4. Set to 0 to disable.
        typical_p: Float that controls the cumulative probability of tokens
            closest in surprise to the expected surprise to consider.
            Must be in (0, 1]. Set to 1 to disable.
        mirostat_mode: Can either be 0 (disabled) or 2 (Mirostat v2).
        mirostat_tau: Target "surprisal" that mirostat works towards.
            Range [0, inf).
        mirostat_eta: Rate at which mirostat updates its internal surprisal
            value. Range [0, inf).
        dynatemp_min: Minimum temperature for dynatemp sampling.
            Range [0, inf).
        dynatemp_max: Maximum temperature for dynatemp sampling.
            Range [0, inf).
        dynatemp_exponent: Exponent for dynatemp sampling. Range [0, inf).
        smoothing_factor: Smoothing factor for Quadratic Sampling.
        smoothing_curve: Smoothing curve for Quadratic (Cubic) Sampling.
        seed: Random seed to use for the generation.
        use_beam_search: Whether to use beam search instead of sampling.
        length_penalty: Float that penalizes sequences based on their length.
            Used in beam search.
        early_stopping: Controls the stopping condition for beam search. It
            accepts the following values: `True`, where the generation stops as
            soon as there are `best_of` complete candidates; `False`, where an
            heuristic is applied and the generation stops when is it very
            unlikely to find better candidates; `"never"`, where the beam search
            procedure only stops when there cannot be better candidates
            (canonical beam search algorithm).
        stop: List of strings that stop the generation when they are generated.
            The returned output will not contain the stop strings.
        stop_token_ids: List of tokens that stop the generation when they are
            generated. The returned output will contain the stop tokens unless
            the stop tokens are special tokens.
        include_stop_str_in_output: Whether to include the stop strings in
            output text. Defaults to False.
        ignore_eos: Whether to ignore the EOS token and continue generating
            tokens after the EOS token is generated.
        max_tokens: Maximum number of tokens to generate per output sequence.
        min_tokens: Minimum number of tokens to generate per output sequence
            before EOS or stop tokens are generated.
        logprobs: Number of log probabilities to return per output token.
            When set to None, no probability is returned. If set to a non-None
            value, the result includes the log probabilities of the specified
            number of most likely tokens, as well as the chosen tokens.
            Note that the implementation follows the OpenAI API: The API will
            always return the log probability of the sampled token, so there
            may be up to `logprobs+1` elements in the response.
        prompt_logprobs: Number of log probabilities to return per prompt token.
        detokenize: Whether to detokenize the output. Defaults to True.
        custom_token_bans: List of token IDs to ban from generating
        skip_special_tokens: Whether to skip special tokens in the output.
            defaults to true.
        spaces_between_special_tokens: Whether to add spaces between special
            tokens in the output. Defaults to True.
        logits_processors: List of functions that modify logits based on
            previously generated tokens, and optionally prompt tokens as
            a first argument.
        truncate_prompt_tokens: If set to an integer k, will use only the last
            k tokens from the prompt (i.e. left-truncation). Defaults to None
            (i.e. no truncation).
        xtc_threshold: In XTC sampling, if 2 or more tokens have probability
            above this threshold, consider removing all but the last one.
        xtc_probability: Probability that the removal will actually happen.
            0 disables the sampler, 1 makes it always happen.
        nsigma: Number of standard deviations from the maximum logit to use
            as a cutoff threshold. Tokens with logits below
            (max_logit - nsgima * std_dev) are filtered out. Higher values
            (e.g. 3.0) keep more tokens, lower values (e.g. 1.0) are more
            selective. Must be positive. 0 to disable.
        dry_multiplier: Float that controls the magnitude of the DRY sampling
            penalty. Higher values create stronger penalties against
            repetition. The penalty is multiplied by this value before being
            applied. Must be non-negative. 0 disables the sampler.
        dry_base: Base for the exponential growth of the DRY sampling penalty.
            Controls how quickly the penalty increases with longer repeated
            sequences. Must be greater than 1. Higher values (e.g. 2.0) create
            more aggressive penalties for longer repetitions. Defaults to 1.75.
        dry_allowed_length: Maximum number of tokens that can be repeated
            without incurring a DRY sampling penalty. Sequences longer than
            this will be penalized exponentially. Must be at least 1.
            Defaults to 2.
        dry_sequence_breaker_ids: List of token IDs that stop
            the matching of repeated content. These tokens will break up the
            input into sections where repetition is evaluated separately.
            Common examples are newlines, quotes, and other structural tokens.
            Defaults to None.
    """

    n: int = 1
    best_of: Optional[int] = None
    presence_penalty: float = 0.0
    frequency_penalty: float = 0.0
    repetition_penalty: float = 1.0
    temperature: float = 1.0
    dynatemp_min: float = 0.0
    dynatemp_max: float = 0.0
    dynatemp_exponent: float = 1.0
    temperature_last: bool = False
    top_p: float = 1.0
    top_k: int = -1
    top_a: float = 0.0
    min_p: float = 0.0
    tfs: float = 1.0
    eta_cutoff: float = 0.0
    epsilon_cutoff: float = 0.0
    typical_p: float = 1.0
    smoothing_factor: float = 0.0
    smoothing_curve: float = 1.0
    seed: Optional[int] = None
    use_beam_search: bool = False
    length_penalty: float = 1.0
    early_stopping: Union[bool, str] = False
    stop: Union[None, str, List[str]] = None
    stop_token_ids: Optional[List[int]] = None
    include_stop_str_in_output: bool = False
    ignore_eos: bool = False
    max_tokens: Optional[int] = 16
    min_tokens: int = 0
    logprobs: Optional[int] = None
    prompt_logprobs: Optional[int] = None
    detokenize: bool = True
    custom_token_bans: Optional[List[int]] = None
    skip_special_tokens: bool = True
    spaces_between_special_tokens: bool = True
    # Optional[List[LogitsProcessorFunc]] type.
    # We use Any here because the type above
    # is not supported by msgspec.
    logits_processors: Optional[Any] = None
    truncate_prompt_tokens: Optional[Annotated[int, msgspec.Meta(ge=1)]] = None
    xtc_threshold: float = 0.1
    xtc_probability: float = 0
    nsigma: float = 0.0
    dry_multiplier: float = 0.0
    dry_base: float = 1.75
    dry_allowed_length: int = 2
    dry_sequence_breaker_ids: List[int] = []
    # The below fields are not supposed to be used as an input.
    # They are set in post_init.
    output_text_buffer_length: int = 0
    _all_stop_token_ids: Set[int] = msgspec.field(default_factory=set)

    default_values = {
        "n": 1,
        "best_of": 1,
        "presence_penalty": 0.0,
        "frequency_penalty": 0.0,
        "repetition_penalty": 1.0,
        "temperature": 1.0,
        "dynatemp_min": 0.0,
        "dynatemp_max": 0.0,
        "dynatemp_exponent": 1.0,
        "temperature_last": False,
        "top_p": 1.0,
        "top_k": -1,
        "top_a": 0.0,
        "min_p": 0.0,
        "tfs": 1.0,
        "eta_cutoff": 0.0,
        "epsilon_cutoff": 0.0,
        "typical_p": 1.0,
        "smoothing_factor": 0.0,
        "smoothing_curve": 1.0,
        "seed": None,
        "use_beam_search": False,
        "length_penalty": 1.0,
        "early_stopping": False,
        "stop": [],
        "stop_token_ids": [],
        "ignore_eos": False,
        "max_tokens": 16,
        "min_tokens": 0,
        "logprobs": None,
        "prompt_logprobs": None,
        "detokenize": True,
        "custom_token_bans": [],
        "skip_special_tokens": True,
        "spaces_between_special_tokens": True,
        "include_stop_str_in_output": False,
        "truncate_prompt_tokens": None,
        "xtc_threshold": 0.1,
        "xtc_probability": 0,
        "nsigma": 0.0,
        "dry_multiplier": 0.0,
        "dry_base": 1.75,
        "dry_allowed_length": 2,
        "dry_sequence_breaker_ids": [],
    }

    def __post_init__(self) -> None:
        self.best_of = self.best_of or self.n
        if 0 < self.temperature < _MAX_TEMP:
            logger.warning(
                f"temperature {self.temperature} is less than {_MAX_TEMP}, "
                "which may cause numerical errors NaN or inf in tensors. We "
                f"have maxed it out to {_MAX_TEMP}.")
            self.temperature = max(self.temperature, _MAX_TEMP)
        if self.seed == -1:
            self.seed = None
        else:
            self.seed = self.seed
        if self.stop is None:
            self.stop = []
        elif isinstance(self.stop, str):
            self.stop = [self.stop]
        else:
            self.stop = list(self.stop)
        if self.stop_token_ids is None:
            self.stop_token_ids = []
        else:
            self.stop_token_ids = list(self.stop_token_ids)
        self.logprobs = 1 if self.logprobs is True else self.logprobs
        self.prompt_logprobs = (1 if self.prompt_logprobs is True else
                                self.prompt_logprobs)

        # Number of characters to hold back for stop string evaluation
        # until sequence is finished.
        if self.stop and not self.include_stop_str_in_output:
            self.output_text_buffer_length = max(len(s) for s in self.stop) - 1

        self._verify_args()
        if self.use_beam_search:
            if not APHRODITE_NO_DEPRECATION_WARNING:
                logger.warning(
                    "[IMPORTANT] We plan to discontinue the support for beam "
                    "search in the next major release. Set "
                    "APHRODITE_NO_DEPRECATION_WARNING=1 to "
                    "suppress this warning.")
            self._verify_beam_search()
        else:
            self._verify_non_beam_search()
            if self.temperature < _SAMPLING_EPS:
                # Zero temperature means greedy sampling.
                self.top_p = 1.0
                self.top_k = -1
                self.min_p = 0.0
                self.top_a = 0.0
                self._verify_greedy_sampling()
        # eos_token_id is added to this by the engine
        self._all_stop_token_ids = set(self.stop_token_ids)

    def _verify_args(self) -> None:
        if self.n < 1:
            raise ValueError(f"n must be at least 1, got {self.n}.")
        assert isinstance(self.best_of, int)
        if self.best_of < self.n:
            raise ValueError(f"best_of must be greater than or equal to n, "
                             f"got n={self.n} and best_of={self.best_of}.")
        if not -2.0 <= self.presence_penalty <= 2.0:
            raise ValueError("presence_penalty must be in [-2, 2], got "
                             f"{self.presence_penalty}.")
        if not -2.0 <= self.frequency_penalty <= 2.0:
            raise ValueError("frequency_penalty must be in [-2, 2], got "
                             f"{self.frequency_penalty}.")
        if self.repetition_penalty < 1.0:
            raise ValueError("repetition_penalty must be in [1, inf), got "
                             f"{self.repetition_penalty}.")
        if self.temperature < 0.0:
            raise ValueError(
                f"temperature must be non-negative, got {self.temperature}.")
        if not 0.0 < self.top_p <= 1.0:
            raise ValueError(f"top_p must be in (0, 1], got {self.top_p}.")
        if self.top_k < -1 or self.top_k == 0:
            raise ValueError(f"top_k must be -1 (disable), or at least 1, "
                             f"got {self.top_k}.")
        if self.top_a < 0:
            raise ValueError(f"top_a must be non negative, got {self.top_a}.")
        if not 0.0 <= self.min_p <= 1.0:
            raise ValueError(f"min_p must be in [0, 1], got {self.min_p}.")
        if not 0.0 < self.tfs <= 1.0:
            raise ValueError(f"tfs must be in (0, 1], got {self.tfs}.")
        if self.epsilon_cutoff < 0.0 or self.epsilon_cutoff > 1000.0:
            raise ValueError("epsilon_cutoff must be in [0, 1000], got "
                             f"{self.epsilon_cutoff}.")
        # pylint: disable=unneeded-not
        if not self.eta_cutoff >= 0:
            raise ValueError(
                f"eta_cutoff must be non negative, got {self.eta_cutoff}.")
        if not 0.0 <= self.typical_p <= 1.0:
            raise ValueError(
                f"typical_p must be in (0, 1], got {self.typical_p}.")
        if self.max_tokens is not None and self.max_tokens < 1:
            raise ValueError(
                f"max_tokens must be at least 1, got {self.max_tokens}.")
        if self.min_tokens < 0:
            raise ValueError(f"min_tokens must be greater than or equal to 0, "
                             f"got {self.min_tokens}.")
        if self.max_tokens is not None and self.min_tokens > self.max_tokens:
            raise ValueError(
                f"min_tokens must be less than or equal to "
                f"max_tokens={self.max_tokens}, got {self.min_tokens}.")
        if self.logprobs is not None and self.logprobs < 0:
            raise ValueError(
                f"logprobs must be non-negative, got {self.logprobs}.")
        if self.prompt_logprobs is not None and self.prompt_logprobs < 0:
            raise ValueError("prompt_logprobs must be non-negative, got "
                             f"{self.prompt_logprobs}.")
        if (self.truncate_prompt_tokens is not None
                and self.truncate_prompt_tokens < 1):
            raise ValueError(f"truncate_prompt_tokens must be >= 1, "
                             f"got {self.truncate_prompt_tokens}")
        assert isinstance(self.stop, list)
        if any(not stop_str for stop_str in self.stop):
            raise ValueError("stop cannot contain an empty string.")
        if self.stop and not self.detokenize:
            raise ValueError(
                "stop strings are only supported when detokenize is True. "
                "Set detokenize=True to use stop.")
        if self.xtc_threshold < 0.0:
            raise ValueError(
                "xtc_threshold must be non-negative, got "
                f"{self.xtc_threshold}.")
        if not 0.0 <= self.xtc_probability <= 1.0:
            raise ValueError(
                "xtc_probability must be in [0, 1], got "
                f"{self.xtc_probability}.")
        if self.nsigma < 0.0:
            raise ValueError(
                "nsigma must be non-negative, got "
                f"{self.nsigma}.")
        if self.dry_multiplier < 0.0:
            raise ValueError(
                "dry_multiplier must be non-negative, got "
                f"{self.dry_multiplier}.")
        if self.dry_base <= 1.0:
            raise ValueError(
                "dry_base must be greater than 1, got "
                f"{self.dry_base}.")
        if self.dry_allowed_length < 0:
            raise ValueError(
                "dry_allowed_length must be non-negative, got "
                f"{self.dry_allowed_length}.")    

    def _verify_beam_search(self) -> None:
        if self.best_of == 1:
            raise ValueError("best_of must be greater than 1 when using beam "
                             f"search. Got {self.best_of}.")
        if self.temperature > _SAMPLING_EPS:
            raise ValueError("temperature must be 0 when using beam search.")
        if self.top_p < 1.0 - _SAMPLING_EPS:
            raise ValueError("top_p must be 1 when using beam search.")
        if self.top_k != -1:
            raise ValueError("top_k must be -1 when using beam search.")
        if self.early_stopping not in [True, False, "never"]:
            raise ValueError(
                f"early_stopping must be True, False, or 'never', "
                f"got {self.early_stopping}.")

    def _verify_non_beam_search(self) -> None:
        if self.early_stopping is not False:
            raise ValueError("early_stopping is not effective and must be "
                             "False when not using beam search.")
        if (self.length_penalty < 1.0 - _SAMPLING_EPS
                or self.length_penalty > 1.0 + _SAMPLING_EPS):
            raise ValueError(
                "length_penalty is not effective and must be the "
                "default value of 1.0 when not using beam search.")

    def _verify_greedy_sampling(self) -> None:
        assert isinstance(self.best_of, int)
        if self.best_of > 1:
            raise ValueError("best_of must be 1 when using greedy sampling."
                             f"Got {self.best_of}.")
        if self.top_p < 1.0 - _SAMPLING_EPS:
            raise ValueError("top_p must be 1 when using greedy sampling.")
        if self.top_k != -1:
            raise ValueError("top_k must be -1 when using greedy sampling.")

    def update_from_generation_config(
            self,
            generation_config: Dict[str, Any],
            model_eos_token_id: Optional[int] = None) -> None:
        """Update if there are non-default values from generation_config"""

        if model_eos_token_id is not None:
            # Add the eos token id into the sampling_params to support
            # min_tokens processing.
            self._all_stop_token_ids.add(model_eos_token_id)

        # Update eos_token_id for generation
        if (eos_ids := generation_config.get("eos_token_id")) is not None:
            # it can be either int or list of int
            eos_ids = {eos_ids} if isinstance(eos_ids, int) else set(eos_ids)
            if model_eos_token_id is not None:
                # We don't need to include the primary eos_token_id in
                # stop_token_ids since it's handled separately for stopping
                # purposes.
                eos_ids.discard(model_eos_token_id)
            if eos_ids:
                self._all_stop_token_ids.update(eos_ids)
                if not self.ignore_eos:
                    assert isinstance(self.stop_token_ids, list)
                    eos_ids.update(self.stop_token_ids)
                    self.stop_token_ids = list(eos_ids)

    @cached_property
    def sampling_type(self) -> SamplingType:
        if self.use_beam_search:
            return SamplingType.BEAM
        if self.temperature < _SAMPLING_EPS:
            return SamplingType.GREEDY
        if self.seed is not None:
            return SamplingType.RANDOM_SEED
        return SamplingType.RANDOM

    @property
    def all_stop_token_ids(self) -> Set[int]:
        return self._all_stop_token_ids

    def clone(self) -> "SamplingParams":
        """Deep copy excluding LogitsProcessor objects.
        LogitsProcessor objects are excluded because they may contain an
        arbitrary, nontrivial amount of data.
        """

        logit_processor_refs = None if self.logits_processors is None else {
            id(lp): lp
            for lp in self.logits_processors
        }
        return copy.deepcopy(self, memo=logit_processor_refs)

    def __repr__(self) -> str:
        repr_str = "SamplingParams("
        for param, default_value in self.default_values.items():
            current_value = getattr(self, param)
            if current_value != default_value:
                repr_str += f"{param}={current_value}, "
        repr_str = repr_str.rstrip(', ') + ")"
        return repr_str