llama_cpp.py

from __future__ import annotations

import sys
import os
import ctypes
import functools
import pathlib

from typing import (
    Any,
    Callable,
    List,
    Union,
    NewType,
    Optional,
    TYPE_CHECKING,
    TypeVar,
    Generic,
)
from typing_extensions import TypeAlias


# Load the library
def _load_shared_library(lib_base_name: str):
    # Construct the paths to the possible shared library names
    _base_path = pathlib.Path(os.path.abspath(os.path.dirname(__file__)))
    # Searching for the library in the current directory under the name "libllama" (default name
    # for llamacpp) and "llama" (default name for this repo)
    _lib_paths: List[pathlib.Path] = []
    # Determine the file extension based on the platform
    if sys.platform.startswith("linux"):
        _lib_paths += [
            _base_path / f"lib{lib_base_name}.so",
        ]
    elif sys.platform == "darwin":
        _lib_paths += [
            _base_path / f"lib{lib_base_name}.so",
            _base_path / f"lib{lib_base_name}.dylib",
        ]
    elif sys.platform == "win32":
        _lib_paths += [
            _base_path / f"{lib_base_name}.dll",
            _base_path / f"lib{lib_base_name}.dll",
        ]
    else:
        raise RuntimeError("Unsupported platform")

    if "LLAMA_CPP_LIB" in os.environ:
        lib_base_name = os.environ["LLAMA_CPP_LIB"]
        _lib = pathlib.Path(lib_base_name)
        _base_path = _lib.parent.resolve()
        _lib_paths = [_lib.resolve()]

    cdll_args = dict()  # type: ignore
    # Add the library directory to the DLL search path on Windows (if needed)
    if sys.platform == "win32" and sys.version_info >= (3, 8):
        os.add_dll_directory(str(_base_path))
        if "CUDA_PATH" in os.environ:
            os.add_dll_directory(os.path.join(os.environ["CUDA_PATH"], "bin"))
            os.add_dll_directory(os.path.join(os.environ["CUDA_PATH"], "lib"))
        if "HIP_PATH" in os.environ:
            os.add_dll_directory(os.path.join(os.environ["HIP_PATH"], "bin"))
            os.add_dll_directory(os.path.join(os.environ["HIP_PATH"], "lib"))
        cdll_args["winmode"] = ctypes.RTLD_GLOBAL

    # Try to load the shared library, handling potential errors
    for _lib_path in _lib_paths:
        if _lib_path.exists():
            try:
                return ctypes.CDLL(str(_lib_path), **cdll_args)  # type: ignore
            except Exception as e:
                raise RuntimeError(f"Failed to load shared library '{_lib_path}': {e}")

    raise FileNotFoundError(
        f"Shared library with base name '{lib_base_name}' not found"
    )


# Specify the base name of the shared library to load
_lib_base_name = "llama"

# Load the library
_lib = _load_shared_library(_lib_base_name)


# ctypes sane type hint helpers
#
# - Generic Pointer and Array types
# - PointerOrRef type with a type hinted byref function
#
# NOTE: Only use these for static type checking not for runtime checks
# no good will come of that

if TYPE_CHECKING:
    CtypesCData = TypeVar("CtypesCData", bound=ctypes._CData)  # type: ignore

    CtypesArray: TypeAlias = ctypes.Array[CtypesCData]  # type: ignore

    CtypesPointer: TypeAlias = ctypes._Pointer[CtypesCData]  # type: ignore

    CtypesVoidPointer: TypeAlias = ctypes.c_void_p

    class CtypesRef(Generic[CtypesCData]):
        pass

    CtypesPointerOrRef: TypeAlias = Union[
        CtypesPointer[CtypesCData], CtypesRef[CtypesCData]
    ]

    CtypesFuncPointer: TypeAlias = ctypes._FuncPointer  # type: ignore

F = TypeVar("F", bound=Callable[..., Any])


def ctypes_function_for_shared_library(lib: ctypes.CDLL):
    def ctypes_function(
        name: str, argtypes: List[Any], restype: Any, enabled: bool = True
    ):
        def decorator(f: F) -> F:
            if enabled:
                func = getattr(lib, name)
                func.argtypes = argtypes
                func.restype = restype
                functools.wraps(f)(func)
                return func
            else:
                return f

        return decorator

    return ctypes_function


ctypes_function = ctypes_function_for_shared_library(_lib)


def byref(obj: CtypesCData, offset: Optional[int] = None) -> CtypesRef[CtypesCData]:
    """Type-annotated version of ctypes.byref"""
    ...


byref = ctypes.byref  # type: ignore

# from ggml.h
# // NOTE: always add types at the end of the enum to keep backward compatibility
# enum ggml_type {
#     GGML_TYPE_F32     = 0,
#     GGML_TYPE_F16     = 1,
#     GGML_TYPE_Q4_0    = 2,
#     GGML_TYPE_Q4_1    = 3,
#     // GGML_TYPE_Q4_2 = 4, support has been removed
#     // GGML_TYPE_Q4_3 = 5, support has been removed
#     GGML_TYPE_Q5_0    = 6,
#     GGML_TYPE_Q5_1    = 7,
#     GGML_TYPE_Q8_0    = 8,
#     GGML_TYPE_Q8_1    = 9,
#     GGML_TYPE_Q2_K    = 10,
#     GGML_TYPE_Q3_K    = 11,
#     GGML_TYPE_Q4_K    = 12,
#     GGML_TYPE_Q5_K    = 13,
#     GGML_TYPE_Q6_K    = 14,
#     GGML_TYPE_Q8_K    = 15,
#     GGML_TYPE_IQ2_XXS = 16,
#     GGML_TYPE_IQ2_XS  = 17,
#     GGML_TYPE_IQ3_XXS = 18,
#     GGML_TYPE_IQ1_S   = 19,
#     GGML_TYPE_IQ4_NL  = 20,
#     GGML_TYPE_IQ3_S   = 21,
#     GGML_TYPE_IQ2_S   = 22,
#     GGML_TYPE_IQ4_XS  = 23,
#     GGML_TYPE_I8      = 24,
#     GGML_TYPE_I16     = 25,
#     GGML_TYPE_I32     = 26,
#     GGML_TYPE_I64     = 27,
#     GGML_TYPE_F64     = 28,
#     GGML_TYPE_IQ1_M   = 29,
#     GGML_TYPE_COUNT,
# };
GGML_TYPE_F32 = 0
GGML_TYPE_F16 = 1
GGML_TYPE_Q4_0 = 2
GGML_TYPE_Q4_1 = 3
GGML_TYPE_Q5_0 = 6
GGML_TYPE_Q5_1 = 7
GGML_TYPE_Q8_0 = 8
GGML_TYPE_Q8_1 = 9
GGML_TYPE_Q2_K = 10
GGML_TYPE_Q3_K = 11
GGML_TYPE_Q4_K = 12
GGML_TYPE_Q5_K = 13
GGML_TYPE_Q6_K = 14
GGML_TYPE_Q8_K = 15
GGML_TYPE_IQ2_XXS = 16
GGML_TYPE_IQ2_XS = 17
GGML_TYPE_IQ3_XXS = 18
GGML_TYPE_IQ1_S = 19
GGML_TYPE_IQ4_NL = 20
GGML_TYPE_IQ3_S = 21
GGML_TYPE_IQ2_S = 22
GGML_TYPE_IQ4_XS = 23
GGML_TYPE_I8 = 24
GGML_TYPE_I16 = 25
GGML_TYPE_I32 = 26
GGML_TYPE_I64 = 27
GGML_TYPE_F64 = 28
GGML_TYPE_IQ1_M = 29
GGML_TYPE_COUNT = 30

# from ggml-backend.h
# typedef bool (*ggml_backend_sched_eval_callback)(struct ggml_tensor * t, bool ask, void * user_data);
ggml_backend_sched_eval_callback = ctypes.CFUNCTYPE(
    ctypes.c_bool, ctypes.c_void_p, ctypes.c_bool, ctypes.c_void_p
)

# // Abort callback
# // If not NULL, called before ggml computation
# // If it returns true, the computation is aborted
# typedef bool (*ggml_abort_callback)(void * data);
ggml_abort_callback = ctypes.CFUNCTYPE(ctypes.c_bool, ctypes.c_void_p)

# llama.h bindings

_lib.llama_max_devices.argtypes = []
_lib.llama_max_devices.restype = ctypes.c_size_t

LLAMA_MAX_DEVICES = _lib.llama_max_devices()

# define LLAMA_DEFAULT_SEED 0xFFFFFFFF
LLAMA_DEFAULT_SEED = 0xFFFFFFFF

# define LLAMA_MAX_RNG_STATE (64*1024)
LLAMA_MAX_RNG_STATE = 64 * 1024

# define LLAMA_FILE_MAGIC_GGLA 0x67676c61u // 'ggla'
LLAMA_FILE_MAGIC_GGLA = 0x67676C61

# define LLAMA_FILE_MAGIC_GGSN 0x6767736eu // 'ggsn'
LLAMA_FILE_MAGIC_GGSN = 0x6767736E

# define LLAMA_FILE_MAGIC_GGSQ 0x67677371u // 'ggsq'
LLAMA_FILE_MAGIC_GGSQ = 0x67677371

# define LLAMA_SESSION_MAGIC   LLAMA_FILE_MAGIC_GGSN
LLAMA_SESSION_MAGIC = LLAMA_FILE_MAGIC_GGSN
# define LLAMA_SESSION_VERSION 5
LLAMA_SESSION_VERSION = 5

# define LLAMA_STATE_SEQ_MAGIC   LLAMA_FILE_MAGIC_GGSQ
LLAMA_STATE_SEQ_MAGIC = LLAMA_FILE_MAGIC_GGSQ
# define LLAMA_STATE_SEQ_VERSION 1
LLAMA_STATE_SEQ_VERSION = 1

# struct llama_model;
llama_model_p = NewType("llama_model_p", int)
llama_model_p_ctypes = ctypes.c_void_p

# struct llama_context;
llama_context_p = NewType("llama_context_p", int)
llama_context_p_ctypes = ctypes.c_void_p


# typedef int32_t llama_pos;
llama_pos = ctypes.c_int32
# typedef int32_t llama_token;
llama_token = ctypes.c_int32
llama_token_p = ctypes.POINTER(llama_token)
# typedef int32_t llama_seq_id;
llama_seq_id = ctypes.c_int32


# enum llama_vocab_type {
#     LLAMA_VOCAB_TYPE_NONE = 0, // For models without vocab
#     LLAMA_VOCAB_TYPE_SPM  = 1, // LLaMA tokenizer based on byte-level BPE with byte fallback
#     LLAMA_VOCAB_TYPE_BPE  = 2, // GPT-2 tokenizer based on byte-level BPE
#     LLAMA_VOCAB_TYPE_WPM  = 3, // BERT tokenizer based on WordPiece
# };
LLAMA_VOCAB_TYPE_NONE = 0
"""For models without vocab"""
LLAMA_VOCAB_TYPE_SPM = 1
"""LLaMA tokenizer based on byte-level BPE with byte fallback"""
LLAMA_VOCAB_TYPE_BPE = 2
"""GPT-2 tokenizer based on byte-level BPE"""
LLAMA_VOCAB_TYPE_WPM = 3
"""BERT tokenizer based on WordPiece"""


# // note: these values should be synchronized with ggml_rope
# // TODO: maybe move this enum to ggml.h (ggml_rope_type)
# enum llama_rope_type {
#     LLAMA_ROPE_TYPE_NONE = -1,
#     LLAMA_ROPE_TYPE_NORM =  0,
#     LLAMA_ROPE_TYPE_NEOX =  2,
#     LLAMA_ROPE_TYPE_GLM  =  4,
# };
LLAMA_ROPE_TYPE_NONE = -1
LLAMA_ROPE_TYPE_NORM = 0
LLAMA_ROPE_TYPE_NEOX = 2
LLAMA_ROPE_TYPE_GLM = 4


# enum llama_token_type {
#     LLAMA_TOKEN_TYPE_UNDEFINED    = 0,
#     LLAMA_TOKEN_TYPE_NORMAL       = 1,
#     LLAMA_TOKEN_TYPE_UNKNOWN      = 2,
#     LLAMA_TOKEN_TYPE_CONTROL      = 3,
#     LLAMA_TOKEN_TYPE_USER_DEFINED = 4,
#     LLAMA_TOKEN_TYPE_UNUSED       = 5,
#     LLAMA_TOKEN_TYPE_BYTE         = 6,
# };
LLAMA_TOKEN_TYPE_UNDEFINED = 0
LLAMA_TOKEN_TYPE_NORMAL = 1
LLAMA_TOKEN_TYPE_UNKNOWN = 2
LLAMA_TOKEN_TYPE_CONTROL = 3
LLAMA_TOKEN_TYPE_USER_DEFINED = 4
LLAMA_TOKEN_TYPE_UNUSED = 5
LLAMA_TOKEN_TYPE_BYTE = 6


# // model file types
# enum llama_ftype {
#     LLAMA_FTYPE_ALL_F32              = 0,
#     LLAMA_FTYPE_MOSTLY_F16           = 1,  // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q4_0          = 2,  // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q4_1          = 3,  // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4,  // tok_embeddings.weight and output.weight are F16
#     // LLAMA_FTYPE_MOSTLY_Q4_2       = 5,  // support has been removed
#     // LLAMA_FTYPE_MOSTLY_Q4_3       = 6,  // support has been removed
#     LLAMA_FTYPE_MOSTLY_Q8_0          = 7,  // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q5_0          = 8,  // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q5_1          = 9,  // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q2_K          = 10, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q3_K_S        = 11, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q3_K_M        = 12, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q3_K_L        = 13, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q4_K_S        = 14, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q4_K_M        = 15, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q5_K_S        = 16, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q5_K_M        = 17, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q6_K          = 18, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ2_XXS       = 19, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ2_XS        = 20, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_Q2_K_S        = 21, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ3_XS        = 22, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ3_XXS       = 23, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ1_S         = 24, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ4_NL        = 25, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ3_S         = 26, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ3_M         = 27, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ2_S         = 28, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ2_M         = 29, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ4_XS        = 30, // except 1d tensors
#     LLAMA_FTYPE_MOSTLY_IQ1_M         = 31, // except 1d tensors

#     LLAMA_FTYPE_GUESSED = 1024, // not specified in the model file
# };
LLAMA_FTYPE_ALL_F32 = 0
LLAMA_FTYPE_MOSTLY_F16 = 1
LLAMA_FTYPE_MOSTLY_Q4_0 = 2
LLAMA_FTYPE_MOSTLY_Q4_1 = 3
LLAMA_FTYPE_MOSTLY_Q4_1_SOME_F16 = 4
LLAMA_FTYPE_MOSTLY_Q8_0 = 7
LLAMA_FTYPE_MOSTLY_Q5_0 = 8
LLAMA_FTYPE_MOSTLY_Q5_1 = 9
LLAMA_FTYPE_MOSTLY_Q2_K = 10
LLAMA_FTYPE_MOSTLY_Q3_K_S = 11
LLAMA_FTYPE_MOSTLY_Q3_K_M = 12
LLAMA_FTYPE_MOSTLY_Q3_K_L = 13
LLAMA_FTYPE_MOSTLY_Q4_K_S = 14
LLAMA_FTYPE_MOSTLY_Q4_K_M = 15
LLAMA_FTYPE_MOSTLY_Q5_K_S = 16
LLAMA_FTYPE_MOSTLY_Q5_K_M = 17
LLAMA_FTYPE_MOSTLY_Q6_K = 18
LLAMA_FTYPE_MOSTLY_IQ2_XXS = 19
LLAMA_FTYPE_MOSTLY_IQ2_XS = 20
LLAMA_FTYPE_MOSTLY_Q2_K_S = 21
LLAMA_FTYPE_MOSTLY_IQ3_XS = 22
LLAMA_FTYPE_MOSTLY_IQ3_XXS = 23
LLAMA_FTYPE_MOSTLY_IQ1_S = 24
LLAMA_FTYPE_MOSTLY_IQ4_NL = 25
LLAMA_FTYPE_MOSTLY_IQ3_S = 26
LLAMA_FTYPE_MOSTLY_IQ3_M = 27
LLAMA_FTYPE_MOSTLY_IQ2_S = 28
LLAMA_FTYPE_MOSTLY_IQ2_M = 29
LLAMA_FTYPE_MOSTLY_IQ4_XS = 30
LLAMA_FTYPE_GUESSED = 1024

# enum llama_rope_scaling_type {
#     LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED = -1,
#     LLAMA_ROPE_SCALING_TYPE_NONE        = 0,
#     LLAMA_ROPE_SCALING_TYPE_LINEAR      = 1,
#     LLAMA_ROPE_SCALING_TYPE_YARN        = 2,
#     LLAMA_ROPE_SCALING_TYPE_MAX_VALUE   = LLAMA_ROPE_SCALING_TYPE_YARN,
# };
LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED = -1
LLAMA_ROPE_SCALING_TYPE_NONE = 0
LLAMA_ROPE_SCALING_TYPE_LINEAR = 1
LLAMA_ROPE_SCALING_TYPE_YARN = 2
LLAMA_ROPE_SCALING_TYPE_MAX_VALUE = LLAMA_ROPE_SCALING_TYPE_YARN

# enum llama_pooling_type {
#     LLAMA_POOLING_TYPE_UNSPECIFIED = -1,
#     LLAMA_POOLING_TYPE_NONE = 0,
#     LLAMA_POOLING_TYPE_MEAN = 1,
#     LLAMA_POOLING_TYPE_CLS  = 2,
# };
LLAMA_POOLING_TYPE_UNSPECIFIED = -1
LLAMA_POOLING_TYPE_NONE = 0
LLAMA_POOLING_TYPE_MEAN = 1
LLAMA_POOLING_TYPE_CLS = 2

# enum llama_split_mode {
#     LLAMA_SPLIT_MODE_NONE    = 0, // single GPU
#     LLAMA_SPLIT_MODE_LAYER   = 1, // split layers and KV across GPUs
#     LLAMA_SPLIT_MODE_ROW     = 2, // split rows across GPUs
# };
LLAMA_SPLIT_MODE_NONE = 0
LLAMA_SPLIT_MODE_LAYER = 1
LLAMA_SPLIT_MODE_ROW = 2


# typedef struct llama_token_data {
#     llama_token id; // token id
#     float logit;    // log-odds of the token
#     float p;        // probability of the token
# } llama_token_data;
class llama_token_data(ctypes.Structure):
    """Used to store token data

    Attributes:
        id (llama_token): token id
        logit (float): log-odds of the token
        p (float): probability of the token"""

    if TYPE_CHECKING:
        id: llama_token
        logit: float
        p: float

    _fields_ = [
        ("id", llama_token),
        ("logit", ctypes.c_float),
        ("p", ctypes.c_float),
    ]


llama_token_data_p = ctypes.POINTER(llama_token_data)


# typedef struct llama_token_data_array {
#     llama_token_data * data;
#     size_t size;
#     bool sorted;
# } llama_token_data_array;
class llama_token_data_array(ctypes.Structure):
    """Used to sample tokens given logits

    Attributes:
        data (ctypes.Array[llama_token_data]): token data
        size (int): size of the array
        sorted (bool): whether the array is sorted"""

    if TYPE_CHECKING:
        data: CtypesArray[llama_token_data]
        size: int
        sorted: bool

    _fields_ = [
        ("data", llama_token_data_p),
        ("size", ctypes.c_size_t),
        ("sorted", ctypes.c_bool),
    ]


llama_token_data_array_p = ctypes.POINTER(llama_token_data_array)

# typedef bool (*llama_progress_callback)(float progress, void *ctx);
llama_progress_callback = ctypes.CFUNCTYPE(
    ctypes.c_bool, ctypes.c_float, ctypes.c_void_p
)


# // Input data for llama_decode
# // A llama_batch object can contain input about one or many sequences
# // The provided arrays (i.e. token, embd, pos, etc.) must have size of n_tokens
# //
# // - token  : the token ids of the input (used when embd is NULL)
# // - embd   : token embeddings (i.e. float vector of size n_embd) (used when token is NULL)
# // - pos    : the positions of the respective token in the sequence
# // - seq_id : the sequence to which the respective token belongs
# // - logits : if zero, the logits (and/or the embeddings) for the respective token will not be output
# //
# typedef struct llama_batch {
#     int32_t n_tokens;

#     llama_token  *  token;
#     float        *  embd;
#     llama_pos    *  pos;
#     int32_t      *  n_seq_id;
#     llama_seq_id ** seq_id;
#     int8_t       *  logits; // TODO: rename this to "output"


#     // NOTE: helpers for smooth API transition - can be deprecated in the future
#     //       for future-proof code, use the above fields instead and ignore everything below
#     //
#     // pos[i] = all_pos_0 + i*all_pos_1
#     //
#     llama_pos    all_pos_0;  // used if pos == NULL
#     llama_pos    all_pos_1;  // used if pos == NULL
#     llama_seq_id all_seq_id; // used if seq_id == NULL
# } llama_batch;
class llama_batch(ctypes.Structure):
    """Input data for llama_decode

    A llama_batch object can contain input about one or many sequences

    The provided arrays (i.e. token, embd, pos, etc.) must have size of n_tokens

    Attributes:
        n_tokens (int): number of tokens
        token (ctypes.Array[llama_token]): the token ids of the input (used when embd is NULL)
        embd (ctypes.Array[ctypes.ctypes.c_float]): token embeddings (i.e. float vector of size n_embd) (used when token is NULL)
        pos (ctypes.Array[ctypes.Array[llama_pos]]): the positions of the respective token in the sequence
        seq_id (ctypes.Array[ctypes.Array[llama_seq_id]]): the sequence to which the respective token belongs
        logits (ctypes.Array[ctypes.ctypes.c_int8]): if zero, the logits for the respective token will not be output
    """

    if TYPE_CHECKING:
        n_tokens: int
        token: CtypesArray[llama_token]
        embd: CtypesArray[ctypes.c_float]
        pos: CtypesArray[CtypesArray[llama_pos]]
        n_seq_id: CtypesArray[ctypes.c_int]
        seq_id: CtypesArray[CtypesArray[llama_seq_id]]
        logits: CtypesArray[ctypes.c_int8]

    _fields_ = [
        ("n_tokens", ctypes.c_int32),
        ("token", ctypes.POINTER(llama_token)),
        ("embd", ctypes.POINTER(ctypes.c_float)),
        ("pos", ctypes.POINTER(llama_pos)),
        ("n_seq_id", ctypes.POINTER(ctypes.c_int32)),
        ("seq_id", ctypes.POINTER(ctypes.POINTER(llama_seq_id))),
        ("logits", ctypes.POINTER(ctypes.c_int8)),
        ("all_pos_0", llama_pos),
        ("all_pos_1", llama_pos),
        ("all_seq_id", llama_seq_id),
    ]


# enum llama_model_kv_override_type {
#     LLAMA_KV_OVERRIDE_TYPE_INT,
#     LLAMA_KV_OVERRIDE_TYPE_FLOAT,
#     LLAMA_KV_OVERRIDE_TYPE_BOOL,
# };
LLAMA_KV_OVERRIDE_TYPE_INT = 0
LLAMA_KV_OVERRIDE_TYPE_FLOAT = 1
LLAMA_KV_OVERRIDE_TYPE_BOOL = 2


# struct llama_model_kv_override {
#     char key[128];
#     enum llama_model_kv_override_type tag;
#     union {
#         int64_t int_value;
#         double float_value;
#         bool bool_value;
#     };
# };
class llama_model_kv_override_value(ctypes.Union):
    _fields_ = [
        ("int_value", ctypes.c_int64),
        ("float_value", ctypes.c_double),
        ("bool_value", ctypes.c_bool),
    ]


class llama_model_kv_override(ctypes.Structure):
    _fields_ = [
        ("key", ctypes.c_char * 128),
        ("tag", ctypes.c_int),
        ("value", llama_model_kv_override_value),
    ]


# struct llama_model_params {
#     int32_t n_gpu_layers; // number of layers to store in VRAM
#     enum llama_split_mode split_mode; // how to split the model across multiple GPUs

#     // main_gpu interpretation depends on split_mode:
#     // LLAMA_SPLIT_NONE: the GPU that is used for the entire model
#     // LLAMA_SPLIT_ROW: the GPU that is used for small tensors and intermediate results
#     // LLAMA_SPLIT_LAYER: ignored
#     int32_t main_gpu;

#     // proportion of the model (layers or rows) to offload to each GPU, size: llama_max_devices()
#     const float * tensor_split;

#     // Called with a progress value between 0.0 and 1.0. Pass NULL to disable.
#     // If the provided progress_callback returns true, model loading continues.
#     // If it returns false, model loading is immediately aborted.
#     llama_progress_callback progress_callback;

#     // context pointer passed to the progress callback
#     void * progress_callback_user_data;

#     // override key-value pairs of the model meta data
#     const struct llama_model_kv_override * kv_overrides;


#     // Keep the booleans together to avoid misalignment during copy-by-value.
#     bool vocab_only; // only load the vocabulary, no weights
#     bool use_mmap;   // use mmap if possible
#     bool use_mlock;  // force system to keep model in RAM
# };
class llama_model_params(ctypes.Structure):
    """Parameters for llama_model

    Attributes:
        n_gpu_layers (int): number of layers to store in VRAM
        split_mode (int): how to split the model across multiple GPUs
        main_gpu (int): the GPU that is used for the entire model. main_gpu interpretation depends on split_mode: LLAMA_SPLIT_NONE: the GPU that is used for the entire model LLAMA_SPLIT_ROW: the GPU that is used for small tensors and intermediate results LLAMA_SPLIT_LAYER: ignored
        tensor_split (ctypes.Array[ctypes.ctypes.c_float]): proportion of the model (layers or rows) to offload to each GPU, size: llama_max_devices()
        progress_callback (llama_progress_callback): called with a progress value between 0.0 and 1.0. Pass NULL to disable. If the provided progress_callback returns true, model loading continues. If it returns false, model loading is immediately aborted.
        progress_callback_user_data (ctypes.ctypes.c_void_p): context pointer passed to the progress callback
        kv_overrides (ctypes.Array[llama_model_kv_override]): override key-value pairs of the model meta data
        vocab_only (bool): only load the vocabulary, no weights
        use_mmap (bool): use mmap if possible
        use_mlock (bool): force system to keep model in RAM"""

    if TYPE_CHECKING:
        n_gpu_layers: int
        split_mode: int
        main_gpu: int
        tensor_split: CtypesArray[ctypes.c_float]
        progress_callback: Callable[[float, ctypes.c_void_p], bool]
        progress_callback_user_data: ctypes.c_void_p
        kv_overrides: CtypesArray[llama_model_kv_override]
        vocab_only: bool
        use_mmap: bool
        use_mlock: bool

    _fields_ = [
        ("n_gpu_layers", ctypes.c_int32),
        ("split_mode", ctypes.c_int),
        ("main_gpu", ctypes.c_int32),
        ("tensor_split", ctypes.POINTER(ctypes.c_float)),
        ("progress_callback", llama_progress_callback),
        ("progress_callback_user_data", ctypes.c_void_p),
        ("kv_overrides", ctypes.POINTER(llama_model_kv_override)),
        ("vocab_only", ctypes.c_bool),
        ("use_mmap", ctypes.c_bool),
        ("use_mlock", ctypes.c_bool),
    ]


# struct llama_context_params {
#     uint32_t seed;              // RNG seed, -1 for random
#     uint32_t n_ctx;             // text context, 0 = from model
#     uint32_t n_batch;           // logical maximum batch size that can be submitted to llama_decode
#     uint32_t n_ubatch;          // physical maximum batch size
#     uint32_t n_seq_max;         // max number of sequences (i.e. distinct states for recurrent models)
#     uint32_t n_threads;         // number of threads to use for generation
#     uint32_t n_threads_batch;   // number of threads to use for batch processing

#     enum llama_rope_scaling_type rope_scaling_type; // RoPE scaling type, from `enum llama_rope_scaling_type`
#     enum llama_pooling_type      pooling_type;      // whether to pool (sum) embedding results by sequence id
#                                                     // (ignored if no pooling layer)

#     // ref: https://github.com/ggerganov/llama.cpp/pull/2054
#     float    rope_freq_base;   // RoPE base frequency, 0 = from model
#     float    rope_freq_scale;  // RoPE frequency scaling factor, 0 = from model
#     float    yarn_ext_factor;  // YaRN extrapolation mix factor, negative = from model
#     float    yarn_attn_factor; // YaRN magnitude scaling factor
#     float    yarn_beta_fast;   // YaRN low correction dim
#     float    yarn_beta_slow;   // YaRN high correction dim
#     uint32_t yarn_orig_ctx;    // YaRN original context size
#     float    defrag_thold;     // defragment the KV cache if holes/size > thold, < 0 disabled (default)

#     ggml_backend_sched_eval_callback cb_eval;
#     void * cb_eval_user_data;

#     enum ggml_type type_k; // data type for K cache
#     enum ggml_type type_v; // data type for V cache

#     // Keep the booleans together to avoid misalignment during copy-by-value.
#     bool logits_all;  // the llama_decode() call computes all logits, not just the last one (DEPRECATED - set llama_batch.logits instead)
#     bool embeddings;  // if true, extract embeddings (together with logits)
#     bool offload_kqv; // whether to offload the KQV ops (including the KV cache) to GPU


#     // Abort callback
#     // if it returns true, execution of llama_decode() will be aborted
#     // currently works only with CPU execution
#     ggml_abort_callback abort_callback;
#     void *              abort_callback_data;
# };
class llama_context_params(ctypes.Structure):
    """Parameters for llama_context

    Attributes:
        seed (int): RNG seed, -1 for random
        n_ctx (int): text context, 0 = from model
        n_batch (int): logical maximum batch size that can be submitted to llama_decode
        n_ubatch (int): physical maximum batch size
        n_seq_max (int): max number of sequences (i.e. distinct states for recurrent models)
        n_threads (int): number of threads to use for generation
        n_threads_batch (int): number of threads to use for batch processing
        rope_scaling_type (int): RoPE scaling type, from `enum llama_rope_scaling_type`
        pooling_type (int): whether to pool (sum) embedding results by sequence id (ignored if no pooling layer)
        rope_freq_base (float): RoPE base frequency, 0 = from model
        rope_freq_scale (float): RoPE frequency scaling factor, 0 = from model
        yarn_ext_factor (float): YaRN extrapolation mix factor, negative = from model
        yarn_attn_factor (float): YaRN magnitude scaling factor
        yarn_beta_fast (float): YaRN low correction dim
        yarn_beta_slow (float): YaRN high correction dim
        yarn_orig_ctx (int): YaRN original context size
        defrag_thold (float): defragment the KV cache if holes/size > thold, < 0 disabled (default)
        cb_eval (ggml_backend_sched_eval_callback): callback for scheduling eval
        cb_eval_user_data (ctypes.ctypes.c_void_p): user data for cb_eval
        type_k (int): data type for K cache
        type_v (int): data type for V cache
        logits_all (bool): the llama_eval() call computes all logits, not just the last one (DEPRECATED - set llama_batch.logits instead)
        embeddings (bool): if true, extract embeddings (together with logits)
        offload_kqv (bool): whether to offload the KQV ops (including the KV cache) to GPU
        abort_callback (ggml_abort_callback): abort callback if it returns true, execution of llama_decode() will be aborted
        abort_callback_data (ctypes.ctypes.c_void_p): data for abort_callback
    """

    if TYPE_CHECKING:
        seed: int
        n_ctx: int
        n_batch: int
        n_ubatch: int
        n_seq_max: int
        n_threads: int
        n_threads_batch: int
        rope_scaling_type: int
        pooling_type: int
        rope_freq_base: float
        rope_freq_scale: float
        yarn_ext_factor: float
        yarn_attn_factor: float
        yarn_beta_fast: float
        yarn_beta_slow: float
        yarn_orig_ctx: int
        defrag_thold: float
        cb_eval: Callable[[ctypes.c_void_p, bool], bool]
        cb_eval_user_data: ctypes.c_void_p
        type_k: int
        type_v: int
        logits_all: bool
        embeddings: bool
        offload_kqv: bool
        abort_callback: Callable[[ctypes.c_void_p], bool]
        abort_callback_data: ctypes.c_void_p

    _fields_ = [
        ("seed", ctypes.c_uint32),
        ("n_ctx", ctypes.c_uint32),
        ("n_batch", ctypes.c_uint32),
        ("n_ubatch", ctypes.c_uint32),
        ("n_seq_max", ctypes.c_uint32),
        ("n_threads", ctypes.c_uint32),
        ("n_threads_batch", ctypes.c_uint32),
        ("rope_scaling_type", ctypes.c_int),
        ("pooling_type", ctypes.c_int),
        ("rope_freq_base", ctypes.c_float),
        ("rope_freq_scale", ctypes.c_float),
        ("yarn_ext_factor", ctypes.c_float),
        ("yarn_attn_factor", ctypes.c_float),
        ("yarn_beta_fast", ctypes.c_float),
        ("yarn_beta_slow", ctypes.c_float),
        ("yarn_orig_ctx", ctypes.c_uint32),
        ("defrag_thold", ctypes.c_float),
        ("cb_eval", ggml_backend_sched_eval_callback),
        ("cb_eval_user_data", ctypes.c_void_p),
        ("type_k", ctypes.c_int),
        ("type_v", ctypes.c_int),
        ("logits_all", ctypes.c_bool),
        ("embeddings", ctypes.c_bool),
        ("offload_kqv", ctypes.c_bool),
        ("abort_callback", ggml_abort_callback),
        ("abort_callback_data", ctypes.c_void_p),
    ]


# // Signature for logging events
# // Note that text includes the new line character at the end for most events.
# // If your logging mechanism cannot handle that, check if the last character is '\n' and strip it
# // if it exists.
# // It might not exist for progress report where '.' is output repeatedly.
# typedef void (*llama_log_callback)(enum llama_log_level level, const char * text, void * user_data);
llama_log_callback = ctypes.CFUNCTYPE(
    None, ctypes.c_int, ctypes.c_char_p, ctypes.c_void_p
)
"""Signature for logging events
Note that text includes the new line character at the end for most events.
If your logging mechanism cannot handle that, check if the last character is '\n' and strip it
if it exists.
It might not exist for progress report where '.' is output repeatedly."""


# // model quantization parameters
# typedef struct llama_model_quantize_params {
#     int32_t nthread;                     // number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
#     enum llama_ftype ftype;              // quantize to this llama_ftype
#     enum ggml_type output_tensor_type;   // output tensor type
#     enum ggml_type token_embedding_type; // itoken embeddings tensor type
#     bool allow_requantize;               // allow quantizing non-f32/f16 tensors
#     bool quantize_output_tensor;         // quantize output.weight
#     bool only_copy;                      // only copy tensors - ftype, allow_requantize and quantize_output_tensor are ignored
#     bool pure;                           // quantize all tensors to the default type
#     bool keep_split;                     // quantize to the same number of shards
#     void * imatrix;                      // pointer to importance matrix data
#     void * kv_overrides;                 // pointer to vector containing overrides
# } llama_model_quantize_params;
class llama_model_quantize_params(ctypes.Structure):
    """Parameters for llama_model_quantize

    Attributes:
        nthread (int): number of threads to use for quantizing, if <=0 will use std::thread::hardware_concurrency()
        ftype (int): quantize to this llama_ftype
        output_tensor_type (int): output tensor type
        token_embedding_type (int): itoken embeddings tensor type
        allow_requantize (bool): allow quantizing non-f32/f16 tensors
        quantize_output_tensor (bool): quantize output.weight
        only_copy (bool): only copy tensors - ftype, allow_requantize and quantize_output_tensor are ignored
        pure (bool): quantize all tensors to the default type
        keep_split (bool): quantize to the same number of shards
        imatrix (ctypes.c_void_p): pointer to importance matrix data
        kv_overrides (ctypes.c_void_p): pointer to vector containing overrides
    """

    if TYPE_CHECKING:
        nthread: int
        ftype: int
        output_tensor_type: int
        token_embedding_type: int
        allow_requantize: bool
        quantize_output_tensor: bool
        only_copy: bool
        pure: bool
        keep_split: bool
        imatrix: ctypes.c_void_p
        kv_overrides: ctypes.c_void_p

    _fields_ = [
        ("nthread", ctypes.c_int32),
        ("ftype", ctypes.c_int),
        ("output_tensor_type", ctypes.c_int),
        ("token_embedding_type", ctypes.c_int),
        ("allow_requantize", ctypes.c_bool),
        ("quantize_output_tensor", ctypes.c_bool),
        ("only_copy", ctypes.c_bool),
        ("pure", ctypes.c_bool),
        ("keep_split", ctypes.c_bool),
        ("imatrix", ctypes.c_void_p),
        ("kv_overrides", ctypes.c_void_p),
    ]


# // grammar types
# struct llama_grammar;
llama_grammar_p = ctypes.c_void_p

# // grammar element type
# enum llama_gretype {
#     // end of rule definition
#     LLAMA_GRETYPE_END            = 0,

#     // start of alternate definition for rule
#     LLAMA_GRETYPE_ALT            = 1,

#     // non-terminal element: reference to rule
#     LLAMA_GRETYPE_RULE_REF       = 2,

#     // terminal element: character (code point)
#     LLAMA_GRETYPE_CHAR           = 3,

#     // inverse char(s) ([^a], [^a-b] [^abc])
#     LLAMA_GRETYPE_CHAR_NOT       = 4,

#     // modifies a preceding LLAMA_GRETYPE_CHAR or LLAMA_GRETYPE_CHAR_ALT to
#     // be an inclusive range ([a-z])
#     LLAMA_GRETYPE_CHAR_RNG_UPPER = 5,

#     // modifies a preceding LLAMA_GRETYPE_CHAR or
#     // LLAMA_GRETYPE_CHAR_RNG_UPPER to add an alternate char to match ([ab], [a-zA])
#     LLAMA_GRETYPE_CHAR_ALT       = 6,
# };
LLAMA_GRETYPE_END = 0
LLAMA_GRETYPE_ALT = 1
LLAMA_GRETYPE_RULE_REF = 2
LLAMA_GRETYPE_CHAR = 3
LLAMA_GRETYPE_CHAR_NOT = 4
LLAMA_GRETYPE_CHAR_RNG_UPPER = 5
LLAMA_GRETYPE_CHAR_ALT = 6


# typedef struct llama_grammar_element {
#     enum llama_gretype type;
#     uint32_t           value; // Unicode code point or rule ID
# } llama_grammar_element;
class llama_grammar_element(ctypes.Structure):
    if TYPE_CHECKING:
        type: int
        value: int

    _fields_ = [
        ("type", ctypes.c_int),
        ("value", ctypes.c_uint32),
    ]


llama_grammar_element_p = ctypes.POINTER(llama_grammar_element)

# // performance timing information
# struct llama_timings {
#     double t_start_ms;
#     double t_end_ms;
#     double t_load_ms;
#     double t_sample_ms;
#     double t_p_eval_ms;
#     double t_eval_ms;


#     int32_t n_sample;
#     int32_t n_p_eval;
#     int32_t n_eval;
# };
class llama_timings(ctypes.Structure):
    if TYPE_CHECKING:
        t_start_ms: float
        t_end_ms: float
        t_load_ms: float
        t_sample_ms: float
        t_p_eval_ms: float
        t_eval_ms: float
        n_sample: int
        n_p_eval: int
        n_eval: int

    _fields_ = [
        ("t_start_ms", ctypes.c_double),
        ("t_end_ms", ctypes.c_double),
        ("t_load_ms", ctypes.c_double),
        ("t_sample_ms", ctypes.c_double),
        ("t_p_eval_ms", ctypes.c_double),
        ("t_eval_ms", ctypes.c_double),
        ("n_sample", ctypes.c_int32),
        ("n_p_eval", ctypes.c_int32),
        ("n_eval", ctypes.c_int32),
    ]


# // used in chat template
# typedef struct llama_chat_message {
#     const char * role;
#     const char * content;
# } llama_chat_message;
class llama_chat_message(ctypes.Structure):
    _fields_ = [
        ("role", ctypes.c_char_p),
        ("content", ctypes.c_char_p),
    ]


# // Helpers for getting default parameters
# LLAMA_API struct llama_model_params llama_model_default_params(void);
@ctypes_function(
    "llama_model_default_params",
    [],
    llama_model_params,
)
def llama_model_default_params() -> llama_model_params:
    """Get default parameters for llama_model"""
    ...


# LLAMA_API struct llama_context_params llama_context_default_params(void);
@ctypes_function(
    "llama_context_default_params",
    [],
    llama_context_params,
)
def llama_context_default_params() -> llama_context_params:
    """Get default parameters for llama_context"""
    ...


# LLAMA_API struct llama_model_quantize_params llama_model_quantize_default_params(void);
@ctypes_function(
    "llama_model_quantize_default_params",
    [],
    llama_model_quantize_params,
)
def llama_model_quantize_default_params() -> llama_model_quantize_params:
    """Get default parameters for llama_model_quantize"""
    ...