llama.cpp/convert_lora_to_gguf.py

377 lines
14 KiB
Python
Raw Normal View History

2024-07-08 16:35:27 +02:00
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
from __future__ import annotations
from dataclasses import dataclass
2024-07-08 16:35:27 +02:00
import logging
import argparse
import os
import sys
import json
from math import prod
2024-07-08 16:35:27 +02:00
from pathlib import Path
from typing import TYPE_CHECKING, Any, Callable, Iterable, Iterator, Sequence, SupportsIndex, cast
2024-07-08 16:35:27 +02:00
import torch
if TYPE_CHECKING:
from torch import Tensor
if 'NO_LOCAL_GGUF' not in os.environ:
sys.path.insert(1, str(Path(__file__).parent / 'gguf-py'))
import gguf
# reuse model definitions from convert_hf_to_gguf.py
from convert_hf_to_gguf import LazyTorchTensor, Model
2024-07-08 16:35:27 +02:00
logger = logging.getLogger("lora-to-gguf")
2024-07-08 21:55:41 +02:00
@dataclass
class PartialLoraTensor:
A: Tensor | None = None
B: Tensor | None = None
# magic to support tensor shape modifications and splitting
class LoraTorchTensor:
_lora_A: Tensor # (n_rank, row_size)
_lora_B: Tensor # (col_size, n_rank)
_rank: int
def __init__(self, A: Tensor, B: Tensor):
assert len(A.shape) == len(B.shape)
assert A.shape[-2] == B.shape[-1]
if A.dtype != B.dtype:
A = A.to(torch.float32)
B = B.to(torch.float32)
self._lora_A = A
self._lora_B = B
self._rank = B.shape[-1]
def get_lora_A_B(self) -> tuple[Tensor, Tensor]:
return (self._lora_A, self._lora_B)
def __getitem__(
self,
indices: (
SupportsIndex
| slice
| tuple[SupportsIndex | slice | Tensor, ...] # TODO: add ellipsis in the type signature
),
) -> LoraTorchTensor:
shape = self.shape
if isinstance(indices, SupportsIndex):
if len(shape) > 2:
return LoraTorchTensor(self._lora_A[indices], self._lora_B[indices])
else:
raise NotImplementedError # can't return a vector
elif isinstance(indices, slice):
if len(shape) > 2:
return LoraTorchTensor(self._lora_A[indices], self._lora_B[indices])
else:
return LoraTorchTensor(self._lora_A, self._lora_B[indices])
elif isinstance(indices, tuple):
assert len(indices) > 0
if indices[-1] is Ellipsis:
return self[indices[:-1]]
# expand ellipsis
indices = tuple(
u
for v in (
(
(slice(None, None) for _ in range(len(indices) - 1))
if i is Ellipsis
else (i,)
)
for i in indices
)
for u in v
)
if len(indices) < len(shape):
indices = (*indices, *(slice(None, None) for _ in range(len(indices), len(shape))))
# TODO: make sure this is correct
indices_A = (
*(
(
j.__index__() % self._lora_A.shape[i]
if isinstance(j, SupportsIndex)
else slice(None, None)
)
for i, j in enumerate(indices[:-2])
),
slice(None, None),
indices[-1],
)
indices_B = indices[:-1]
return LoraTorchTensor(self._lora_A[indices_A], self._lora_B[indices_B])
else:
raise NotImplementedError # unknown indice type
@property
def dtype(self) -> torch.dtype:
assert self._lora_A.dtype == self._lora_B.dtype
return self._lora_A.dtype
@property
def shape(self) -> tuple[int, ...]:
assert len(self._lora_A.shape) == len(self._lora_B.shape)
return (*self._lora_B.shape[:-1], self._lora_A.shape[-1])
def size(self, dim=None):
assert dim is None
return self.shape
def reshape(self, *shape: int | tuple[int, ...]) -> LoraTorchTensor:
if isinstance(shape[0], tuple):
new_shape: tuple[int, ...] = shape[0]
else:
new_shape = cast(tuple[int, ...], shape)
orig_shape = self.shape
if len(new_shape) < 2:
raise NotImplementedError # can't become a vector
# expand -1 in the shape
if any(dim == -1 for dim in new_shape):
n_elems = prod(orig_shape)
n_new_elems = prod(dim if dim != -1 else 1 for dim in new_shape)
assert n_elems % n_new_elems == 0
new_shape = (*(dim if dim != -1 else n_elems // n_new_elems for dim in new_shape),)
if new_shape[-1] != orig_shape[-1]:
raise NotImplementedError # can't reshape the row size trivially
shape_A = (*(1 for _ in new_shape[:-2]), self._rank, orig_shape[-1])
shape_B = (*new_shape[:-1], self._rank)
return LoraTorchTensor(
self._lora_A.reshape(shape_A),
self._lora_B.reshape(shape_B),
)
def reshape_as(self, other: Tensor) -> LoraTorchTensor:
return self.reshape(*other.shape)
def view(self, *size: int) -> LoraTorchTensor:
return self.reshape(*size)
def permute(self, *dims: int) -> LoraTorchTensor:
shape = self.shape
dims = tuple(dim - len(shape) if dim >= 0 else dim for dim in dims)
if dims[-1] == -1:
# TODO: support higher dimensional A shapes bigger than 1
assert all(dim == 1 for dim in self._lora_A.shape[:-2])
return LoraTorchTensor(self._lora_A, self._lora_B.permute(*dims))
if len(shape) == 2 and dims[-1] == -2 and dims[-2] == -1:
return LoraTorchTensor(self._lora_B.permute(*dims), self._lora_A.permute(*dims))
else:
# TODO: compose the above two
raise NotImplementedError
def transpose(self, dim0: int, dim1: int) -> LoraTorchTensor:
shape = self.shape
dims = [i for i in range(len(shape))]
dims[dim0], dims[dim1] = dims[dim1], dims[dim0]
return self.permute(*dims)
def swapaxes(self, axis0: int, axis1: int) -> LoraTorchTensor:
return self.transpose(axis0, axis1)
def to(self, *args, **kwargs):
return LoraTorchTensor(self._lora_A.to(*args, **kwargs), self._lora_B.to(*args, **kwargs))
@classmethod
def __torch_function__(cls, func: Callable, types, args=(), kwargs=None):
del types # unused
if kwargs is None:
kwargs = {}
if func is torch.permute:
return type(args[0]).permute(*args, **kwargs)
elif func is torch.reshape:
return type(args[0]).reshape(*args, **kwargs)
elif func is torch.stack:
assert isinstance(args[0], Sequence)
dim = kwargs.get("dim", 0)
assert dim == 0
return LoraTorchTensor(
torch.stack([a._lora_A for a in args[0]], dim),
torch.stack([b._lora_B for b in args[0]], dim),
)
elif func is torch.cat:
assert isinstance(args[0], Sequence)
dim = kwargs.get("dim", 0)
assert dim == 0
if len(args[0][0].shape) > 2:
return LoraTorchTensor(
torch.cat([a._lora_A for a in args[0]], dim),
torch.cat([b._lora_B for b in args[0]], dim),
)
elif all(torch.equal(args[0][0]._lora_A, t._lora_A) for t in args[0][1:]):
return LoraTorchTensor(
args[0][0]._lora_A,
torch.cat([b._lora_B for b in args[0]], dim),
)
else:
raise NotImplementedError
else:
raise NotImplementedError
2024-07-10 00:23:07 +02:00
def get_base_tensor_name(lora_tensor_name: str) -> str:
base_name = lora_tensor_name.replace("base_model.model.", "")
base_name = base_name.replace(".lora_A.weight", ".weight")
base_name = base_name.replace(".lora_B.weight", ".weight")
return base_name
2024-07-08 16:35:27 +02:00
def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser(
2024-07-08 21:55:41 +02:00
description="Convert a huggingface PEFT LoRA adapter to a GGML compatible file")
2024-07-08 16:35:27 +02:00
parser.add_argument(
"--outfile", type=Path,
2024-07-08 22:05:35 +02:00
help="path to write to; default: based on input. {ftype} will be replaced by the outtype.",
2024-07-08 16:35:27 +02:00
)
parser.add_argument(
"--outtype", type=str, choices=["f32", "f16", "bf16", "q8_0", "auto"], default="f16",
help="output format - use f32 for float32, f16 for float16, bf16 for bfloat16, q8_0 for Q8_0, auto for the highest-fidelity 16-bit float type depending on the first loaded tensor type",
2024-07-08 16:35:27 +02:00
)
parser.add_argument(
"--bigendian", action="store_true",
help="model is executed on big endian machine",
)
parser.add_argument(
"--no-lazy", action="store_true",
help="use more RAM by computing all outputs before writing (use in case lazy evaluation is broken)",
)
2024-07-08 16:35:27 +02:00
parser.add_argument(
"--verbose", action="store_true",
help="increase output verbosity",
)
parser.add_argument(
"--base", type=Path, required=True,
help="directory containing base model file",
)
parser.add_argument(
"lora_path", type=Path,
help="directory containing LoRA adapter file",
)
return parser.parse_args()
if __name__ == '__main__':
args = parse_args()
logging.basicConfig(level=logging.DEBUG if args.verbose else logging.INFO)
ftype_map: dict[str, gguf.LlamaFileType] = {
"f32": gguf.LlamaFileType.ALL_F32,
"f16": gguf.LlamaFileType.MOSTLY_F16,
"bf16": gguf.LlamaFileType.MOSTLY_BF16,
"q8_0": gguf.LlamaFileType.MOSTLY_Q8_0,
"auto": gguf.LlamaFileType.GUESSED,
2024-07-08 16:35:27 +02:00
}
2024-07-08 21:55:41 +02:00
ftype = ftype_map[args.outtype]
2024-07-08 16:35:27 +02:00
dir_base_model: Path = args.base
dir_lora: Path = args.lora_path
lora_config = dir_lora / "adapter_config.json"
input_model = dir_lora / "adapter_model.safetensors"
2024-07-08 16:35:27 +02:00
if args.outfile is not None:
fname_out = args.outfile
else:
# output in the same directory as the model by default
2024-07-08 22:05:35 +02:00
fname_out = dir_lora / 'ggml-lora-{ftype}.gguf'
2024-07-08 16:35:27 +02:00
if os.path.exists(input_model):
# lazy import load_file only if lora is in safetensors format.
from safetensors.torch import load_file
2024-07-08 16:35:27 +02:00
lora_model = load_file(input_model, device="cpu")
else:
input_model = os.path.join(dir_lora, "adapter_model.bin")
lora_model = torch.load(input_model, map_location="cpu", weights_only=True)
2024-07-08 16:35:27 +02:00
# load base model
logger.info(f"Loading base model: {dir_base_model.name}")
hparams = Model.load_hparams(dir_base_model)
with torch.inference_mode():
try:
model_class = Model.from_model_architecture(hparams["architectures"][0])
except NotImplementedError:
logger.error(f"Model {hparams['architectures'][0]} is not supported")
sys.exit(1)
class LoraModel(model_class):
model_arch = model_class.model_arch
2024-07-08 16:35:27 +02:00
def get_tensors(self) -> Iterator[tuple[str, Tensor]]:
tensor_map: dict[str, PartialLoraTensor] = {}
2024-07-08 16:35:27 +02:00
for name, tensor in lora_model.items():
if self.lazy:
tensor = LazyTorchTensor.from_eager(tensor)
base_name = get_base_tensor_name(name)
is_lora_a = ".lora_A.weight" in name
is_lora_b = ".lora_B.weight" in name
if not is_lora_a and not is_lora_b:
if ".base_layer.weight" in name:
continue
logger.error(f"Unexpected name '{name}': Not a lora_A or lora_B tensor")
sys.exit(1)
2024-07-08 16:35:27 +02:00
if base_name in tensor_map:
if is_lora_a:
tensor_map[base_name].A = tensor
else:
tensor_map[base_name].B = tensor
else:
if is_lora_a:
tensor_map[base_name] = PartialLoraTensor(A=tensor)
else:
tensor_map[base_name] = PartialLoraTensor(B=tensor)
2024-07-08 16:35:27 +02:00
for name, tensor in tensor_map.items():
assert tensor.A is not None
assert tensor.B is not None
yield (name, cast(torch.Tensor, LoraTorchTensor(tensor.A, tensor.B)))
2024-07-08 17:05:17 +02:00
def modify_tensors(self, data_torch: Tensor, name: str, bid: int | None) -> Iterable[tuple[str, Tensor]]:
dest = super().modify_tensors(data_torch, name, bid)
for dest_name, dest_data in dest:
assert isinstance(dest_data, LoraTorchTensor)
lora_a, lora_b = dest_data.get_lora_A_B()
yield (dest_name + ".lora_a", lora_a)
yield (dest_name + ".lora_b", lora_b)
model_instance = LoraModel(
dir_base_model,
ftype,
fname_out,
is_big_endian=args.bigendian,
use_temp_file=False,
eager=args.no_lazy,
model_name=None,
)
logger.info("Set model parameters")
model_instance.set_gguf_parameters()
2024-07-10 00:23:07 +02:00
with open(lora_config, "r") as f:
lparams: dict[str, Any] = json.load(f)
alpha = lparams["lora_alpha"]
model_instance.gguf_writer.add_string("training.type", "finetune_lora")
model_instance.gguf_writer.add_float32("training.lora.alpha", float(alpha))
2024-07-08 21:55:41 +02:00
model_instance.gguf_writer.add_quantization_version(gguf.GGML_QUANT_VERSION)
logger.info("Exporting model...")
model_instance.write()
logger.info(f"Model successfully exported to {model_instance.fname_out}")