llama.cpp/convert-lora-to-ggml.py

142 lines
4.6 KiB
Python
Raw Permalink Normal View History

#!/usr/bin/env python3
from __future__ import annotations
2023-04-17 15:28:55 +00:00
import json
import os
import struct
import sys
from typing import Any, BinaryIO, Sequence
2023-04-17 15:28:55 +00:00
2023-08-23 14:46:54 +00:00
import numpy as np
2023-04-17 15:28:55 +00:00
import torch
from pathlib import Path
if 'NO_LOCAL_GGUF' not in os.environ:
sys.path.insert(1, str(Path(__file__).parent / 'gguf-py' / 'gguf'))
import gguf
2023-04-17 15:28:55 +00:00
NUMPY_TYPE_TO_FTYPE: dict[str, int] = {"float32": 0, "float16": 1}
2023-04-17 15:28:55 +00:00
def write_file_header(fout: BinaryIO, params: dict[str, Any]) -> None:
2023-04-17 15:28:55 +00:00
fout.write(b"ggla"[::-1]) # magic (ggml lora)
fout.write(struct.pack("i", 1)) # file version
fout.write(struct.pack("i", params["r"]))
# https://opendelta.readthedocs.io/en/latest/modules/deltas.html says that `lora_alpha` is an int
# but some models ship a float value instead
# let's convert to int, but fail if lossless conversion is not possible
2023-08-23 14:46:54 +00:00
assert (
int(params["lora_alpha"]) == params["lora_alpha"]
), "cannot convert float to int losslessly"
fout.write(struct.pack("i", int(params["lora_alpha"])))
2023-04-17 15:28:55 +00:00
def write_tensor_header(fout: BinaryIO, name: str, shape: Sequence[int], data_type: np.dtype[Any]) -> None:
2023-04-17 15:28:55 +00:00
sname = name.encode("utf-8")
fout.write(
struct.pack(
"iii",
len(shape),
len(sname),
2023-08-23 14:46:54 +00:00
NUMPY_TYPE_TO_FTYPE[data_type.name],
2023-04-17 15:28:55 +00:00
)
)
fout.write(struct.pack("i" * len(shape), *shape[::-1]))
fout.write(sname)
fout.seek((fout.tell() + 31) & -32)
if len(sys.argv) < 2:
print(f"Usage: python {sys.argv[0]} <path> [arch]")
2023-04-17 15:28:55 +00:00
print(
"Path must contain HuggingFace PEFT LoRA files 'adapter_config.json' and 'adapter_model.bin'"
)
print(f"Arch must be one of {list(gguf.MODEL_ARCH_NAMES.values())} (default: llama)")
2023-04-17 15:28:55 +00:00
sys.exit(1)
input_json = os.path.join(sys.argv[1], "adapter_config.json")
input_model = os.path.join(sys.argv[1], "adapter_model.bin")
output_path = os.path.join(sys.argv[1], "ggml-adapter-model.bin")
model = torch.load(input_model, map_location="cpu")
arch_name = sys.argv[2] if len(sys.argv) == 3 else "llama"
if arch_name not in gguf.MODEL_ARCH_NAMES.values():
print(f"Error: unsupported architecture {arch_name}")
sys.exit(1)
arch = list(gguf.MODEL_ARCH_NAMES.keys())[list(gguf.MODEL_ARCH_NAMES.values()).index(arch_name)]
name_map = gguf.TensorNameMap(arch, 200) # 200 layers ought to be enough for anyone
2023-04-17 15:28:55 +00:00
with open(input_json, "r") as f:
params = json.load(f)
if params["peft_type"] != "LORA":
print(f"Error: unsupported adapter type {params['peft_type']}, expected LORA")
sys.exit(1)
if params["fan_in_fan_out"] is True:
2023-04-17 15:28:55 +00:00
print("Error: param fan_in_fan_out is not supported")
sys.exit(1)
if params["bias"] is not None and params["bias"] != "none":
print("Error: param bias is not supported")
sys.exit(1)
# TODO: these seem to be layers that have been trained but without lora.
# doesn't seem widely used but eventually should be supported
if params["modules_to_save"] is not None and len(params["modules_to_save"]) > 0:
print("Error: param modules_to_save is not supported")
sys.exit(1)
with open(output_path, "wb") as fout:
fout.truncate()
write_file_header(fout, params)
for k, v in model.items():
orig_k = k
if k.endswith(".default.weight"):
k = k.replace(".default.weight", ".weight")
if k in ["llama_proj.weight", "llama_proj.bias"]:
continue
2023-04-17 15:28:55 +00:00
if k.endswith("lora_A.weight"):
if v.dtype != torch.float16 and v.dtype != torch.float32:
v = v.float()
v = v.T
else:
v = v.float()
t = v.detach().numpy()
prefix = "base_model.model."
if k.startswith(prefix):
k = k[len(prefix) :]
lora_suffixes = (".lora_A.weight", ".lora_B.weight")
if k.endswith(lora_suffixes):
suffix = k[-len(lora_suffixes[0]):]
k = k[: -len(lora_suffixes[0])]
else:
print(f"Error: unrecognized tensor name {orig_k}")
sys.exit(1)
tname = name_map.get_name(k)
if tname is None:
print(f"Error: could not map tensor name {orig_k}")
print(" Note: the arch parameter must be specified if the model is not llama")
sys.exit(1)
if suffix == ".lora_A.weight":
tname += ".weight.loraA"
elif suffix == ".lora_B.weight":
tname += ".weight.loraB"
else:
assert False
2023-04-17 15:28:55 +00:00
print(f"{k} => {tname} {t.shape} {t.dtype} {t.nbytes/1024/1024:.2f}MB")
write_tensor_header(fout, tname, t.shape, t.dtype)
t.tofile(fout)
print(f"Converted {input_json} and {input_model} to {output_path}")