Skip to content

feat: add chat template to existed gguf file #6156

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from
Closed
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
69 changes: 69 additions & 0 deletions gguf-py/scripts/gguf-add-chat-template.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,69 @@
import sys
import struct
import numpy as np
from gguf import GGUFReader, GGUFValueType, GGUF_DEFAULT_ALIGNMENT


# add chat template to existed gguf and retrun new file path
def add_chat_template_to_existed_gguf(model_path: str, chat_template: str, new_file_path: str) -> None:
reader = GGUFReader(model_path, "r+")

# chat template key
CHAT_TEMPLATE = "tokenizer.chat_template"

# generate and adjust chat template data from chat template with alignment
alignment = GGUF_DEFAULT_ALIGNMENT
new_align = reader.fields.get('general.alignment')
if new_align is not None:
alignment = new_align.parts[-1][0]

add_data = bytearray()

name_data = CHAT_TEMPLATE.encode("utf-8")
add_data += struct.pack("Q", len(name_data))
add_data += name_data
add_data += struct.pack("I", GGUFValueType.STRING.value)

# calculate the padding length
raw_len = len(add_data) + 8 + len(chat_template)
add_len = alignment - (raw_len % alignment)
if add_len != 0:
chat_template += " " * add_len

raw_data = chat_template.encode("utf-8")
add_data += struct.pack("Q", len(raw_data))
add_data += raw_data

# insert raw bytes into file
# find insert index
kv = reader.fields
last_field = list(kv.values())[-1]
insert_offset = last_field.offset

# copy original data
new_data = reader.data.copy()
new_data = np.concatenate(
(new_data[:insert_offset], add_data, new_data[insert_offset:]))

# add kv_count
kv_count_idx = reader.fields["GGUF.kv_count"].parts[0][0]
new_data[kv_count_idx] += 1

# save file
with open(new_file_path, "wb") as file:
file.write(new_data.tobytes())

# sample usage: python guff-add-chat-template.py "./model.gguf" "chat template" "./new_model.gguf"
def main() -> None:
parser = argparse.ArgumentParser(description="Add chat template to existed gguf file")
parser.add_argument("model", type=str, help="GGUF format model filepath")
parser.add_argument("chat_template", type=str, help="chat template for this model")
parser.add_argument("new_model", type=str, help="New GGUF model filepath")
args = parser.parse_args(None if len(sys.argv) > 1 else ["--help"])
print(f'adding chat template to {args.model}')
add_chat_template_to_existed_gguf(args.model, args.chat_template, args.new_model)
print(f'new model {args.new_model} with chat template added successfully!')


if __name__ == '__main__':
main()