File size: 1,565 Bytes
6dc0c9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
"""
Apply the LoRA weights on top of a base model.

Usage:
python3 -m fastchat.model.apply_lora --base ~/model_weights/llama-7b --target ~/model_weights/baize-7b --lora project-baize/baize-lora-7B

Dependency:
pip3 install git+https://github.com/huggingface/peft.git@2822398fbe896f25d4dac5e468624dc5fd65a51b
"""
import argparse

import torch
from peft import PeftModel
from transformers import AutoTokenizer, AutoModelForCausalLM


def apply_lora(base_model_path, target_model_path, lora_path):
    print(f"Loading the base model from {base_model_path}")
    base = AutoModelForCausalLM.from_pretrained(
        base_model_path, torch_dtype=torch.float16, low_cpu_mem_usage=True
    )
    base_tokenizer = AutoTokenizer.from_pretrained(base_model_path, use_fast=False)

    print(f"Loading the LoRA adapter from {lora_path}")

    lora_model = PeftModel.from_pretrained(
        base,
        lora_path,
        # torch_dtype=torch.float16
    )

    print("Applying the LoRA")
    model = lora_model.merge_and_unload()

    print(f"Saving the target model to {target_model_path}")
    model.save_pretrained(target_model_path)
    base_tokenizer.save_pretrained(target_model_path)


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument("--base-model-path", type=str, required=True)
    parser.add_argument("--target-model-path", type=str, required=True)
    parser.add_argument("--lora-path", type=str, required=True)

    args = parser.parse_args()

    apply_lora(args.base_model_path, args.target_model_path, args.lora_path)