-
Notifications
You must be signed in to change notification settings - Fork 3
Expand file tree
/
Copy pathmain.py
More file actions
45 lines (37 loc) · 2.04 KB
/
Copy pathmain.py
File metadata and controls
45 lines (37 loc) · 2.04 KB
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
from lora_merging.merge_two_lora import merge_lora_models
from lora_merging.merge_lora_checkpoint import merge_lora_into_checkpoint
from checkpoint_merging.merge_checkpoints import merge_checkpoints
from helpers.file_utils import load_model, save_model
def main():
import argparse
parser = argparse.ArgumentParser(description="Merging LoRA and Checkpoint Models")
parser.add_argument("--operation", type=str, required=True, choices=["merge_loras", "merge_lora_checkpoint", "merge_checkpoints"], help="Choose operation")
parser.add_argument("--model1", type=str, required=True, help="Path to the first model file")
parser.add_argument("--model2", type=str, required=False, help="Path to the second model file")
parser.add_argument("--output", type=str, required=True, help="Path for the output merged model")
parser.add_argument("--alpha", type=float, default=0.5, help="Blend ratio for merging two models")
parser.add_argument("--merge_weight", type=float, default=0.5, help="Weight for merging LoRA into a checkpoint")
args = parser.parse_args()
# Load models
model1 = load_model(args.model1)
model2 = load_model(args.model2) if args.model2 else None
# Perform merging operations
if args.operation == "merge_loras":
if model2 is None:
print("Error: Please provide two LoRA models for merging.")
return
merged_model = merge_lora_models(model1, model2, args.alpha)
elif args.operation == "merge_lora_checkpoint":
if model2 is None:
print("Error: Please provide a checkpoint to merge the LoRA into.")
return
merged_model = merge_lora_into_checkpoint(model1, model2, args.merge_weight)
elif args.operation == "merge_checkpoints":
if model2 is None:
print("Error: Please provide two checkpoints for merging.")
return
merged_model = merge_checkpoints(model1, model2, args.alpha)
# Save the merged model
save_model(merged_model, args.output)
if __name__ == "__main__":
main()