Switch to fast RMS/LN Norm (#603)

* use nn.RMSNorm, use sdpa, cleanup

* bump mlx versions

* minor update

* use fast layer norm

* version bump

* update requirement for whisper

* update requirement for gguf
This commit is contained in:
Awni Hannun
2024-03-23 07:13:51 -07:00
committed by GitHub
parent fbed720d6f
commit b8a348c1b8
44 changed files with 144 additions and 1155 deletions

View File

@@ -1,4 +1,4 @@
# Copyright © 2023 Apple Inc.
# Copyright © 2023-2024 Apple Inc.
import argparse
from pathlib import Path
@@ -7,7 +7,7 @@ import mlx.core as mx
import mlx.nn as nn
import utils
from mlx.utils import tree_flatten, tree_unflatten
from models.lora import LoRALinear
from models import LoRALinear
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="LoRA or QLoRA finetuning.")