Buckets:
# LoRA
LoRA adalah metode pelatihan yang cepat dan ringan, yang menyisipkan dan melatih jumlah parameter yang jauh lebih sedikit dibandingkan seluruh parameter model. Hal ini menghasilkan file yang lebih kecil (~100 MB) dan memudahkan pelatihan model dengan cepat untuk mempelajari konsep baru. Bobot LoRA biasanya dimuat ke dalam denoiser, text encoder, atau keduanya. Denoiser biasanya merujuk pada UNet (misalnya [UNet2DConditionModel](/docs/diffusers/main/id/api/models/unet2d-cond#diffusers.UNet2DConditionModel)) atau Transformer (misalnya [SD3Transformer2DModel](/docs/diffusers/main/id/api/models/sd3_transformer2d#diffusers.SD3Transformer2DModel)). Terdapat beberapa kelas untuk memuat bobot LoRA:
- `StableDiffusionLoraLoaderMixin` menyediakan fungsi untuk memuat dan melepas, menggabungkan dan memisahkan, mengaktifkan dan menonaktifkan, serta fungsi lain untuk mengelola bobot LoRA. Kelas ini dapat digunakan dengan model apa pun.
- `StableDiffusionXLLoraLoaderMixin` adalah versi [Stable Diffusion (SDXL)](../../api/pipelines/stable_diffusion/stable_diffusion_xl) dari kelas `StableDiffusionLoraLoaderMixin` untuk memuat dan menyimpan bobot LoRA. Kelas ini hanya dapat digunakan dengan model SDXL.
- `SD3LoraLoaderMixin` menyediakan fungsi serupa untuk [Stable Diffusion 3](https://huggingface.co/blog/sd3).
- `FluxLoraLoaderMixin` menyediakan fungsi serupa untuk [Flux](../pipelines/flux).
- `CogVideoXLoraLoaderMixin` menyediakan fungsi serupa untuk [CogVideoX](../pipelines/cogvideox).
- `Mochi1LoraLoaderMixin` menyediakan fungsi serupa untuk [Mochi](../pipelines/mochi).
- `AuraFlowLoraLoaderMixin` menyediakan fungsi serupa untuk [AuraFlow](https://huggingface.co/fal/AuraFlow).
- `LTXVideoLoraLoaderMixin` menyediakan fungsi serupa untuk [LTX-Video](../pipelines/ltx_video).
- `SanaLoraLoaderMixin` menyediakan fungsi serupa untuk [Sana](../pipelines/sana).
- `HeliosLoraLoaderMixin` menyediakan fungsi serupa untuk [HunyuanVideo](../pipelines/helios).
- `HunyuanVideoLoraLoaderMixin` menyediakan fungsi serupa untuk [HunyuanVideo](../pipelines/hunyuan_video).
- `Lumina2LoraLoaderMixin` menyediakan fungsi serupa untuk [Lumina2](../pipelines/lumina2).
- `WanLoraLoaderMixin` menyediakan fungsi serupa untuk [Wan](../pipelines/wan).
- `SkyReelsV2LoraLoaderMixin` menyediakan fungsi serupa untuk [SkyReels-V2](../pipelines/skyreels_v2).
- `CogView4LoraLoaderMixin` menyediakan fungsi serupa untuk [CogView4](../pipelines/cogview4).
- `AmusedLoraLoaderMixin` adalah untuk `AmusedPipeline`.
- `AnimaLoraLoaderMixin` menyediakan fungsi serupa untuk [Anima](../pipelines/anima).
- `AceStepLoraLoaderMixin` menyediakan fungsi serupa untuk [ACE-Step](../pipelines/ace_step).
- `HiDreamImageLoraLoaderMixin` menyediakan fungsi serupa untuk [HiDream Image](../pipelines/hidream)
- `QwenImageLoraLoaderMixin` menyediakan fungsi serupa untuk [Qwen Image](../pipelines/qwenimage).
- `ZImageLoraLoaderMixin` menyediakan fungsi serupa untuk [Z-Image](../pipelines/z_image).
- `Flux2LoraLoaderMixin` menyediakan fungsi serupa untuk [Flux2](../pipelines/flux2).
- `ErnieImageLoraLoaderMixin` menyediakan fungsi serupa untuk [Ernie-Image](../pipelines/ernie_image).
- `LTX2LoraLoaderMixin` menyediakan fungsi serupa untuk [Flux2](../pipelines/ltx2).
- `MiniMaxH3LoraLoaderMixin` menyediakan fungsi serupa untuk [MiniMax-H3](../pipelines/minimax_h3).
- `PikaH3FlexLoraLoaderMixin` menyediakan fungsi serupa untuk [Pika H3 Flex](../pipelines/pika_h3_flex).
- `RFD4ProteinaLoraLoaderMixin` menyediakan fungsi serupa untuk [RFD4-Proteína](../pipelines/rfd4_proteina).
- `BriaFibo2LoraLoaderMixin` menyediakan fungsi serupa untuk [fibo-2](../pipelines/bria_fibo2).
- `LoraBaseMixin` menyediakan kelas dasar dengan beberapa metode utilitas untuk menggabungkan, memisahkan, memuat ulang, LoRA, dan lainnya.
> [!TIP]
> Untuk mempelajari lebih lanjut tentang cara memuat bobot LoRA, lihat panduan pemuatan [LoRA](../../tutorials/using_peft_for_inference).
## RFD4ProteinaLoraLoaderMixin[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin]]
#### diffusers.loaders.RFD4ProteinaLoraLoaderMixin[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin]]
```python
diffusers.loaders.RFD4ProteinaLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7532)
Muat lapisan LoRA ke dalam [RFD4ProteinaTransformer](/docs/diffusers/main/id/api/models/rfd4_proteina_transformer#diffusers.RFD4ProteinaTransformer).
Autoencoder dipasangkan dengan checkpoint denoiser dan tidak diadaptasi, sehingga transformer adalah satu-satunya modul yang dapat dimuat.
#### fuse_lora[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7694)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7631)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7595)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7543)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7658)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7714)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## LoraBaseMixin[[diffusers.loaders.lora_base.LoraBaseMixin]]
#### diffusers.loaders.lora_base.LoraBaseMixin[[diffusers.loaders.lora_base.LoraBaseMixin]]
```python
diffusers.loaders.lora_base.LoraBaseMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L491)
Kelas utilitas untuk menangani LoRA.
#### delete_adapters[[diffusers.loaders.lora_base.LoraBaseMixin.delete_adapters]]
```python
delete_adapters(adapter_names: list[str] | str)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L873)
**Parameter:**
adapter_names (`list[str, str]`) : Nama-nama adapter yang akan dihapus.
Hapus lapisan LoRA dari adapter di pipeline.
Contoh:
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
"jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", adapter_names="cinematic"
)
pipeline.delete_adapters("cinematic")
```
#### disable_lora[[diffusers.loaders.lora_base.LoraBaseMixin.disable_lora]]
```python
disable_lora()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L813)
Menonaktifkan lapisan LoRA aktif dari pipeline.
Contoh:
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
"jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", adapter_name="cinematic"
)
pipeline.disable_lora()
```
#### enable_lora[[diffusers.loaders.lora_base.LoraBaseMixin.enable_lora]]
```python
enable_lora()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L843)
Mengaktifkan lapisan LoRA aktif dari pipeline.
Contoh:
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
"jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", adapter_name="cinematic"
)
pipeline.enable_lora()
```
#### enable_lora_hotswap[[diffusers.loaders.lora_base.LoraBaseMixin.enable_lora_hotswap]]
```python
enable_lora_hotswap(**kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L1020)
**Parameter:**
target_rank (`int`) : Pangkat tertinggi di antara semua adapter yang akan dimuat.
check_compiled (`str`, *opsional*, defaultnya `"error"`) : Cara menangani model yang sudah dikompilasi. Pengecekan dapat mengembalikan pesan berikut: - "error" (default): lempar error - "warn": terbitkan peringatan - "ignore": tidak melakukan apa pun
Hotswap adapter tanpa memicu kompilasi ulang model atau jika rank dari adapter yang dimuat berbeda.
#### fuse_lora[[diffusers.loaders.lora_base.LoraBaseMixin.fuse_lora]]
```python
fuse_lora(components: list[str] | None = None, lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L549)
**Parameter:**
components : (`list[str]`): daftar komponen yang dapat diinjeksi LoRA untuk menggabungkan LoRA ke dalamnya.
lora_scale (`float`, defaultnya 1.0) : Mengontrol seberapa besar pengaruh parameter LoRA terhadap output.
safe_fusing (`bool`, defaultnya `False`) : Apakah akan memeriksa bobot yang sudah digabungkan (fused) untuk nilai NaN sebelum penggabungan, dan jika nilainya NaN, tidak melakukan penggabungan.
adapter_names (`list[str]`, *opsional*) : Nama-nama adapter yang akan digunakan untuk fusi. Jika tidak ada yang diteruskan, semua adapter aktif akan difusi.
Menggabungkan parameter LoRA ke dalam parameter asli dari blok yang sesuai.
Contoh:
```py
from diffusers import DiffusionPipeline
import torch
pipeline = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights("nerijs/pixel-art-xl", weight_name="pixel-art-xl.safetensors", adapter_name="pixel")
pipeline.fuse_lora(lora_scale=0.7)
```
#### get_active_adapters[[diffusers.loaders.lora_base.LoraBaseMixin.get_active_adapters]]
```python
get_active_adapters()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L911)
Mendapatkan daftar adapter aktif saat ini.
Contoh:
```python
from diffusers import DiffusionPipeline
pipeline = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0",
).to("cuda")
pipeline.load_lora_weights("CiroN2022/toy-face", weight_name="toy_face_sdxl.safetensors", adapter_name="toy")
pipeline.get_active_adapters()
```
#### get_list_adapters[[diffusers.loaders.lora_base.LoraBaseMixin.get_list_adapters]]
```python
get_list_adapters()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L944)
Mendapatkan daftar saat ini dari semua adapter yang tersedia di pipeline.
#### set_adapters[[diffusers.loaders.lora_base.LoraBaseMixin.set_adapters]]
```python
set_adapters(adapter_names: list[str] | str, adapter_weights: float | dict[str, float | dict | None] | list[float | dict[str, float | dict | None] | None] | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L698)
**Parameter:**
adapter_names (`list[str]` atau `str`) : Nama-nama adapter yang akan digunakan.
adapter_weights (`float`, `dict` atau `list`, *opsional*) : Bobot untuk adapter. Skalar menskalakan semua komponen yang kompatibel dengan LoRA. Kamus memetakan nama komponen (misalnya `"unet"`, `"transformer"` atau `"text_encoder"`) ke bobot; bobot komponen UNet juga dapat berupa kamus bersarang dari skala blok. Skalar atau kamus tunggal diterapkan ke setiap adapter. Daftar harus berisi satu skalar, kamus, atau `None` per adapter, sesuai urutan `adapter_names`. `None` secara default menjadi `1.0`, termasuk entri daftar dan bobot komponen. Komponen yang tidak ditentukan dan blok UNet juga secara default menjadi `1.0`.
Atur adapter yang saat ini aktif untuk digunakan dalam pipeline.
Contoh:
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
"jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", adapter_name="cinematic"
)
pipeline.load_lora_weights(
"troveski/sdxl-lora", weight_name="pytorch_lora_weights.safetensors", adapter_name="illustration"
)
pipeline.set_adapters(["cinematic", "illustration"], adapter_weights=[0.5, 0.5])
pipeline.set_adapters(
["cinematic", "illustration"],
adapter_weights=[{"unet": 0.5}, {"unet": 0.8, "text_encoder": 0.6}],
)
```
#### set_lora_device[[diffusers.loaders.lora_base.LoraBaseMixin.set_lora_device]]
```python
set_lora_device(adapter_names: list[str], device: torch.device | str | int)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L966)
**Parameter:**
adapter_names (`list[str]`) : daftar adapter untuk dikirim ke perangkat.
device (`torch.device | str | int`) : Perangkat untuk mengirim adapter. Bisa berupa perangkat torch, str, atau integer.
Memindahkan LoRA yang tercantum dalam `adapter_names` ke perangkat target. Berguna untuk offloading LoRA ke CPU jika Anda ingin memuat beberapa adapter dan membebaskan sebagian memori GPU.
Setelah meng-offload adapter LoRA ke CPU, selama bagian lain dari model masih berada di GPU, adapter LoRA tidak dapat lagi digunakan untuk inferensi, karena hal itu akan menyebabkan ketidakcocokan perangkat. Ingatlah untuk mengatur perangkat kembali ke GPU sebelum menggunakan adapter LoRA tersebut untuk inferensi.
```python
>>> pipe.load_lora_weights(path_1, adapter_name="adapter-1")
>>> pipe.load_lora_weights(path_2, adapter_name="adapter-2")
>>> pipe.set_adapters("adapter-1")
>>> image_1 = pipe(**kwargs)
>>> # switch to adapter-2, offload adapter-1
>>> pipeline.set_lora_device(adapter_names=["adapter-1"], device="cpu")
>>> pipeline.set_lora_device(adapter_names=["adapter-2"], device="cuda:0")
>>> pipe.set_adapters("adapter-2")
>>> image_2 = pipe(**kwargs)
>>> # switch back to adapter-1, offload adapter-2
>>> pipeline.set_lora_device(adapter_names=["adapter-2"], device="cpu")
>>> pipeline.set_lora_device(adapter_names=["adapter-1"], device="cuda:0")
>>> pipe.set_adapters("adapter-1")
>>> ...
```
#### unfuse_lora[[diffusers.loaders.lora_base.LoraBaseMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L636)
**Parameter:**
components (`list[str]`) : daftar komponen yang dapat diinjeksi LoRA untuk melepas fusi LoRA.
unfuse_unet (`bool`, defaultnya `True`) : Apakah akan memisahkan parameter LoRA dari UNet.
unfuse_text_encoder (`bool`, defaultnya `True`) : Apakah akan memisahkan (unfuse) parameter LoRA dari text encoder. Jika text encoder tidak di-patch secara monkey dengan parameter LoRA, maka tidak akan ada efeknya.
Membalikkan efek dari [`pipe.fuse_lora()`](https://huggingface.co/docs/diffusers/main/en/api/loaders#diffusers.loaders.LoraBaseMixin.fuse_lora).
#### unload_lora_weights[[diffusers.loaders.lora_base.LoraBaseMixin.unload_lora_weights]]
```python
unload_lora_weights()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L526)
Memuat ulang parameter LoRA.
Contoh:
```python
>>> # Assuming `pipeline` is already loaded with the LoRA parameters.
>>> pipeline.unload_lora_weights()
>>> ...
```
#### write_lora_layers[[diffusers.loaders.lora_base.LoraBaseMixin.write_lora_layers]]
```python
write_lora_layers(state_dict: dict[str, torch.Tensor], save_directory: str, is_main_process: bool, weight_name: str, save_function: Callable, safe_serialization: bool, lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L1043)
Menulis state dict dari lapisan LoRA (opsional dengan metadata) ke disk.
## StableDiffusionLoraLoaderMixin[[diffusers.loaders.StableDiffusionLoraLoaderMixin]]
#### diffusers.loaders.StableDiffusionLoraLoaderMixin[[diffusers.loaders.StableDiffusionLoraLoaderMixin]]
```python
diffusers.loaders.StableDiffusionLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L138)
Muat lapisan LoRA ke dalam [UNet2DConditionModel](/docs/diffusers/main/id/api/models/unet2d-cond#diffusers.UNet2DConditionModel) dan [`CLIPTextModel`](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel) Stable Diffusion.
#### load_lora_into_text_encoder[[diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_text_encoder]]
```python
load_lora_into_text_encoder(state_dict, network_alphas, text_encoder, prefix = None, lora_scale = 1.0, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L408)
**Parameter:**
state_dict (`dict`) : state dict standar yang berisi parameter lapisan lora. Kuncinya harus diberi awalan tambahan `text_encoder` untuk membedakan lapisan lora unet.
network_alphas (`dict[str, float]`) : Nilai network alpha yang digunakan untuk pembelajaran yang stabil dan mencegah underflow. Nilai ini memiliki arti yang sama dengan opsi `--network_alpha` dalam skrip trainer kohya-ss. Lihat [tautan ini](https://github.com/darkstorm2150/sd-scripts/blob/main/docs/train_network_README-en.md#execute-learning).
text_encoder (`CLIPTextModel`) : Model encoder teks untuk memuat lapisan LoRA.
prefix (`str`) : Prefiks yang diharapkan dari `text_encoder` di dalam `state_dict`.
lora_scale (`float`) : Berapa skala yang diterapkan pada output dari lora linear layer sebelum ditambahkan dengan output dari lora layer reguler.
adapter_name (`str`, *opsional*) : Nama adapter yang akan digunakan untuk merujuk model adapter yang dimuat. Jika tidak ditentukan, akan digunakan `default_{i}` di mana i adalah total jumlah adapter yang dimuat.
low_cpu_mem_usage (`bool`, *opsional*) : Mempercepat pemuatan model dengan hanya memuat bobot LoRA pretrained dan tidak menginisialisasi bobot acak.
hotswap (`bool`, *opsional*) : Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights).
metadata (`dict`) : Metadata adapter LoRA opsional. Jika diberikan, argumen `LoraConfig` dari `peft` tidak akan diturunkan dari state dict.
Ini akan memuat lapisan LoRA yang ditentukan dalam `state_dict` ke dalam `text_encoder`
#### load_lora_into_unet[[diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet]]
```python
load_lora_into_unet(state_dict, network_alphas, unet, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L354)
**Parameter:**
state_dict (`dict`) : state dict standar yang berisi parameter lapisan lora. Kunci-kuncinya dapat diindeks langsung ke unet atau diberi awalan tambahan `unet` yang dapat digunakan untuk membedakan lapisan lora text encoder.
network_alphas (`dict[str, float]`) : Nilai network alpha yang digunakan untuk pembelajaran yang stabil dan mencegah underflow. Nilai ini memiliki arti yang sama dengan opsi `--network_alpha` dalam skrip trainer kohya-ss. Lihat [tautan ini](https://github.com/darkstorm2150/sd-scripts/blob/main/docs/train_network_README-en.md#execute-learning).
unet (`UNet2DConditionModel`) : Model UNet untuk memuat lapisan LoRA.
adapter_name (`str`, *opsional*) : Nama adapter yang akan digunakan untuk merujuk model adapter yang dimuat. Jika tidak ditentukan, akan digunakan `default_{i}` di mana i adalah total jumlah adapter yang dimuat.
low_cpu_mem_usage (`bool`, *opsional*) : Mempercepat pemuatan model dengan hanya memuat bobot LoRA pretrained dan tidak menginisialisasi bobot acak.
hotswap (`bool`, *opsional*) : Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights).
metadata (`dict`) : Metadata adapter LoRA opsional. Jika diberikan, argumen `LoraConfig` dari `peft` tidak akan diturunkan dari state dict.
Ini akan memuat lapisan LoRA yang ditentukan dalam `state_dict` ke dalam `unet`.
#### load_lora_weights[[diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L148)
**Parameter:**
pretrained_model_name_or_path_or_dict (`str` atau `os.PathLike` atau `dict`) : Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict).
adapter_name (`str`, *opsional*) : Nama adapter yang akan digunakan untuk merujuk model adapter yang dimuat. Jika tidak ditentukan, akan digunakan `default_{i}` di mana i adalah total jumlah adapter yang dimuat.
low_cpu_mem_usage (`bool`, *opsional*) : Mempercepat pemuatan model dengan hanya memuat bobot LoRA pretrained dan tidak menginisialisasi bobot acak.
hotswap (`bool`, *opsional*) : Bawaan adalah `False`. Apakah akan mengganti adapter (LoRA) yang sudah ada dengan adapter yang baru dimuat secara in-place. Artinya, alih-alih memuat adapter tambahan, proses ini akan mengambil bobot adapter yang ada dan menggantinya dengan bobot dari adapter baru. Hal ini bisa lebih cepat dan lebih efisien dalam penggunaan memori. Namun, keunggulan utama dari hotswapping adalah ketika model dikompilasi dengan torch.compile, pemuatan adapter baru tidak memerlukan kompilasi ulang model. Saat menggunakan hotswapping, `adapter_name` yang diberikan harus berupa nama dari adapter yang sudah dimuat. Jika adapter baru dan adapter lama memiliki rank dan/atau LoRA alphas (yaitu scaling) yang berbeda, Anda perlu memanggil metode tambahan sebelum memuat adapter: ```py pipeline = ... # load diffusers pipeline max_rank = ... # the highest rank among all LoRAs that you want to load # call *before* compiling and loading the LoRA adapter pipeline.enable_lora_hotswap(target_rank=max_rank) pipeline.load_lora_weights(file_name) # optionally compile the model now ``` Perhatikan bahwa hotswapping adapter untuk text encoder belum didukung. Ada beberapa keterbatasan lain untuk teknik ini, yang didokumentasikan di sini: https://huggingface.co/docs/peft/main/en/package_reference/hotswap
kwargs (`dict`, *opsional*) : Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict).
Muat bobot LoRA yang ditentukan dalam `pretrained_model_name_or_path_or_dict` ke `self.unet` dan `self.text_encoder`.
Semua kwargs diteruskan ke `self.lora_state_dict`.
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut tentang cara state dict dimuat.
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut tentang bagaimana state dict dimuat ke dalam `self.unet`.
Lihat [load_lora_into_text_encoder()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_text_encoder) untuk detail lebih lanjut tentang bagaimana state dict dimuat ke dalam `self.text_encoder`.
#### lora_state_dict[[diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L243)
**Parameter:**
pretrained_model_name_or_path_or_dict (`str` atau `os.PathLike` atau `dict`) : Bisa berupa: - String, *model id* (misalnya `google/ddpm-celebahq-256`) dari model pretrained yang dihosting di Hub. - Path ke *directory* (misalnya `./my_model_directory`) yang berisi bobot model yang disimpan dengan [ModelMixin.save_pretrained()](/docs/diffusers/main/id/api/models/overview#diffusers.ModelMixin.save_pretrained). - [torch state dict](https://pytorch.org/tutorials/beginner/saving_loading_models.html#what-is-a-state-dict).
cache_dir (`str | os.PathLike`, *opsional*) : Path ke direktori tempat konfigurasi model pretrained yang diunduh di-cache jika cache standar tidak digunakan.
force_download (`bool`, *opsional*, defaultnya `False`) : Apakah akan memaksa (mengunduh ulang) bobot model dan file konfigurasi, menimpa versi yang di-cache jika ada.
proxies (`dict[str, str]`, *opsional*) : Kamus server proxy yang akan digunakan berdasarkan protokol atau endpoint, misalnya `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. Proxy digunakan pada setiap permintaan.
local_files_only (`bool`, *opsional*, defaultnya `False`) : Apakah hanya memuat bobot model dan file konfigurasi lokal atau tidak. Jika diatur ke `True`, model tidak akan diunduh dari Hub.
token (`str` atau *bool*, *opsional*) : Token yang digunakan sebagai otorisasi bearer HTTP untuk file jarak jauh. Jika `True`, token yang dihasilkan dari `diffusers-cli login` (disimpan di `~/.huggingface`) akan digunakan.
revision (`str`, *opsional*, defaultnya `"main"`) : Versi model spesifik yang akan digunakan. Bisa berupa nama branch, nama tag, id commit, atau identifikasi apa pun yang diizinkan oleh Git.
subfolder (`str`, *opsional*, defaultnya `""`) : Lokasi subfolder dari file model di dalam repositori model yang lebih besar di Hub atau secara lokal.
weight_name (`str`, *opsional*, defaultnya None) : Nama file state dict yang diserialisasi.
return_lora_metadata (`bool`, *opsional*, defaultnya False) : Jika diaktifkan, kembalikan juga metadata adapter LoRA, yang biasanya ditemukan dalam state dict.
Kembalikan state dict untuk bobot lora dan alfa jaringan.
> [!WARNING] > Kami mendukung pemuatan checkpoint LoRA berformat A1111 dengan kapasitas terbatas. > > Fungsi ini
eksperimental dan mungkin berubah di masa depan.
#### save_lora_weights[[diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, unet_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, text_encoder_lora_layers: dict[str, torch.nn.Module] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, unet_lora_adapter_metadata = None, text_encoder_lora_adapter_metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L466)
**Parameter:**
save_directory (`str` atau `os.PathLike`) : Direktori untuk menyimpan parameter LoRA. Akan dibuat jika belum ada.
unet_lora_layers (`dict[str, torch.nn.Module]` atau `dict[str, torch.Tensor]`) : State dict dari lapisan LoRA yang sesuai dengan `unet`.
text_encoder_lora_layers (`dict[str, torch.nn.Module]` atau `dict[str, torch.Tensor]`) : State dict dari lapisan LoRA yang sesuai dengan `text_encoder`. Anda harus secara eksplisit meneruskan state dict LoRA untuk text encoder karena berasal dari 🤗 Transformers.
is_main_process (`bool`, *opsional*, defaultnya `True`) : Apakah proses yang memanggil ini adalah proses utama atau tidak. Berguna selama pelatihan terdistribusi dan Anda perlu memanggil fungsi ini di semua proses. Dalam kasus ini, atur `is_main_process=True` hanya pada proses utama untuk menghindari kondisi balapan.
save_function (`Callable`) : Fungsi yang digunakan untuk menyimpan state dictionary. Berguna selama pelatihan terdistribusi ketika Anda perlu mengganti `torch.save` dengan metode lain. Dapat dikonfigurasi dengan variabel lingkungan `DIFFUSERS_SAVE_MODE`.
safe_serialization (`bool`, *opsional*, defaultnya `True`) : Apakah model disimpan menggunakan `safetensors` atau cara tradisional PyTorch dengan `pickle`.
unet_lora_adapter_metadata : Metadata adapter LoRA yang terkait dengan unet untuk diserialisasi bersama state dict.
text_encoder_lora_adapter_metadata : Metadata adapter LoRA yang terkait dengan text encoder untuk diserialisasi bersama state dict.
Simpan parameter LoRA yang sesuai dengan UNet dan text encoder.
## StableDiffusionXLLoraLoaderMixin[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin]]
#### diffusers.loaders.StableDiffusionXLLoraLoaderMixin[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin]]
```python
diffusers.loaders.StableDiffusionXLLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L585)
Muat lapisan LoRA ke dalam [UNet2DConditionModel](/docs/diffusers/main/id/api/models/unet2d-cond#diffusers.UNet2DConditionModel), [`CLIPTextModel`](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel), dan [`CLIPTextModelWithProjection`](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModelWithProjection) Stable Diffusion XL.
#### fuse_lora[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['unet', 'text_encoder', 'text_encoder_2'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L938)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_text_encoder[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin.load_lora_into_text_encoder]]
```python
load_lora_into_text_encoder(state_dict, network_alphas, text_encoder, prefix = None, lora_scale = 1.0, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L831)
**Parameter:**
state_dict (`dict`) : state dict standar yang berisi parameter lapisan lora. Kuncinya harus diberi awalan tambahan `text_encoder` untuk membedakan lapisan lora unet.
network_alphas (`dict[str, float]`) : Nilai network alpha yang digunakan untuk pembelajaran yang stabil dan mencegah underflow. Nilai ini memiliki arti yang sama dengan opsi `--network_alpha` dalam skrip trainer kohya-ss. Lihat [tautan ini](https://github.com/darkstorm2150/sd-scripts/blob/main/docs/train_network_README-en.md#execute-learning).
text_encoder (`CLIPTextModel`) : Model encoder teks untuk memuat lapisan LoRA.
prefix (`str`) : Prefiks yang diharapkan dari `text_encoder` di dalam `state_dict`.
lora_scale (`float`) : Berapa skala yang diterapkan pada output dari lora linear layer sebelum ditambahkan dengan output dari lora layer reguler.
adapter_name (`str`, *opsional*) : Nama adapter yang akan digunakan untuk merujuk model adapter yang dimuat. Jika tidak ditentukan, akan digunakan `default_{i}` di mana i adalah total jumlah adapter yang dimuat.
low_cpu_mem_usage (`bool`, *opsional*) : Mempercepat pemuatan model dengan hanya memuat bobot LoRA pretrained dan tidak menginisialisasi bobot acak.
hotswap (`bool`, *opsional*) : Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights).
metadata (`dict`) : Metadata adapter LoRA opsional. Jika diberikan, argumen `LoraConfig` dari `peft` tidak akan diturunkan dari state dict.
Ini akan memuat lapisan LoRA yang ditentukan dalam `state_dict` ke dalam `text_encoder`
#### load_lora_into_unet[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin.load_lora_into_unet]]
```python
load_lora_into_unet(state_dict, network_alphas, unet, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L776)
**Parameter:**
state_dict (`dict`) : state dict standar yang berisi parameter lapisan lora. Kunci-kuncinya dapat diindeks langsung ke unet atau diberi awalan tambahan `unet` yang dapat digunakan untuk membedakan lapisan lora text encoder.
network_alphas (`dict[str, float]`) : Nilai network alpha yang digunakan untuk pembelajaran yang stabil dan mencegah underflow. Nilai ini memiliki arti yang sama dengan opsi `--network_alpha` dalam skrip trainer kohya-ss. Lihat [tautan ini](https://github.com/darkstorm2150/sd-scripts/blob/main/docs/train_network_README-en.md#execute-learning).
unet (`UNet2DConditionModel`) : Model UNet untuk memuat lapisan LoRA.
adapter_name (`str`, *opsional*) : Nama adapter yang akan digunakan untuk merujuk model adapter yang dimuat. Jika tidak ditentukan, akan digunakan `default_{i}` di mana i adalah total jumlah adapter yang dimuat.
low_cpu_mem_usage (`bool`, *opsional*) : Mempercepat pemuatan model dengan hanya memuat bobot LoRA pretrained dan tidak menginisialisasi bobot acak.
hotswap (`bool`, *opsional*) : Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights).
metadata (`dict`) : Metadata adapter LoRA opsional. Jika diberikan, argumen `LoraConfig` dari `peft` tidak akan diturunkan dari state dict.
Ini akan memuat lapisan LoRA yang ditentukan dalam `state_dict` ke dalam `unet`.
#### load_lora_weights[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L596)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L664)
**Parameter:**
pretrained_model_name_or_path_or_dict (`str` atau `os.PathLike` atau `dict`) : Bisa berupa: - String, *model id* (misalnya `google/ddpm-celebahq-256`) dari model pretrained yang dihosting di Hub. - Path ke *directory* (misalnya `./my_model_directory`) yang berisi bobot model yang disimpan dengan [ModelMixin.save_pretrained()](/docs/diffusers/main/id/api/models/overview#diffusers.ModelMixin.save_pretrained). - [torch state dict](https://pytorch.org/tutorials/beginner/saving_loading_models.html#what-is-a-state-dict).
cache_dir (`str | os.PathLike`, *opsional*) : Path ke direktori tempat konfigurasi model pretrained yang diunduh di-cache jika cache standar tidak digunakan.
force_download (`bool`, *opsional*, defaultnya `False`) : Apakah akan memaksa (mengunduh ulang) bobot model dan file konfigurasi, menimpa versi yang di-cache jika ada.
proxies (`dict[str, str]`, *opsional*) : Kamus server proxy yang akan digunakan berdasarkan protokol atau endpoint, misalnya `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. Proxy digunakan pada setiap permintaan.
local_files_only (`bool`, *opsional*, defaultnya `False`) : Apakah hanya memuat bobot model dan file konfigurasi lokal atau tidak. Jika diatur ke `True`, model tidak akan diunduh dari Hub.
token (`str` atau *bool*, *opsional*) : Token yang digunakan sebagai otorisasi bearer HTTP untuk file jarak jauh. Jika `True`, token yang dihasilkan dari `diffusers-cli login` (disimpan di `~/.huggingface`) akan digunakan.
revision (`str`, *opsional*, defaultnya `"main"`) : Versi model spesifik yang akan digunakan. Bisa berupa nama branch, nama tag, id commit, atau identifikasi apa pun yang diizinkan oleh Git.
subfolder (`str`, *opsional*, defaultnya `""`) : Lokasi subfolder dari file model di dalam repositori model yang lebih besar di Hub atau secara lokal.
weight_name (`str`, *opsional*, defaultnya None) : Nama file state dict yang diserialisasi.
return_lora_metadata (`bool`, *opsional*, defaultnya False) : Jika diaktifkan, kembalikan juga metadata adapter LoRA, yang biasanya ditemukan dalam state dict.
Kembalikan state dict untuk bobot lora dan alfa jaringan.
> [!WARNING] > Kami mendukung pemuatan checkpoint LoRA berformat A1111 dengan kapasitas terbatas. > > Fungsi ini
eksperimental dan mungkin berubah di masa depan.
#### save_lora_weights[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, unet_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, text_encoder_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, text_encoder_2_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, unet_lora_adapter_metadata = None, text_encoder_lora_adapter_metadata = None, text_encoder_2_lora_adapter_metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L890)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.StableDiffusionXLLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['unet', 'text_encoder', 'text_encoder_2'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L957)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## SD3LoraLoaderMixin[[diffusers.loaders.SD3LoraLoaderMixin]]
#### diffusers.loaders.SD3LoraLoaderMixin[[diffusers.loaders.SD3LoraLoaderMixin]]
```python
diffusers.loaders.SD3LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L964)
Muat lapisan LoRA ke dalam [SD3Transformer2DModel](/docs/diffusers/main/id/api/models/sd3_transformer2d#diffusers.SD3Transformer2DModel), [`CLIPTextModel`](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel), dan [`CLIPTextModelWithProjection`](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModelWithProjection).
Khusus untuk [StableDiffusion3Pipeline](/docs/diffusers/main/id/api/pipelines/stable_diffusion/stable_diffusion_3#diffusers.StableDiffusion3Pipeline).
#### fuse_lora[[diffusers.loaders.SD3LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer', 'text_encoder', 'text_encoder_2'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1225)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_text_encoder[[diffusers.loaders.SD3LoraLoaderMixin.load_lora_into_text_encoder]]
```python
load_lora_into_text_encoder(state_dict, network_alphas, text_encoder, prefix = None, lora_scale = 1.0, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1116)
**Parameter:**
state_dict (`dict`) : state dict standar yang berisi parameter lapisan lora. Kuncinya harus diberi awalan tambahan `text_encoder` untuk membedakan lapisan lora unet.
network_alphas (`dict[str, float]`) : Nilai network alpha yang digunakan untuk pembelajaran yang stabil dan mencegah underflow. Nilai ini memiliki arti yang sama dengan opsi `--network_alpha` dalam skrip trainer kohya-ss. Lihat [tautan ini](https://github.com/darkstorm2150/sd-scripts/blob/main/docs/train_network_README-en.md#execute-learning).
text_encoder (`CLIPTextModel`) : Model encoder teks untuk memuat lapisan LoRA.
prefix (`str`) : Prefiks yang diharapkan dari `text_encoder` di dalam `state_dict`.
lora_scale (`float`) : Berapa skala yang diterapkan pada output dari lora linear layer sebelum ditambahkan dengan output dari lora layer reguler.
adapter_name (`str`, *opsional*) : Nama adapter yang akan digunakan untuk merujuk model adapter yang dimuat. Jika tidak ditentukan, akan digunakan `default_{i}` di mana i adalah total jumlah adapter yang dimuat.
low_cpu_mem_usage (`bool`, *opsional*) : Mempercepat pemuatan model dengan hanya memuat bobot LoRA pretrained dan tidak menginisialisasi bobot acak.
hotswap (`bool`, *opsional*) : Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights).
metadata (`dict`) : Metadata adapter LoRA opsional. Jika diberikan, argumen `LoraConfig` dari `peft` tidak akan diturunkan dari state dict.
Ini akan memuat lapisan LoRA yang ditentukan dalam `state_dict` ke dalam `text_encoder`
#### load_lora_into_transformer[[diffusers.loaders.SD3LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1090)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.SD3LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1031)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.SD3LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L977)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.SD3LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, text_encoder_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, text_encoder_2_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata = None, text_encoder_lora_adapter_metadata = None, text_encoder_2_lora_adapter_metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1175)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.SD3LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer', 'text_encoder', 'text_encoder_2'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1245)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## FluxLoraLoaderMixin[[diffusers.loaders.FluxLoraLoaderMixin]]
#### diffusers.loaders.FluxLoraLoaderMixin[[diffusers.loaders.FluxLoraLoaderMixin]]
```python
diffusers.loaders.FluxLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1441)
Muat lapisan LoRA ke dalam [FluxTransformer2DModel](/docs/diffusers/main/id/api/models/flux_transformer#diffusers.FluxTransformer2DModel), [`CLIPTextModel`](https://huggingface.co/docs/transformers/model_doc/clip#transformers.CLIPTextModel).
Khusus untuk [FluxPipeline](/docs/diffusers/main/id/api/pipelines/flux#diffusers.FluxPipeline).
#### fuse_lora[[diffusers.loaders.FluxLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1879)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### load_lora_into_text_encoder[[diffusers.loaders.FluxLoraLoaderMixin.load_lora_into_text_encoder]]
```python
load_lora_into_text_encoder(state_dict, network_alphas, text_encoder, prefix = None, lora_scale = 1.0, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1756)
**Parameter:**
state_dict (`dict`) : state dict standar yang berisi parameter lapisan lora. Kuncinya harus diberi awalan tambahan `text_encoder` untuk membedakan lapisan lora unet.
network_alphas (`dict[str, float]`) : Nilai network alpha yang digunakan untuk pembelajaran yang stabil dan mencegah underflow. Nilai ini memiliki arti yang sama dengan opsi `--network_alpha` dalam skrip trainer kohya-ss. Lihat [tautan ini](https://github.com/darkstorm2150/sd-scripts/blob/main/docs/train_network_README-en.md#execute-learning).
text_encoder (`CLIPTextModel`) : Model encoder teks untuk memuat lapisan LoRA.
prefix (`str`) : Prefiks yang diharapkan dari `text_encoder` di dalam `state_dict`.
lora_scale (`float`) : Berapa skala yang diterapkan pada output dari lora linear layer sebelum ditambahkan dengan output dari lora layer reguler.
adapter_name (`str`, *opsional*) : Nama adapter yang akan digunakan untuk merujuk model adapter yang dimuat. Jika tidak ditentukan, akan digunakan `default_{i}` di mana i adalah total jumlah adapter yang dimuat.
low_cpu_mem_usage (`bool`, *opsional*) : Mempercepat pemuatan model dengan hanya memuat bobot LoRA pretrained dan tidak menginisialisasi bobot acak.
hotswap (`bool`, *opsional*) : Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights).
metadata (`dict`) : Metadata adapter LoRA opsional. Jika diberikan, argumen `LoraConfig` dari `peft` tidak akan diturunkan dari state dict.
Ini akan memuat lapisan LoRA yang ditentukan dalam `state_dict` ke dalam `text_encoder`
#### load_lora_into_transformer[[diffusers.loaders.FluxLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, network_alphas, transformer, adapter_name = None, metadata = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1675)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.FluxLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1579)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.FluxLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], return_alphas: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1454)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.FluxLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, text_encoder_lora_layers: dict[str, torch.nn.Module] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata = None, text_encoder_lora_adapter_metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1815)
**Parameter:**
save_directory (`str` atau `os.PathLike`) : Direktori untuk menyimpan parameter LoRA. Akan dibuat jika belum ada.
transformer_lora_layers (`dict[str, torch.nn.Module]` atau `dict[str, torch.Tensor]`) : State dict dari lapisan LoRA yang sesuai dengan `transformer`.
text_encoder_lora_layers (`dict[str, torch.nn.Module]` atau `dict[str, torch.Tensor]`) : State dict dari lapisan LoRA yang sesuai dengan `text_encoder`. Anda harus secara eksplisit meneruskan state dict LoRA untuk text encoder karena berasal dari 🤗 Transformers.
is_main_process (`bool`, *opsional*, defaultnya `True`) : Apakah proses yang memanggil ini adalah proses utama atau tidak. Berguna selama pelatihan terdistribusi dan Anda perlu memanggil fungsi ini di semua proses. Dalam kasus ini, atur `is_main_process=True` hanya pada proses utama untuk menghindari kondisi balapan.
save_function (`Callable`) : Fungsi yang digunakan untuk menyimpan state dictionary. Berguna selama pelatihan terdistribusi ketika Anda perlu mengganti `torch.save` dengan metode lain. Dapat dikonfigurasi dengan variabel lingkungan `DIFFUSERS_SAVE_MODE`.
safe_serialization (`bool`, *opsional*, defaultnya `True`) : Apakah model disimpan menggunakan `safetensors` atau cara tradisional PyTorch dengan `pickle`.
transformer_lora_adapter_metadata : Metadata adapter LoRA yang terkait dengan transformer untuk diserialisasi bersama state dict.
text_encoder_lora_adapter_metadata : Metadata adapter LoRA yang terkait dengan text encoder untuk diserialisasi bersama state dict.
Simpan parameter LoRA yang sesuai dengan UNet dan text encoder.
#### unfuse_lora[[diffusers.loaders.FluxLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer', 'text_encoder'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1911)
**Parameter:**
components (`list[str]`) : daftar komponen yang dapat diinjeksi LoRA untuk melepas fusi LoRA.
Membalikkan efek dari [`pipe.fuse_lora()`](https://huggingface.co/docs/diffusers/main/en/api/loaders#diffusers.loaders.LoraBaseMixin.fuse_lora).
#### unload_lora_weights[[diffusers.loaders.FluxLoraLoaderMixin.unload_lora_weights]]
```python
unload_lora_weights(reset_to_overwritten_params = False)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1926)
**Parameter:**
reset_to_overwritten_params (`bool`, defaultnya `False`) : Apakah modul yang dimuat LoRA akan direset ke parameter aslinya. Lihat [dokumentasi Flux](https://huggingface.co/docs/diffusers/main/en/api/pipelines/flux) untuk mempelajari lebih lanjut.
Memuat ulang parameter LoRA.
Contoh:
```python
>>> # Assuming `pipeline` is already loaded with the LoRA parameters.
>>> pipeline.unload_lora_weights()
>>> ...
```
## Flux2LoraLoaderMixin[[diffusers.loaders.Flux2LoraLoaderMixin]]
#### diffusers.loaders.Flux2LoraLoaderMixin[[diffusers.loaders.Flux2LoraLoaderMixin]]
```python
diffusers.loaders.Flux2LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5773)
Muat lapisan LoRA ke dalam [Flux2Transformer2DModel](/docs/diffusers/main/id/api/models/flux2_transformer#diffusers.Flux2Transformer2DModel). Khusus untuk [Flux2Pipeline](/docs/diffusers/main/id/api/pipelines/flux2#diffusers.Flux2Pipeline).
#### fuse_lora[[diffusers.loaders.Flux2LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5957)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.Flux2LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5894)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.Flux2LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5859)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.Flux2LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5781)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.Flux2LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5921)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.Flux2LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5977)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## Flux3LoraLoaderMixin[[diffusers.loaders.Flux3LoraLoaderMixin]]
#### diffusers.loaders.Flux3LoraLoaderMixin[[diffusers.loaders.Flux3LoraLoaderMixin]]
```python
diffusers.loaders.Flux3LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5984)
#### fuse_lora[[diffusers.loaders.Flux3LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6139)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.Flux3LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6076)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.Flux3LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6040)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.Flux3LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6103)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.Flux3LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6159)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## BriaFibo2LoraLoaderMixin[[diffusers.loaders.BriaFibo2LoraLoaderMixin]]
#### diffusers.loaders.BriaFibo2LoraLoaderMixin[[diffusers.loaders.BriaFibo2LoraLoaderMixin]]
```python
diffusers.loaders.BriaFibo2LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6166)
Muat lapisan LoRA ke dalam [BriaFibo2Transformer2DModel](/docs/diffusers/main/id/api/models/transformer_bria_fibo2#diffusers.BriaFibo2Transformer2DModel). Khusus untuk [BriaFibo2Pipeline](/docs/diffusers/main/id/api/pipelines/bria_fibo2#diffusers.BriaFibo2Pipeline).
#### fuse_lora[[diffusers.loaders.BriaFibo2LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6328)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.BriaFibo2LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6265)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.BriaFibo2LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6229)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.BriaFibo2LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6174)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.BriaFibo2LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6292)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.BriaFibo2LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6348)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## ErnieImageLoraLoaderMixin[[diffusers.loaders.ErnieImageLoraLoaderMixin]]
#### diffusers.loaders.ErnieImageLoraLoaderMixin[[diffusers.loaders.ErnieImageLoraLoaderMixin]]
```python
diffusers.loaders.ErnieImageLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6551)
Muat lapisan LoRA ke dalam [ErnieImageTransformer2DModel](/docs/diffusers/main/id/api/models/ernie_image_transformer2d#diffusers.ErnieImageTransformer2DModel). Khusus untuk `ErnieImagePipeline`.
#### fuse_lora[[diffusers.loaders.ErnieImageLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6724)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.ErnieImageLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6661)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.ErnieImageLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6625)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.ErnieImageLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6559)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.ErnieImageLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6688)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.ErnieImageLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6744)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## LTX2LoraLoaderMixin[[diffusers.loaders.LTX2LoraLoaderMixin]]
#### diffusers.loaders.LTX2LoraLoaderMixin[[diffusers.loaders.LTX2LoraLoaderMixin]]
```python
diffusers.loaders.LTX2LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2919)
Muat lapisan LoRA ke dalam [LTX2VideoTransformer3DModel](/docs/diffusers/main/id/api/models/ltx2_video_transformer3d#diffusers.LTX2VideoTransformer3DModel). Khusus untuk [LTX2Pipeline](/docs/diffusers/main/id/api/pipelines/ltx2#diffusers.LTX2Pipeline).
#### fuse_lora[[diffusers.loaders.LTX2LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3108)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.LTX2LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None, prefix: str = 'transformer')
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3044)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.LTX2LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2992)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.LTX2LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2928)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.LTX2LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3072)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.LTX2LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3128)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## CogVideoXLoraLoaderMixin[[diffusers.loaders.CogVideoXLoraLoaderMixin]]
#### diffusers.loaders.CogVideoXLoraLoaderMixin[[diffusers.loaders.CogVideoXLoraLoaderMixin]]
```python
diffusers.loaders.CogVideoXLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2353)
Muat lapisan LoRA ke dalam [CogVideoXTransformer3DModel](/docs/diffusers/main/id/api/models/cogvideox_transformer3d#diffusers.CogVideoXTransformer3DModel). Khusus untuk [CogVideoXPipeline](/docs/diffusers/main/id/api/pipelines/cogvideox#diffusers.CogVideoXPipeline).
#### fuse_lora[[diffusers.loaders.CogVideoXLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2512)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.CogVideoXLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2451)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.CogVideoXLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2416)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.CogVideoXLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2361)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.CogVideoXLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2478)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.CogVideoXLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2531)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## Mochi1LoraLoaderMixin[[diffusers.loaders.Mochi1LoraLoaderMixin]]
#### diffusers.loaders.Mochi1LoraLoaderMixin[[diffusers.loaders.Mochi1LoraLoaderMixin]]
```python
diffusers.loaders.Mochi1LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2538)
Muat lapisan LoRA ke dalam [MochiTransformer3DModel](/docs/diffusers/main/id/api/models/mochi_transformer3d#diffusers.MochiTransformer3DModel). Khusus untuk [MochiPipeline](/docs/diffusers/main/id/api/pipelines/mochi#diffusers.MochiPipeline).
#### fuse_lora[[diffusers.loaders.Mochi1LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2700)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.Mochi1LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2637)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.Mochi1LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2601)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.Mochi1LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2546)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.Mochi1LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2664)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.Mochi1LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2720)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## AuraFlowLoraLoaderMixin[[diffusers.loaders.AuraFlowLoraLoaderMixin]]
#### diffusers.loaders.AuraFlowLoraLoaderMixin[[diffusers.loaders.AuraFlowLoraLoaderMixin]]
```python
diffusers.loaders.AuraFlowLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1252)
Muat lapisan LoRA ke dalam [AuraFlowTransformer2DModel](/docs/diffusers/main/id/api/models/aura_flow_transformer2d#diffusers.AuraFlowTransformer2DModel) Khusus untuk [AuraFlowPipeline](/docs/diffusers/main/id/api/pipelines/aura_flow#diffusers.AuraFlowPipeline).
#### fuse_lora[[diffusers.loaders.AuraFlowLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1414)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.AuraFlowLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1351)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.AuraFlowLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1315)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.AuraFlowLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1260)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.AuraFlowLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1378)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.AuraFlowLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer', 'text_encoder'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L1434)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## LTXVideoLoraLoaderMixin[[diffusers.loaders.LTXVideoLoraLoaderMixin]]
#### diffusers.loaders.LTXVideoLoraLoaderMixin[[diffusers.loaders.LTXVideoLoraLoaderMixin]]
```python
diffusers.loaders.LTXVideoLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2727)
Muat lapisan LoRA ke dalam [LTXVideoTransformer3DModel](/docs/diffusers/main/id/api/models/ltx_video_transformer3d#diffusers.LTXVideoTransformer3DModel). Khusus untuk [LTXPipeline](/docs/diffusers/main/id/api/pipelines/ltx_video#diffusers.LTXPipeline).
#### fuse_lora[[diffusers.loaders.LTXVideoLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2892)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.LTXVideoLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2829)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.LTXVideoLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2793)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.LTXVideoLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2735)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.LTXVideoLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2856)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.LTXVideoLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2912)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## SanaLoraLoaderMixin[[diffusers.loaders.SanaLoraLoaderMixin]]
#### diffusers.loaders.SanaLoraLoaderMixin[[diffusers.loaders.SanaLoraLoaderMixin]]
```python
diffusers.loaders.SanaLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3135)
Muat lapisan LoRA ke dalam [SanaTransformer2DModel](/docs/diffusers/main/id/api/models/sana_transformer2d#diffusers.SanaTransformer2DModel). Khusus untuk [SanaPipeline](/docs/diffusers/main/id/api/pipelines/sana#diffusers.SanaPipeline).
#### fuse_lora[[diffusers.loaders.SanaLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3297)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.SanaLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3234)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.SanaLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3198)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.SanaLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3143)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.SanaLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3261)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.SanaLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3317)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## HeliosLoraLoaderMixin[[diffusers.loaders.HeliosLoraLoaderMixin]]
#### diffusers.loaders.HeliosLoraLoaderMixin[[diffusers.loaders.HeliosLoraLoaderMixin]]
```python
diffusers.loaders.HeliosLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3324)
Muat lapisan LoRA ke dalam [HeliosTransformer3DModel](/docs/diffusers/main/id/api/models/helios_transformer3d#diffusers.HeliosTransformer3DModel). Khusus untuk [HeliosPipeline](/docs/diffusers/main/id/api/pipelines/helios#diffusers.HeliosPipeline) dan [HeliosPyramidPipeline](/docs/diffusers/main/id/api/pipelines/helios#diffusers.HeliosPyramidPipeline).
#### fuse_lora[[diffusers.loaders.HeliosLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3487)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.HeliosLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3424)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.HeliosLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3390)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.HeliosLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3332)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.HeliosLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3451)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.HeliosLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3507)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## HunyuanVideoLoraLoaderMixin[[diffusers.loaders.HunyuanVideoLoraLoaderMixin]]
#### diffusers.loaders.HunyuanVideoLoraLoaderMixin[[diffusers.loaders.HunyuanVideoLoraLoaderMixin]]
```python
diffusers.loaders.HunyuanVideoLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3514)
Muat lapisan LoRA ke dalam [HunyuanVideoTransformer3DModel](/docs/diffusers/main/id/api/models/hunyuan_video_transformer_3d#diffusers.HunyuanVideoTransformer3DModel). Khusus untuk [HunyuanVideoPipeline](/docs/diffusers/main/id/api/pipelines/hunyuan_video#diffusers.HunyuanVideoPipeline).
#### fuse_lora[[diffusers.loaders.HunyuanVideoLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3679)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.HunyuanVideoLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3616)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.HunyuanVideoLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3580)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.HunyuanVideoLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3522)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.HunyuanVideoLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3643)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.HunyuanVideoLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3699)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## Lumina2LoraLoaderMixin[[diffusers.loaders.Lumina2LoraLoaderMixin]]
#### diffusers.loaders.Lumina2LoraLoaderMixin[[diffusers.loaders.Lumina2LoraLoaderMixin]]
```python
diffusers.loaders.Lumina2LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3706)
Muat lapisan LoRA ke dalam [Lumina2Transformer2DModel](/docs/diffusers/main/id/api/models/lumina2_transformer2d#diffusers.Lumina2Transformer2DModel). Khusus untuk [Lumina2Pipeline](/docs/diffusers/main/id/api/pipelines/lumina2#diffusers.Lumina2Pipeline).
#### fuse_lora[[diffusers.loaders.Lumina2LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3872)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.Lumina2LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3809)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.Lumina2LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3773)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.Lumina2LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3714)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.Lumina2LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3836)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.Lumina2LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3892)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## CogView4LoraLoaderMixin[[diffusers.loaders.CogView4LoraLoaderMixin]]
#### diffusers.loaders.CogView4LoraLoaderMixin[[diffusers.loaders.CogView4LoraLoaderMixin]]
```python
diffusers.loaders.CogView4LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4617)
Muat lapisan LoRA ke dalam [WanTransformer3DModel](/docs/diffusers/main/id/api/models/wan_transformer_3d#diffusers.WanTransformer3DModel). Khusus untuk [CogView4Pipeline](/docs/diffusers/main/id/api/pipelines/cogview4#diffusers.CogView4Pipeline).
#### fuse_lora[[diffusers.loaders.CogView4LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4779)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.CogView4LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4716)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.CogView4LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4680)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.CogView4LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4625)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.CogView4LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4743)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.CogView4LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4799)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## WanLoraLoaderMixin[[diffusers.loaders.WanLoraLoaderMixin]]
#### diffusers.loaders.WanLoraLoaderMixin[[diffusers.loaders.WanLoraLoaderMixin]]
```python
diffusers.loaders.WanLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4088)
Muat lapisan LoRA ke dalam [WanTransformer3DModel](/docs/diffusers/main/id/api/models/wan_transformer_3d#diffusers.WanTransformer3DModel). Khusus untuk [WanPipeline](/docs/diffusers/main/id/api/pipelines/wan#diffusers.WanPipeline) dan `[WanImageToVideoPipeline`].
#### fuse_lora[[diffusers.loaders.WanLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4324)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.WanLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4261)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.WanLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4202)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.WanLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4096)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.WanLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4288)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.WanLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4344)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## SkyReelsV2LoraLoaderMixin[[diffusers.loaders.SkyReelsV2LoraLoaderMixin]]
#### diffusers.loaders.SkyReelsV2LoraLoaderMixin[[diffusers.loaders.SkyReelsV2LoraLoaderMixin]]
```python
diffusers.loaders.SkyReelsV2LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4351)
Muat lapisan LoRA ke dalam [SkyReelsV2Transformer3DModel](/docs/diffusers/main/id/api/models/skyreels_v2_transformer_3d#diffusers.SkyReelsV2Transformer3DModel).
#### fuse_lora[[diffusers.loaders.SkyReelsV2LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4590)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.SkyReelsV2LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4527)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.SkyReelsV2LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4467)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.SkyReelsV2LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4359)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.SkyReelsV2LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4554)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.SkyReelsV2LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4610)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## AmusedLoraLoaderMixin[[diffusers.loaders.AmusedLoraLoaderMixin]]
#### diffusers.loaders.AmusedLoraLoaderMixin[[diffusers.loaders.AmusedLoraLoaderMixin]]
```python
diffusers.loaders.AmusedLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2206)
#### load_lora_into_transformer[[diffusers.loaders.AmusedLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, network_alphas, transformer, adapter_name = None, metadata = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2211)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.AmusedLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, text_encoder_lora_layers: dict[str, torch.nn.Module] = None, transformer_lora_layers: dict[str, torch.nn.Module] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L2298)
**Parameter:**
save_directory (`str` atau `os.PathLike`) : Direktori untuk menyimpan parameter LoRA. Akan dibuat jika belum ada.
unet_lora_layers (`dict[str, torch.nn.Module]` atau `dict[str, torch.Tensor]`) : State dict dari lapisan LoRA yang sesuai dengan `unet`.
text_encoder_lora_layers (`dict[str, torch.nn.Module]` atau `dict[str, torch.Tensor]`) : State dict dari lapisan LoRA yang sesuai dengan `text_encoder`. Anda harus secara eksplisit meneruskan state dict LoRA untuk text encoder karena berasal dari 🤗 Transformers.
is_main_process (`bool`, *opsional*, defaultnya `True`) : Apakah proses yang memanggil ini adalah proses utama atau tidak. Berguna selama pelatihan terdistribusi dan Anda perlu memanggil fungsi ini di semua proses. Dalam kasus ini, atur `is_main_process=True` hanya pada proses utama untuk menghindari kondisi balapan.
save_function (`Callable`) : Fungsi yang digunakan untuk menyimpan state dictionary. Berguna selama pelatihan terdistribusi ketika Anda perlu mengganti `torch.save` dengan metode lain. Dapat dikonfigurasi dengan variabel lingkungan `DIFFUSERS_SAVE_MODE`.
safe_serialization (`bool`, *opsional*, defaultnya `True`) : Apakah model disimpan menggunakan `safetensors` atau cara tradisional PyTorch dengan `pickle`.
Simpan parameter LoRA yang sesuai dengan UNet dan text encoder.
## AnimaLoraLoaderMixin[[diffusers.loaders.AnimaLoraLoaderMixin]]
#### diffusers.loaders.AnimaLoraLoaderMixin[[diffusers.loaders.AnimaLoraLoaderMixin]]
```python
diffusers.loaders.AnimaLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5587)
Muat lapisan LoRA ke dalam [CosmosTransformer3DModel](/docs/diffusers/main/id/api/models/cosmos_transformer3d#diffusers.CosmosTransformer3DModel) dan [AnimaTextConditioner](/docs/diffusers/main/id/api/pipelines/anima#diffusers.AnimaTextConditioner).
#### fuse_lora[[diffusers.loaders.AnimaLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer', 'text_conditioner'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5747)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.AnimaLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5652)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.AnimaLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5596)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### unfuse_lora[[diffusers.loaders.AnimaLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer', 'text_conditioner'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5766)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## AceStepLoraLoaderMixin[[diffusers.loaders.AceStepLoraLoaderMixin]]
#### diffusers.loaders.AceStepLoraLoaderMixin[[diffusers.loaders.AceStepLoraLoaderMixin]]
```python
diffusers.loaders.AceStepLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6941)
Muat lapisan LoRA ke dalam [AceStepTransformer1DModel](/docs/diffusers/main/id/api/models/ace_step_transformer#diffusers.AceStepTransformer1DModel). Khusus untuk [AceStepPipeline](/docs/diffusers/main/id/api/pipelines/ace_step#diffusers.AceStepPipeline).
#### fuse_lora[[diffusers.loaders.AceStepLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7105)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.AceStepLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7042)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.AceStepLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7006)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.AceStepLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6949)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.AceStepLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7069)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.AceStepLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7125)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## HiDreamImageLoraLoaderMixin[[diffusers.loaders.HiDreamImageLoraLoaderMixin]]
#### diffusers.loaders.HiDreamImageLoraLoaderMixin[[diffusers.loaders.HiDreamImageLoraLoaderMixin]]
```python
diffusers.loaders.HiDreamImageLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4806)
Muat lapisan LoRA ke dalam [HiDreamImageTransformer2DModel](/docs/diffusers/main/id/api/models/hidream_image_transformer#diffusers.HiDreamImageTransformer2DModel). Khusus untuk [HiDreamImagePipeline](/docs/diffusers/main/id/api/pipelines/hidream#diffusers.HiDreamImagePipeline).
#### fuse_lora[[diffusers.loaders.HiDreamImageLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4971)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.HiDreamImageLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4908)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.HiDreamImageLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4872)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.HiDreamImageLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4814)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.HiDreamImageLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4935)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.HiDreamImageLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4991)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## QwenImageLoraLoaderMixin[[diffusers.loaders.QwenImageLoraLoaderMixin]]
#### diffusers.loaders.QwenImageLoraLoaderMixin[[diffusers.loaders.QwenImageLoraLoaderMixin]]
```python
diffusers.loaders.QwenImageLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4998)
Muat lapisan LoRA ke dalam [QwenImageTransformer2DModel](/docs/diffusers/main/id/api/models/qwenimage_transformer2d#diffusers.QwenImageTransformer2DModel). Khusus untuk [QwenImagePipeline](/docs/diffusers/main/id/api/pipelines/qwenimage#diffusers.QwenImagePipeline).
#### fuse_lora[[diffusers.loaders.QwenImageLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5166)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.QwenImageLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5103)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.QwenImageLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5067)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.QwenImageLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5006)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.QwenImageLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5130)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.QwenImageLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5186)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## ZImageLoraLoaderMixin[[diffusers.loaders.ZImageLoraLoaderMixin]]
#### diffusers.loaders.ZImageLoraLoaderMixin[[diffusers.loaders.ZImageLoraLoaderMixin]]
```python
diffusers.loaders.ZImageLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5387)
Muat lapisan LoRA ke dalam [ZImageTransformer2DModel](/docs/diffusers/main/id/api/models/z_image_transformer2d#diffusers.ZImageTransformer2DModel). Khusus untuk [ZImagePipeline](/docs/diffusers/main/id/api/pipelines/z_image#diffusers.ZImagePipeline).
#### fuse_lora[[diffusers.loaders.ZImageLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5560)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.ZImageLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5497)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.ZImageLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5461)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.ZImageLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5395)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.ZImageLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5524)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.ZImageLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5580)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## CosmosLoraLoaderMixin[[diffusers.loaders.CosmosLoraLoaderMixin]]
#### diffusers.loaders.CosmosLoraLoaderMixin[[diffusers.loaders.CosmosLoraLoaderMixin]]
```python
diffusers.loaders.CosmosLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6751)
Muat lapisan LoRA ke dalam [CosmosTransformer3DModel](/docs/diffusers/main/id/api/models/cosmos_transformer3d#diffusers.CosmosTransformer3DModel), Khusus untuk [Cosmos2_5_PredictBasePipeline](/docs/diffusers/main/id/api/pipelines/cosmos#diffusers.Cosmos2_5_PredictBasePipeline).
#### fuse_lora[[diffusers.loaders.CosmosLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6914)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.CosmosLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6851)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.CosmosLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6815)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.CosmosLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6760)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.CosmosLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6878)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.CosmosLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6934)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## KandinskyLoraLoaderMixin[[diffusers.loaders.KandinskyLoraLoaderMixin]]
#### diffusers.loaders.KandinskyLoraLoaderMixin[[diffusers.loaders.KandinskyLoraLoaderMixin]]
```python
diffusers.loaders.KandinskyLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3899)
Muat lapisan LoRA ke dalam `Kandinsky5Transformer3DModel`,
#### fuse_lora[[diffusers.loaders.KandinskyLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4061)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.KandinskyLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3998)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.KandinskyLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3962)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.KandinskyLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L3907)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.KandinskyLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4025)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.KandinskyLoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L4081)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## Ideogram4LoraLoaderMixin[[diffusers.loaders.Ideogram4LoraLoaderMixin]]
#### diffusers.loaders.Ideogram4LoraLoaderMixin[[diffusers.loaders.Ideogram4LoraLoaderMixin]]
```python
diffusers.loaders.Ideogram4LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6355)
Muat lapisan LoRA ke dalam [Ideogram4Transformer2DModel](/docs/diffusers/main/id/api/models/ideogram4_transformer2d#diffusers.Ideogram4Transformer2DModel). Khusus untuk [Ideogram4Pipeline](/docs/diffusers/main/id/api/pipelines/ideogram4#diffusers.Ideogram4Pipeline).
#### fuse_lora[[diffusers.loaders.Ideogram4LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6524)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.Ideogram4LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6461)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.Ideogram4LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6425)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.Ideogram4LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6363)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.Ideogram4LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6488)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.Ideogram4LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L6544)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## Krea2LoraLoaderMixin[[diffusers.loaders.Krea2LoraLoaderMixin]]
#### diffusers.loaders.Krea2LoraLoaderMixin[[diffusers.loaders.Krea2LoraLoaderMixin]]
```python
diffusers.loaders.Krea2LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5193)
Muat lapisan LoRA ke dalam [Krea2Transformer2DModel](/docs/diffusers/main/id/api/models/krea2_transformer2d#diffusers.Krea2Transformer2DModel). Khusus untuk [Krea2Pipeline](/docs/diffusers/main/id/api/pipelines/krea2#diffusers.Krea2Pipeline).
#### fuse_lora[[diffusers.loaders.Krea2LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5360)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.Krea2LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5297)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.Krea2LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5261)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.Krea2LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5201)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.Krea2LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5324)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.Krea2LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L5380)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## MiniMaxH3LoraLoaderMixin[[diffusers.loaders.MiniMaxH3LoraLoaderMixin]]
#### diffusers.loaders.MiniMaxH3LoraLoaderMixin[[diffusers.loaders.MiniMaxH3LoraLoaderMixin]]
```python
diffusers.loaders.MiniMaxH3LoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7132)
Muat lapisan LoRA ke dalam [MiniMaxH3Transformer3DModel](/docs/diffusers/main/id/api/models/minimax_h3_transformer3d#diffusers.MiniMaxH3Transformer3DModel). Khusus untuk [MiniMaxH3ModularPipeline](/docs/diffusers/main/id/api/pipelines/minimax_h3#diffusers.MiniMaxH3ModularPipeline).
MiniMax-H3 menyediakan dua partisi DiT independen dengan nama modul yang identik (LoRA untuk satu partisi akan dimuat ke partisi lainnya dan menurunkan kualitas output), sehingga routing dilakukan secara eksplisit: state dict yang dikonversi menargetkan `transformer.`; akses ke `transformer_ref` dilakukan dengan file berawalan `transformer_ref.` atau `load_into_transformer_ref=True`.
LoRA dari DiffSynth-Studio (misalnya [DiffSynth-Studio/MiniMax-H3-LoRA-LineartAnime](https://huggingface.co/DiffSynth-Studio/MiniMax-H3-LoRA-LineartAnime)) dilatih terhadap QKV fused per-head-interleaved dari checkpoint mentah dan di-de-interleave saat konversi. Faktor fp32-nya membuat jalur LoRA yang tidak ter-fuse menghitung dalam fp32; `.to(torch.bfloat16)` pada model setelah dimuat memulihkan anggaran memori bf16.
LoRA yang dilatih terhadap checkpoint yang dipangkas (file `*_pruned_*` di [Comfy-Org/MiniMax-H3](https://huggingface.co/Comfy-Org/MiniMax-H3); [joyfox/MiniMax-H3-Turbo](https://huggingface.co/joyfox/MiniMax-H3-Turbo) adalah salah satunya) gagal dengan kesalahan ukuran yang tidak cocok. File tanpa `alpha` dimuat pada `alpha == rank` (skala 1.0, konvensi yang dinyatakan oleh misalnya [larryvrh/MiniMax-H3-Turbo-Lora](https://huggingface.co/larryvrh/MiniMax-H3-Turbo-Lora)); entri `alpha` dalam `__metadata__` (misalnya file 8-langkah dari [lightx2v/Minimax-h3-Turbo](https://huggingface.co/lightx2v/Minimax-h3-Turbo)), jika ada, akan dihormati sebagai gantinya.
#### fuse_lora[[diffusers.loaders.MiniMaxH3LoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer', 'transformer_ref'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7489)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.MiniMaxH3LoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7389)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_into_transformer_ref[[diffusers.loaders.MiniMaxH3LoraLoaderMixin.load_lora_into_transformer_ref]]
```python
load_lora_into_transformer_ref(state_dict, transformer_ref, prefix, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7416)
Muat lapisan LoRA ke partisi `transformer_ref`. `prefix` adalah nama komponen yang dibawa oleh kunci, yaitu `transformer_ref` untuk file yang menamai partisi dan `transformer` untuk file yang diarahkan ke sini oleh `load_into_transformer_ref=True`. Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.MiniMaxH3LoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, load_into_transformer_ref: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7274)
**Parameter:**
load_into_transformer_ref (`bool`, defaultnya `False`) : Muat lapisan berawalan `transformer.` ke partisi `transformer_ref` — yang digunakan oleh alur kerja `ref2va` untuk denoising — alih-alih `transformer`. Hanya diperlukan ketika kedua partisi dimuat: pipeline yang hanya memegang `transformer_ref` akan mengarah ke sana secara otomatis.
Muat lapisan LoRA ke dalam `transformer` atau, dengan `load_into_transformer_ref=True`, ke dalam `transformer_ref`. Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.MiniMaxH3LoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7159)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.MiniMaxH3LoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, transformer_ref_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None, transformer_ref_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7446)
Simpan lapisan LoRA dari satu atau kedua partisi MiniMax-H3. Partisi mana yang dimiliki oleh sebuah LoRA tidak dapat dipulihkan dari kuncinya, sehingga ini adalah satu-satunya cara untuk menerbitkan LoRA H3 yang mencatatnya. Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
#### unfuse_lora[[diffusers.loaders.MiniMaxH3LoraLoaderMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] = ['transformer', 'transformer_ref'], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7508)
Lihat `unfuse_lora()` untuk detail lebih lanjut.
## PikaH3FlexLoraLoaderMixin[[diffusers.loaders.PikaH3FlexLoraLoaderMixin]]
#### diffusers.loaders.PikaH3FlexLoraLoaderMixin[[diffusers.loaders.PikaH3FlexLoraLoaderMixin]]
```python
diffusers.loaders.PikaH3FlexLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7515)
Muat lapisan LoRA ke dalam [MiniMaxH3Transformer3DModel](/docs/diffusers/main/id/api/models/minimax_h3_transformer3d#diffusers.MiniMaxH3Transformer3DModel). Khusus untuk [PikaH3FlexModularPipeline](/docs/diffusers/main/id/api/pipelines/pika_h3_flex#diffusers.PikaH3FlexModularPipeline).
Rilis Pika H3 Flex *adalah* sepasang adapter rank-128 di atas dua partisi transformer MiniMax-H3, diterbitkan sebagai dump PEFT polos dari nama modul diffusers (`fl2va/adapter.safetensors` untuk `transformer`, `ref2va/adapter.safetensors` untuk `transformer_ref`). File tanpa alpha dimuat pada `alpha == rank` (skala 1.0), kontrak bawaan adapter, dan proses pemuatan, routing, serta penyimpanan mengikuti `MiniMaxH3LoraLoaderMixin`: adapter `fl2va` masuk ke `transformer`, dan adapter `ref2va` ke `transformer_ref` — secara mandiri ketika pipeline hanya memegang partisi tersebut (`workflow="ref2va"`), atau dengan `load_into_transformer_ref=True` ketika pipeline memegang keduanya.
```py
pipe.load_lora_weights("Pika-Labs/Pika-H3-Flex", weight_name="fl2va/adapter.safetensors")
```
## LoraBaseMixin[[diffusers.loaders.lora_base.LoraBaseMixin]]
#### diffusers.loaders.lora_base.LoraBaseMixin[[diffusers.loaders.lora_base.LoraBaseMixin]]
```python
diffusers.loaders.lora_base.LoraBaseMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L491)
Kelas utilitas untuk menangani LoRA.
#### delete_adapters[[diffusers.loaders.lora_base.LoraBaseMixin.delete_adapters]]
```python
delete_adapters(adapter_names: list[str] | str)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L873)
**Parameter:**
adapter_names (`list[str, str]`) : Nama-nama adapter yang akan dihapus.
Hapus lapisan LoRA dari adapter di pipeline.
Contoh:
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
"jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", adapter_names="cinematic"
)
pipeline.delete_adapters("cinematic")
```
#### disable_lora[[diffusers.loaders.lora_base.LoraBaseMixin.disable_lora]]
```python
disable_lora()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L813)
Menonaktifkan lapisan LoRA aktif dari pipeline.
Contoh:
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
"jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", adapter_name="cinematic"
)
pipeline.disable_lora()
```
#### enable_lora[[diffusers.loaders.lora_base.LoraBaseMixin.enable_lora]]
```python
enable_lora()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L843)
Mengaktifkan lapisan LoRA aktif dari pipeline.
Contoh:
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
"jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", adapter_name="cinematic"
)
pipeline.enable_lora()
```
#### enable_lora_hotswap[[diffusers.loaders.lora_base.LoraBaseMixin.enable_lora_hotswap]]
```python
enable_lora_hotswap(**kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L1020)
**Parameter:**
target_rank (`int`) : Pangkat tertinggi di antara semua adapter yang akan dimuat.
check_compiled (`str`, *opsional*, defaultnya `"error"`) : Cara menangani model yang sudah dikompilasi. Pengecekan dapat mengembalikan pesan berikut: - "error" (default): lempar error - "warn": terbitkan peringatan - "ignore": tidak melakukan apa pun
Hotswap adapter tanpa memicu kompilasi ulang model atau jika rank dari adapter yang dimuat berbeda.
#### fuse_lora[[diffusers.loaders.lora_base.LoraBaseMixin.fuse_lora]]
```python
fuse_lora(components: list[str] | None = None, lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L549)
**Parameter:**
components : (`list[str]`): daftar komponen yang dapat diinjeksi LoRA untuk menggabungkan LoRA ke dalamnya.
lora_scale (`float`, defaultnya 1.0) : Mengontrol seberapa besar pengaruh parameter LoRA terhadap output.
safe_fusing (`bool`, defaultnya `False`) : Apakah akan memeriksa bobot yang sudah digabungkan (fused) untuk nilai NaN sebelum penggabungan, dan jika nilainya NaN, tidak melakukan penggabungan.
adapter_names (`list[str]`, *opsional*) : Nama-nama adapter yang akan digunakan untuk fusi. Jika tidak ada yang diteruskan, semua adapter aktif akan difusi.
Menggabungkan parameter LoRA ke dalam parameter asli dari blok yang sesuai.
Contoh:
```py
from diffusers import DiffusionPipeline
import torch
pipeline = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights("nerijs/pixel-art-xl", weight_name="pixel-art-xl.safetensors", adapter_name="pixel")
pipeline.fuse_lora(lora_scale=0.7)
```
#### get_active_adapters[[diffusers.loaders.lora_base.LoraBaseMixin.get_active_adapters]]
```python
get_active_adapters()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L911)
Mendapatkan daftar adapter aktif saat ini.
Contoh:
```python
from diffusers import DiffusionPipeline
pipeline = DiffusionPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0",
).to("cuda")
pipeline.load_lora_weights("CiroN2022/toy-face", weight_name="toy_face_sdxl.safetensors", adapter_name="toy")
pipeline.get_active_adapters()
```
#### get_list_adapters[[diffusers.loaders.lora_base.LoraBaseMixin.get_list_adapters]]
```python
get_list_adapters()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L944)
Mendapatkan daftar saat ini dari semua adapter yang tersedia di pipeline.
#### set_adapters[[diffusers.loaders.lora_base.LoraBaseMixin.set_adapters]]
```python
set_adapters(adapter_names: list[str] | str, adapter_weights: float | dict[str, float | dict | None] | list[float | dict[str, float | dict | None] | None] | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L698)
**Parameter:**
adapter_names (`list[str]` atau `str`) : Nama-nama adapter yang akan digunakan.
adapter_weights (`float`, `dict` atau `list`, *opsional*) : Bobot untuk adapter. Skalar menskalakan semua komponen yang kompatibel dengan LoRA. Kamus memetakan nama komponen (misalnya `"unet"`, `"transformer"` atau `"text_encoder"`) ke bobot; bobot komponen UNet juga dapat berupa kamus bersarang dari skala blok. Skalar atau kamus tunggal diterapkan ke setiap adapter. Daftar harus berisi satu skalar, kamus, atau `None` per adapter, sesuai urutan `adapter_names`. `None` secara default menjadi `1.0`, termasuk entri daftar dan bobot komponen. Komponen yang tidak ditentukan dan blok UNet juga secara default menjadi `1.0`.
Atur adapter yang saat ini aktif untuk digunakan dalam pipeline.
Contoh:
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
"jbilcke-hf/sdxl-cinematic-1", weight_name="pytorch_lora_weights.safetensors", adapter_name="cinematic"
)
pipeline.load_lora_weights(
"troveski/sdxl-lora", weight_name="pytorch_lora_weights.safetensors", adapter_name="illustration"
)
pipeline.set_adapters(["cinematic", "illustration"], adapter_weights=[0.5, 0.5])
pipeline.set_adapters(
["cinematic", "illustration"],
adapter_weights=[{"unet": 0.5}, {"unet": 0.8, "text_encoder": 0.6}],
)
```
#### set_lora_device[[diffusers.loaders.lora_base.LoraBaseMixin.set_lora_device]]
```python
set_lora_device(adapter_names: list[str], device: torch.device | str | int)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L966)
**Parameter:**
adapter_names (`list[str]`) : daftar adapter untuk dikirim ke perangkat.
device (`torch.device | str | int`) : Perangkat untuk mengirim adapter. Bisa berupa perangkat torch, str, atau integer.
Memindahkan LoRA yang tercantum dalam `adapter_names` ke perangkat target. Berguna untuk offloading LoRA ke CPU jika Anda ingin memuat beberapa adapter dan membebaskan sebagian memori GPU.
Setelah meng-offload adapter LoRA ke CPU, selama bagian lain dari model masih berada di GPU, adapter LoRA tidak dapat lagi digunakan untuk inferensi, karena hal itu akan menyebabkan ketidakcocokan perangkat. Ingatlah untuk mengatur perangkat kembali ke GPU sebelum menggunakan adapter LoRA tersebut untuk inferensi.
```python
>>> pipe.load_lora_weights(path_1, adapter_name="adapter-1")
>>> pipe.load_lora_weights(path_2, adapter_name="adapter-2")
>>> pipe.set_adapters("adapter-1")
>>> image_1 = pipe(**kwargs)
>>> # switch to adapter-2, offload adapter-1
>>> pipeline.set_lora_device(adapter_names=["adapter-1"], device="cpu")
>>> pipeline.set_lora_device(adapter_names=["adapter-2"], device="cuda:0")
>>> pipe.set_adapters("adapter-2")
>>> image_2 = pipe(**kwargs)
>>> # switch back to adapter-1, offload adapter-2
>>> pipeline.set_lora_device(adapter_names=["adapter-2"], device="cpu")
>>> pipeline.set_lora_device(adapter_names=["adapter-1"], device="cuda:0")
>>> pipe.set_adapters("adapter-1")
>>> ...
```
#### unfuse_lora[[diffusers.loaders.lora_base.LoraBaseMixin.unfuse_lora]]
```python
unfuse_lora(components: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L636)
**Parameter:**
components (`list[str]`) : daftar komponen yang dapat diinjeksi LoRA untuk melepas fusi LoRA.
unfuse_unet (`bool`, defaultnya `True`) : Apakah akan memisahkan parameter LoRA dari UNet.
unfuse_text_encoder (`bool`, defaultnya `True`) : Apakah akan memisahkan (unfuse) parameter LoRA dari text encoder. Jika text encoder tidak di-patch secara monkey dengan parameter LoRA, maka tidak akan ada efeknya.
Membalikkan efek dari [`pipe.fuse_lora()`](https://huggingface.co/docs/diffusers/main/en/api/loaders#diffusers.loaders.LoraBaseMixin.fuse_lora).
#### unload_lora_weights[[diffusers.loaders.lora_base.LoraBaseMixin.unload_lora_weights]]
```python
unload_lora_weights()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L526)
Memuat ulang parameter LoRA.
Contoh:
```python
>>> # Assuming `pipeline` is already loaded with the LoRA parameters.
>>> pipeline.unload_lora_weights()
>>> ...
```
#### write_lora_layers[[diffusers.loaders.lora_base.LoraBaseMixin.write_lora_layers]]
```python
write_lora_layers(state_dict: dict[str, torch.Tensor], save_directory: str, is_main_process: bool, weight_name: str, save_function: Callable, safe_serialization: bool, lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_base.py#L1043)
Menulis state dict dari lapisan LoRA (opsional dengan metadata) ke disk.
# LoRA
LoRA adalah metode pelatihan yang cepat dan ringan, yang menyisipkan dan melatih jumlah parameter yang jauh lebih sedikit dibandingkan seluruh parameter model. Hal ini menghasilkan file yang lebih kecil (~100 MB) dan memudahkan pelatihan model dengan cepat untuk mempelajari konsep baru. Bobot LoRA biasanya dimuat ke dalam denoiser, text encoder, atau keduanya. Denoiser biasanya merujuk pada UNet (misalnya [UNet2DConditionModel](/docs/diffusers/main/id/api/models/unet2d-cond#diffusers.UNet2DConditionModel)) atau Transformer (misalnya [SD3Transformer2DModel](/docs/diffusers/main/id/api/models/sd3_transformer2d#diffusers.SD3Transformer2DModel)). Terdapat beberapa kelas untuk memuat bobot LoRA:
- `StableDiffusionLoraLoaderMixin` menyediakan fungsi untuk memuat dan melepas, menggabungkan dan memisahkan, mengaktifkan dan menonaktifkan, serta fungsi lain untuk mengelola bobot LoRA. Kelas ini dapat digunakan dengan model apa pun.
- `StableDiffusionXLLoraLoaderMixin` adalah versi [Stable Diffusion (SDXL)](../../api/pipelines/stable_diffusion/stable_diffusion_xl) dari kelas `StableDiffusionLoraLoaderMixin` untuk memuat dan menyimpan bobot LoRA. Kelas ini hanya dapat digunakan dengan model SDXL.
- `SD3LoraLoaderMixin` menyediakan fungsi serupa untuk [Stable Diffusion 3](https://huggingface.co/blog/sd3).
- `FluxLoraLoaderMixin` menyediakan fungsi serupa untuk [Flux](../pipelines/flux).
- `CogVideoXLoraLoaderMixin` menyediakan fungsi serupa untuk [CogVideoX](../pipelines/cogvideox).
- `Mochi1LoraLoaderMixin` menyediakan fungsi serupa untuk [Mochi](../pipelines/mochi).
- `AuraFlowLoraLoaderMixin` menyediakan fungsi serupa untuk [AuraFlow](https://huggingface.co/fal/AuraFlow).
- `LTXVideoLoraLoaderMixin` menyediakan fungsi serupa untuk [LTX-Video](../pipelines/ltx_video).
- `SanaLoraLoaderMixin` menyediakan fungsi serupa untuk [Sana](../pipelines/sana).
- `HeliosLoraLoaderMixin` menyediakan fungsi serupa untuk [HunyuanVideo](../pipelines/helios).
- `HunyuanVideoLoraLoaderMixin` menyediakan fungsi serupa untuk [HunyuanVideo](../pipelines/hunyuan_video).
- `Lumina2LoraLoaderMixin` menyediakan fungsi serupa untuk [Lumina2](../pipelines/lumina2).
- `WanLoraLoaderMixin` menyediakan fungsi serupa untuk [Wan](../pipelines/wan).
- `SkyReelsV2LoraLoaderMixin` menyediakan fungsi serupa untuk [SkyReels-V2](../pipelines/skyreels_v2).
- `CogView4LoraLoaderMixin` menyediakan fungsi serupa untuk [CogView4](../pipelines/cogview4).
- `AmusedLoraLoaderMixin` adalah untuk `AmusedPipeline`.
- `AnimaLoraLoaderMixin` menyediakan fungsi serupa untuk [Anima](../pipelines/anima).
- `AceStepLoraLoaderMixin` menyediakan fungsi serupa untuk [ACE-Step](../pipelines/ace_step).
- `HiDreamImageLoraLoaderMixin` menyediakan fungsi serupa untuk [HiDream Image](../pipelines/hidream)
- `QwenImageLoraLoaderMixin` menyediakan fungsi serupa untuk [Qwen Image](../pipelines/qwenimage).
- `ZImageLoraLoaderMixin` menyediakan fungsi serupa untuk [Z-Image](../pipelines/z_image).
- `Flux2LoraLoaderMixin` menyediakan fungsi serupa untuk [Flux2](../pipelines/flux2).
- `ErnieImageLoraLoaderMixin` menyediakan fungsi serupa untuk [Ernie-Image](../pipelines/ernie_image).
- `LTX2LoraLoaderMixin` menyediakan fungsi serupa untuk [Flux2](../pipelines/ltx2).
- `MiniMaxH3LoraLoaderMixin` menyediakan fungsi serupa untuk [MiniMax-H3](../pipelines/minimax_h3).
- `PikaH3FlexLoraLoaderMixin` menyediakan fungsi serupa untuk [Pika H3 Flex](../pipelines/pika_h3_flex).
- `RFD4ProteinaLoraLoaderMixin` menyediakan fungsi serupa untuk [RFD4-Proteína](../pipelines/rfd4_proteina).
- `BriaFibo2LoraLoaderMixin` menyediakan fungsi serupa untuk [fibo-2](../pipelines/bria_fibo2).
- `LoraBaseMixin` menyediakan kelas dasar dengan beberapa metode utilitas untuk menggabungkan, memisahkan, memuat ulang, LoRA, dan lainnya.
> [!TIP]
> Untuk mempelajari lebih lanjut tentang cara memuat bobot LoRA, lihat panduan pemuatan [LoRA](../../tutorials/using_peft_for_inference).
## RFD4ProteinaLoraLoaderMixin[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin]]
#### diffusers.loaders.RFD4ProteinaLoraLoaderMixin[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin]]
```python
diffusers.loaders.RFD4ProteinaLoraLoaderMixin()
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7532)
Muat lapisan LoRA ke dalam [RFD4ProteinaTransformer](/docs/diffusers/main/id/api/models/rfd4_proteina_transformer#diffusers.RFD4ProteinaTransformer).
Autoencoder dipasangkan dengan checkpoint denoiser dan tidak diadaptasi, sehingga transformer adalah satu-satunya modul yang dapat dimuat.
#### fuse_lora[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.fuse_lora]]
```python
fuse_lora(components: list[str] = ['transformer'], lora_scale: float = 1.0, safe_fusing: bool = False, adapter_names: list[str] | None = None, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7694)
Lihat `fuse_lora()` untuk detail lebih lanjut.
#### load_lora_into_transformer[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.load_lora_into_transformer]]
```python
load_lora_into_transformer(state_dict, transformer, adapter_name = None, _pipeline = None, low_cpu_mem_usage = False, hotswap: bool = False, metadata = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7631)
Lihat [load_lora_into_unet()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_into_unet) untuk detail lebih lanjut.
#### load_lora_weights[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.load_lora_weights]]
```python
load_lora_weights(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], adapter_name: str | None = None, hotswap: bool = False, **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7595)
Lihat [load_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.load_lora_weights) untuk detail lebih lanjut.
#### lora_state_dict[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.lora_state_dict]]
```python
lora_state_dict(pretrained_model_name_or_path_or_dict: str | dict[str, torch.Tensor], **kwargs)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7543)
Lihat [lora_state_dict()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.lora_state_dict) untuk detail lebih lanjut.
#### save_lora_weights[[diffusers.loaders.RFD4ProteinaLoraLoaderMixin.save_lora_weights]]
```python
save_lora_weights(save_directory: str | os.PathLike, transformer_lora_layers: dict[str, typing.Union[torch.nn.Module, torch.Tensor]] = None, is_main_process: bool = True, weight_name: str = None, save_function: typing.Callable = None, safe_serialization: bool = True, transformer_lora_adapter_metadata: dict | None = None)
```
[Sumber](https://github.com/huggingface/diffusers/blob/main/src/diffusers/loaders/lora_pipeline.py#L7658)
Lihat [save_lora_weights()](/docs/diffusers/main/id/api/loaders/lora#diffusers.loaders.StableDiffusionLoraLoaderMixin.save_lora_weights) untuk informasi lebih lanjut.
Xet Storage Details
- Size:
- 162 kB
- Xet hash:
- 01a3a1e1b1e49c82cf6de81b7d313bd90aaf1afebe7fd60a3f7a23b32bdc6662
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.