Buckets:
| # AutoPipeline | |
| [AutoPipeline](../api/pipelines/auto_pipeline) adalah pipeline *task-and-model* yang secara otomatis memilih subclass pipeline yang benar berdasarkan tugasnya. Komponen ini menangani kompleksitas memuat subclass pipeline yang berbeda tanpa perlu mengetahui nama subclass pipeline yang spesifik. | |
| Hal ini berbeda dengan [DiffusionPipeline](/docs/diffusers/v0.43.0/id/api/pipelines/overview#diffusers.DiffusionPipeline), sebuah pipeline *hanya-model* yang secara otomatis memilih subclass pipeline berdasarkan model. | |
| ```text | |
| AutoPipelineForImage2Image.from_pretrained(model_id) | |
| | | |
| +-- read model_index.json (e.g. StableDiffusionXLPipeline) | |
| +-- task mapping (image-to-image) | |
| | | |
| v | |
| StableDiffusionXLImg2ImgPipeline // returned instance | |
| ``` | |
| [AutoPipelineForImage2Image](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForImage2Image) mengembalikan subclass yang spesifik untuk tugas (misalnya, [StableDiffusionXLImg2ImgPipeline](/docs/diffusers/v0.43.0/id/api/pipelines/stable_diffusion/stable_diffusion_xl#diffusers.StableDiffusionXLImg2ImgPipeline)), yang hanya dapat digunakan untuk tugas image-to-image. | |
| ```py | |
| import torch | |
| from diffusers import AutoPipelineForImage2Image | |
| pipeline = AutoPipelineForImage2Image.from_pretrained( | |
| "RunDiffusion/Juggernaut-XL-v9", dtype=torch.bfloat16, device_map="cuda", # or "mps", "xpu", "cpu" | |
| ) | |
| print(pipeline) | |
| # StableDiffusionXLImg2ImgPipeline { | |
| # "_class_name": "StableDiffusionXLImg2ImgPipeline", | |
| # ... | |
| # } | |
| ``` | |
| Memuat model yang sama dengan [DiffusionPipeline](/docs/diffusers/v0.43.0/id/api/pipelines/overview#diffusers.DiffusionPipeline) mengembalikan subclass teks-ke-gambar default, yaitu [StableDiffusionXLPipeline](/docs/diffusers/v0.43.0/id/api/pipelines/stable_diffusion/stable_diffusion_xl#diffusers.StableDiffusionXLPipeline). Pipeline tersebut ditujukan untuk teks-ke-gambar. Untuk gambar-ke-gambar atau inpainting, muat AutoPipeline tugas seperti [AutoPipelineForImage2Image](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForImage2Image) atau [AutoPipelineForInpainting](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForInpainting), atau subclass spesifik tugas yang sesuai. | |
| ```py | |
| import torch | |
| from diffusers import DiffusionPipeline | |
| pipeline = DiffusionPipeline.from_pretrained( | |
| "RunDiffusion/Juggernaut-XL-v9", dtype=torch.bfloat16, device_map="cuda", # or "mps", "xpu", "cpu" | |
| ) | |
| print(pipeline) | |
| # StableDiffusionXLPipeline { | |
| # "_class_name": "StableDiffusionXLPipeline", | |
| # ... | |
| # } | |
| ``` | |
| ## Ganti tugas dengan from_pipe | |
| Muat tugas AutoPipeline sekali, lalu ganti tugas dengan [from_pipe()](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForImage2Image.from_pipe) tanpa mengunduh bobotnya lagi. Komponen digunakan ulang dari pipeline sumber. | |
| ```py | |
| import torch | |
| from diffusers import AutoPipelineForText2Image, AutoPipelineForImage2Image | |
| pipeline_t2i = AutoPipelineForText2Image.from_pretrained( | |
| "RunDiffusion/Juggernaut-XL-v9", dtype=torch.bfloat16, device_map="cuda", # or "mps", "xpu", "cpu" | |
| ) | |
| pipeline_i2i = AutoPipelineForImage2Image.from_pipe(pipeline_t2i) | |
| ``` | |
| Lihat [Memanfaatkan ulang model dalam beberapa pipeline](../using-diffusers/loading#reusing-models-in-multiple-pipelines) untuk detail lebih lanjut. | |
| Periksa [mappings](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/auto_pipeline.py) untuk melihat apakah suatu model didukung atau tidak. Mencoba memuat model yang tidak didukung akan mengembalikan error. | |
| ```py | |
| import torch | |
| from diffusers import AutoPipelineForImage2Image | |
| pipeline = AutoPipelineForImage2Image.from_pretrained( | |
| "openai/shap-e-img2img", dtype=torch.float16, | |
| ) | |
| "ValueError: AutoPipeline can't find a pipeline linked to ShapEImg2ImgPipeline for None" | |
| ``` | |
| Ada empat jenis kelas [AutoPipeline](../api/pipelines/auto_pipeline): | |
| - [AutoPipelineForText2Image](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForText2Image) | |
| - [AutoPipelineForImage2Image](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForImage2Image) | |
| - [AutoPipelineForInpainting](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForInpainting) | |
| - [AutoPipelineForText2Audio](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForText2Audio) | |
| Setiap kelas ini memiliki pemetaan yang telah ditentukan sebelumnya, yang menghubungkan pipeline ke subkelas spesifik tugas mereka. | |
| Ketika [from_pretrained()](/docs/diffusers/v0.43.0/id/api/pipelines/auto_pipeline#diffusers.AutoPipelineForText2Image.from_pretrained) dipanggil, ia mengekstrak nama kelas dari file `model_index.json` dan memilih subkelas pipeline yang sesuai untuk tugas tersebut berdasarkan pemetaan. | |
Xet Storage Details
- Size:
- 4.86 kB
- Xet hash:
- 30bb0614ab18db8d5060b0c906502da1d5ca7895f9799c3bb33a4cc707a2fb67
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.