diffusers/pr_13231/en/_app/immutable/nodes/89.e12ac171.js

import{s as ee,n as te,o as ne}from"../chunks/scheduler.53228c21.js";import{S as ae,i as re,e as l,s as r,c as v,h as se,a as m,d as n,b as s,f as q,g as b,j as G,k as W,l as C,m as a,n as T,t as D,o as M,p as E}from"../chunks/index.cac5d66a.js";import{C as ie}from"../chunks/CopyLLMTxtMenu.54ac3dec.js";import{D as oe}from"../chunks/Docstring.5a04f810.js";import{C as le}from"../chunks/CodeBlock.606cbaf4.js";import{H as N,E as me}from"../chunks/MermaidChart.svelte_svelte_type_style_lang.bbffb525.js";function de(K){let o,x,A,L,d,U,f,Z,p,O=`The <code>RAEDiT2DModel</code> is the Stage-2 latent diffusion transformer introduced in
<a href="https://huggingface.co/papers/2510.11690" rel="nofollow">Diffusion Transformers with Representation Autoencoders</a>.`,F,u,X=`Unlike DiT models that operate on VAE latents, this transformer denoises the latent space learned by
<a href="./autoencoder_rae"><code>AutoencoderRAE</code></a>. It is designed to be used with <a href="/docs/diffusers/pr_13231/en/api/schedulers/flow_match_euler_discrete#diffusers.FlowMatchEulerDiscreteScheduler">FlowMatchEulerDiscreteScheduler</a> and
decoded back to RGB with <a href="/docs/diffusers/pr_13231/en/api/models/autoencoder_rae#diffusers.AutoencoderRAE">AutoencoderRAE</a>.`,H,c,P,h,S,$,J,i,g,B,w,Q="Stage-2 latent diffusion transformer used by the RAE paper.",I,y,Y=`The architecture mirrors the upstream two-stream <code>DiTwDDTHead</code> design:
an encoder path first builds conditioning tokens from the latent input,
then a decoder path denoises the latent tokens conditioned on those
encoded tokens.`,V,_,j,k,z;return d=new ie({props:{containerStyle:"float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"}}),f=new N({props:{title:"RAEDiT2DModel",local:"raedit2dmodel",headingTag:"h1"}}),c=new N({props:{title:"Loading a pretrained transformer",local:"loading-a-pretrained-transformer",headingTag:"h2"}}),h=new le({props:{code:"ZnJvbSUyMGRpZmZ1c2VycyUyMGltcG9ydCUyMFJBRURpVDJETW9kZWwlMEElMEF0cmFuc2Zvcm1lciUyMCUzRCUyMFJBRURpVDJETW9kZWwuZnJvbV9wcmV0cmFpbmVkKCUyMnBhdGglMkZ0byUyRmNvbnZlcnRlZC1zdGFnZTItdHJhbnNmb3JtZXIlMjIp",highlighted:`<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> RAEDiT2DModel

transformer = RAEDiT2DModel.from_pretrained(<span class="hljs-string">&quot;path/to/converted-stage2-transformer&quot;</span>)`,lang:"python",wrap:!1}}),$=new N({props:{title:"RAEDiT2DModel",local:"diffusers.RAEDiT2DModel",headingTag:"h2"}}),g=new oe({props:{name:"class diffusers.RAEDiT2DModel",anchor:"diffusers.RAEDiT2DModel",parameters:[{name:"sample_size",val:": int = 16"},{name:"patch_size",val:": int | tuple[int, int] | list[int] = 1"},{name:"in_channels",val:": int = 768"},{name:"hidden_size",val:": int | tuple[int, int] | list[int] = (1152, 2048)"},{name:"depth",val:": int | tuple[int, int] | list[int] = (28, 2)"},{name:"num_heads",val:": int | tuple[int, int] | list[int] = (16, 16)"},{name:"mlp_ratio",val:": float = 4.0"},{name:"class_dropout_prob",val:": float = 0.1"},{name:"num_classes",val:": int = 1000"},{name:"use_qknorm",val:": bool = False"},{name:"use_swiglu",val:": bool = True"},{name:"use_rope",val:": bool = True"},{name:"use_rmsnorm",val:": bool = True"},{name:"wo_shift",val:": bool = False"},{name:"use_pos_embed",val:": bool = True"}],source:"https://github.com/huggingface/diffusers/blob/vr_13231/src/diffusers/models/transformers/transformer_rae_dit.py#L377"}}),_=new me({props:{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/models/rae_dit_transformer2d.md"}}),{c(){o=l("meta"),x=r(),A=l("p"),L=r(),v(d.$$.fragment),U=r(),v(f.$$.fragment),Z=r(),p=l("p"),p.innerHTML=O,F=r(),u=l("p"),u.innerHTML=X,H=r(),v(c.$$.fragment),P=r(),v(h.$$.fragment),S=r(),v($.$$.fragment),J=r(),i=l("div"),v(g.$$.fragment),B=r(),w=l("p"),w.textContent=Q,I=r(),y=l("p"),y.innerHTML=Y,V=r(),v(_.$$.fragment),j=r(),k=l("p"),this.h()},l(e){const t=se("svelte-u9bgzb",document.head);o=m(t,"META",{name:!0,content:!0}),t.forEach(n),x=s(e),A=m(e,"P",{}),q(A).forEach(n),L=s(e),b(d.$$.fragment,e),U=s(e),b(f.$$.fragment,e),Z=s(e),p=m(e,"P",{"data-svelte-h":!0}),G(p)!=="svelte-4sdrgk"&&(p.innerHTML=O),F=s(e),u=m(e,"P",{"data-svelte-h":!0}),G(u)!=="svelte-fwu3kj"&&(u.innerHTML=X),H=s(e),b(c.$$.fragment,e),P=s(e),b(h.$$.fragment,e),S=s(e),b($.$$.fragment,e),J=s(e),i=m(e,"DIV",{class:!0});var R=q(i);b(g.$$.fragment,R),B=s(R),w=m(R,"P",{"data-svelte-h":!0}),G(w)!=="svelte-6887hg"&&(w.textContent=Q),I=s(R),y=m(R,"P",{"data-svelte-h":!0}),G(y)!=="svelte-y0evfj"&&(y.innerHTML=Y),R.forEach(n),V=s(e),b(_.$$.fragment,e),j=s(e),k=m(e,"P",{}),q(k).forEach(n),this.h()},h(){W(o,"name","hf:doc:metadata"),W(o,"content",fe),W(i,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8")},m(e,t){C(document.head,o),a(e,x,t),a(e,A,t),a(e,L,t),T(d,e,t),a(e,U,t),T(f,e,t),a(e,Z,t),a(e,p,t),a(e,F,t),a(e,u,t),a(e,H,t),T(c,e,t),a(e,P,t),T(h,e,t),a(e,S,t),T($,e,t),a(e,J,t),a(e,i,t),T(g,i,null),C(i,B),C(i,w),C(i,I),C(i,y),a(e,V,t),T(_,e,t),a(e,j,t),a(e,k,t),z=!0},p:te,i(e){z||(D(d.$$.fragment,e),D(f.$$.fragment,e),D(c.$$.fragment,e),D(h.$$.fragment,e),D($.$$.fragment,e),D(g.$$.fragment,e),D(_.$$.fragment,e),z=!0)},o(e){M(d.$$.fragment,e),M(f.$$.fragment,e),M(c.$$.fragment,e),M(h.$$.fragment,e),M($.$$.fragment,e),M(g.$$.fragment,e),M(_.$$.fragment,e),z=!1},d(e){e&&(n(x),n(A),n(L),n(U),n(Z),n(p),n(F),n(u),n(H),n(P),n(S),n(J),n(i),n(V),n(j),n(k)),n(o),E(d,e),E(f,e),E(c,e),E(h,e),E($,e),E(g),E(_,e)}}}const fe='{"title":"RAEDiT2DModel","local":"raedit2dmodel","sections":[{"title":"Loading a pretrained transformer","local":"loading-a-pretrained-transformer","sections":[],"depth":2},{"title":"RAEDiT2DModel","local":"diffusers.RAEDiT2DModel","sections":[],"depth":2}],"depth":1}';function pe(K){return ne(()=>{new URLSearchParams(window.location.search).get("fw")}),[]}class ve extends ae{constructor(o){super(),re(this,o,pe,de,ee,{})}}export{ve as component};