diffusers/pr_13751/en/_app/immutable/nodes/76.34b5daa4.js

import{s as te,n as ne,o as oe}from"../chunks/scheduler.53228c21.js";import{S as ie,i as ae,e as m,s as i,c as $,h as se,a as d,d as n,b as a,f as W,g as b,j as X,k as F,l,m as s,n as L,t as M,o as y,p as w}from"../chunks/index.100fac89.js";import{C as re}from"../chunks/CopyLLMTxtMenu.af3e1493.js";import{D as Z}from"../chunks/Docstring.147b33f1.js";import{H as ee,E as me}from"../chunks/MermaidChart.svelte_svelte_type_style_lang.b5eefd91.js";function de(B){let r,C,z,P,c,E,h,q,p,J='A Next Version of Diffusion Transformer model for 2D data from <a href="https://github.com/Alpha-VLLM/Lumina-T2X" rel="nofollow">Lumina-T2X</a>.',A,_,I,o,g,G,D,K="LuminaNextDiT: Diffusion model with a Transformer backbone.",O,v,Q="Inherit ModelMixin and ConfigMixin to be compatible with the sampler StableDiffusionPipeline of diffusers.",R,f,x,U,N,Y="Forward pass of LuminaNextDiT.",H,T,S,k,V;return c=new re({props:{containerStyle:"float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"}}),h=new ee({props:{title:"LuminaNextDiT2DModel",local:"luminanextdit2dmodel",headingTag:"h1"}}),_=new ee({props:{title:"LuminaNextDiT2DModel",local:"diffusers.LuminaNextDiT2DModel",headingTag:"h2"}}),g=new Z({props:{name:"class diffusers.LuminaNextDiT2DModel",anchor:"diffusers.LuminaNextDiT2DModel",parameters:[{name:"sample_size",val:": int = 128"},{name:"patch_size",val:": int | None = 2"},{name:"in_channels",val:": int | None = 4"},{name:"hidden_size",val:": int | None = 2304"},{name:"num_layers",val:": int | None = 32"},{name:"num_attention_heads",val:": int | None = 32"},{name:"num_kv_heads",val:": int | None = None"},{name:"multiple_of",val:": int | None = 256"},{name:"ffn_dim_multiplier",val:": float | None = None"},{name:"norm_eps",val:": float | None = 1e-05"},{name:"learn_sigma",val:": bool | None = True"},{name:"qk_norm",val:": bool | None = True"},{name:"cross_attention_dim",val:": int | None = 2048"},{name:"scaling_factor",val:": float | None = 1.0"}],parametersDescription:[{anchor:"diffusers.LuminaNextDiT2DModel.sample_size",description:`<strong>sample_size</strong> (<code>int</code>) &#x2014; The width of the latent images. This is fixed during training since
it is used to learn a number of position embeddings.`,name:"sample_size"},{anchor:"diffusers.LuminaNextDiT2DModel.patch_size",description:`<strong>patch_size</strong> (<code>int</code>, <em>optional</em>, (<code>int</code>, <em>optional</em>, defaults to 2) &#x2014;
The size of each patch in the image. This parameter defines the resolution of patches fed into the model.`,name:"patch_size"},{anchor:"diffusers.LuminaNextDiT2DModel.in_channels",description:`<strong>in_channels</strong> (<code>int</code>, <em>optional</em>, defaults to 4) &#x2014;
The number of input channels for the model. Typically, this matches the number of channels in the input
images.`,name:"in_channels"},{anchor:"diffusers.LuminaNextDiT2DModel.hidden_size",description:`<strong>hidden_size</strong> (<code>int</code>, <em>optional</em>, defaults to 4096) &#x2014;
The dimensionality of the hidden layers in the model. This parameter determines the width of the model&#x2019;s
hidden representations.`,name:"hidden_size"},{anchor:"diffusers.LuminaNextDiT2DModel.num_layers",description:`<strong>num_layers</strong> (<code>int</code>, <em>optional</em>, default to 32) &#x2014;
The number of layers in the model. This defines the depth of the neural network.`,name:"num_layers"},{anchor:"diffusers.LuminaNextDiT2DModel.num_attention_heads",description:`<strong>num_attention_heads</strong> (<code>int</code>, <em>optional</em>, defaults to 32) &#x2014;
The number of attention heads in each attention layer. This parameter specifies how many separate attention
mechanisms are used.`,name:"num_attention_heads"},{anchor:"diffusers.LuminaNextDiT2DModel.num_kv_heads",description:`<strong>num_kv_heads</strong> (<code>int</code>, <em>optional</em>, defaults to 8) &#x2014;
The number of key-value heads in the attention mechanism, if different from the number of attention heads.
If None, it defaults to num_attention_heads.`,name:"num_kv_heads"},{anchor:"diffusers.LuminaNextDiT2DModel.multiple_of",description:`<strong>multiple_of</strong> (<code>int</code>, <em>optional</em>, defaults to 256) &#x2014;
A factor that the hidden size should be a multiple of. This can help optimize certain hardware
configurations.`,name:"multiple_of"},{anchor:"diffusers.LuminaNextDiT2DModel.ffn_dim_multiplier",description:`<strong>ffn_dim_multiplier</strong> (<code>float</code>, <em>optional</em>) &#x2014;
A multiplier for the dimensionality of the feed-forward network. If None, it uses a default value based on
the model configuration.`,name:"ffn_dim_multiplier"},{anchor:"diffusers.LuminaNextDiT2DModel.norm_eps",description:`<strong>norm_eps</strong> (<code>float</code>, <em>optional</em>, defaults to 1e-5) &#x2014;
A small value added to the denominator for numerical stability in normalization layers.`,name:"norm_eps"},{anchor:"diffusers.LuminaNextDiT2DModel.learn_sigma",description:`<strong>learn_sigma</strong> (<code>bool</code>, <em>optional</em>, defaults to True) &#x2014;
Whether the model should learn the sigma parameter, which might be related to uncertainty or variance in
predictions.`,name:"learn_sigma"},{anchor:"diffusers.LuminaNextDiT2DModel.qk_norm",description:`<strong>qk_norm</strong> (<code>bool</code>, <em>optional</em>, defaults to True) &#x2014;
Indicates if the queries and keys in the attention mechanism should be normalized.`,name:"qk_norm"},{anchor:"diffusers.LuminaNextDiT2DModel.cross_attention_dim",description:`<strong>cross_attention_dim</strong> (<code>int</code>, <em>optional</em>, defaults to 2048) &#x2014;
The dimensionality of the text embeddings. This parameter defines the size of the text representations used
in the model.`,name:"cross_attention_dim"},{anchor:"diffusers.LuminaNextDiT2DModel.scaling_factor",description:`<strong>scaling_factor</strong> (<code>float</code>, <em>optional</em>, defaults to 1.0) &#x2014;
A scaling factor applied to certain parameters or layers in the model. This can be used for adjusting the
overall scale of the model&#x2019;s operations.`,name:"scaling_factor"}],source:"https://github.com/huggingface/diffusers/blob/vr_13751/src/diffusers/models/transformers/lumina_nextdit2d.py#L178"}}),x=new Z({props:{name:"forward",anchor:"diffusers.LuminaNextDiT2DModel.forward",parameters:[{name:"hidden_states",val:": Tensor"},{name:"timestep",val:": Tensor"},{name:"encoder_hidden_states",val:": Tensor"},{name:"encoder_mask",val:": Tensor"},{name:"image_rotary_emb",val:": Tensor"},{name:"cross_attention_kwargs",val:": dict = None"},{name:"return_dict",val:" = True"}],parametersDescription:[{anchor:"diffusers.LuminaNextDiT2DModel.forward.hidden_states",description:"<strong>hidden_states</strong> (torch.Tensor) &#x2014; Input tensor of shape (N, C, H, W).",name:"hidden_states"},{anchor:"diffusers.LuminaNextDiT2DModel.forward.timestep",description:"<strong>timestep</strong> (torch.Tensor) &#x2014; Tensor of diffusion timesteps of shape (N,).",name:"timestep"},{anchor:"diffusers.LuminaNextDiT2DModel.forward.encoder_hidden_states",description:"<strong>encoder_hidden_states</strong> (torch.Tensor) &#x2014; Tensor of caption features of shape (N, D).",name:"encoder_hidden_states"},{anchor:"diffusers.LuminaNextDiT2DModel.forward.encoder_mask",description:"<strong>encoder_mask</strong> (torch.Tensor) &#x2014; Tensor of caption masks of shape (N, L).",name:"encoder_mask"}],source:"https://github.com/huggingface/diffusers/blob/vr_13751/src/diffusers/models/transformers/lumina_nextdit2d.py#L291"}}),T=new me({props:{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/models/lumina_nextdit2d.md"}}),{c(){r=m("meta"),C=i(),z=m("p"),P=i(),$(c.$$.fragment),E=i(),$(h.$$.fragment),q=i(),p=m("p"),p.innerHTML=J,A=i(),$(_.$$.fragment),I=i(),o=m("div"),$(g.$$.fragment),G=i(),D=m("p"),D.textContent=K,O=i(),v=m("p"),v.textContent=Q,R=i(),f=m("div"),$(x.$$.fragment),U=i(),N=m("p"),N.textContent=Y,H=i(),$(T.$$.fragment),S=i(),k=m("p"),this.h()},l(e){const t=se("svelte-u9bgzb",document.head);r=d(t,"META",{name:!0,content:!0}),t.forEach(n),C=a(e),z=d(e,"P",{}),W(z).forEach(n),P=a(e),b(c.$$.fragment,e),E=a(e),b(h.$$.fragment,e),q=a(e),p=d(e,"P",{"data-svelte-h":!0}),X(p)!=="svelte-13vz9fn"&&(p.innerHTML=J),A=a(e),b(_.$$.fragment,e),I=a(e),o=d(e,"DIV",{class:!0});var u=W(o);b(g.$$.fragment,u),G=a(u),D=d(u,"P",{"data-svelte-h":!0}),X(D)!=="svelte-8adszf"&&(D.textContent=K),O=a(u),v=d(u,"P",{"data-svelte-h":!0}),X(v)!=="svelte-wuyqug"&&(v.textContent=Q),R=a(u),f=d(u,"DIV",{class:!0});var j=W(f);b(x.$$.fragment,j),U=a(j),N=d(j,"P",{"data-svelte-h":!0}),X(N)!=="svelte-vv6okb"&&(N.textContent=Y),j.forEach(n),u.forEach(n),H=a(e),b(T.$$.fragment,e),S=a(e),k=d(e,"P",{}),W(k).forEach(n),this.h()},h(){F(r,"name","hf:doc:metadata"),F(r,"content",le),F(f,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),F(o,"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){l(document.head,r),s(e,C,t),s(e,z,t),s(e,P,t),L(c,e,t),s(e,E,t),L(h,e,t),s(e,q,t),s(e,p,t),s(e,A,t),L(_,e,t),s(e,I,t),s(e,o,t),L(g,o,null),l(o,G),l(o,D),l(o,O),l(o,v),l(o,R),l(o,f),L(x,f,null),l(f,U),l(f,N),s(e,H,t),L(T,e,t),s(e,S,t),s(e,k,t),V=!0},p:ne,i(e){V||(M(c.$$.fragment,e),M(h.$$.fragment,e),M(_.$$.fragment,e),M(g.$$.fragment,e),M(x.$$.fragment,e),M(T.$$.fragment,e),V=!0)},o(e){y(c.$$.fragment,e),y(h.$$.fragment,e),y(_.$$.fragment,e),y(g.$$.fragment,e),y(x.$$.fragment,e),y(T.$$.fragment,e),V=!1},d(e){e&&(n(C),n(z),n(P),n(E),n(q),n(p),n(A),n(I),n(o),n(H),n(S),n(k)),n(r),w(c,e),w(h,e),w(_,e),w(g),w(x),w(T,e)}}}const le='{"title":"LuminaNextDiT2DModel","local":"luminanextdit2dmodel","sections":[{"title":"LuminaNextDiT2DModel","local":"diffusers.LuminaNextDiT2DModel","sections":[],"depth":2}],"depth":1}';function fe(B){return oe(()=>{new URLSearchParams(window.location.search).get("fw")}),[]}class ge extends ie{constructor(r){super(),ae(this,r,fe,de,te,{})}}export{ge as component};