diffusers/pr_13751/en/_app/immutable/nodes/80.aa4f5512.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 d,s as i,c as N,h as se,a as m,d as n,b as a,f as W,g as $,j,k as X,l,m as s,n as L,t as M,o as w,p as y}from"../chunks/index.cac5d66a.js";import{C as re}from"../chunks/CopyLLMTxtMenu.127444ce.js";import{D as Z}from"../chunks/Docstring.3f02c614.js";import{H as ee,E as de}from"../chunks/MermaidChart.svelte_svelte_type_style_lang.1e8e5da3.js";function me(B){let r,C,k,P,u,A,p,E,h,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>.',I,_,q,o,g,F,D,K="LuminaNextDiT: Diffusion model with a Transformer backbone.",G,b,Q="Inherit ModelMixin and ConfigMixin to be compatible with the sampler StableDiffusionPipeline of diffusers.",R,f,x,U,v,Y="Forward pass of LuminaNextDiT.",H,T,S,z,O;return u=new re({props:{containerStyle:"float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"}}),p=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"},{anchor:"diffusers.LuminaNextDiT2DModel.forward.image_rotary_emb",description:`<strong>image_rotary_emb</strong> (<code>torch.Tensor</code>) &#x2014;
Pre-computed rotary positional embeddings.`,name:"image_rotary_emb"},{anchor:"diffusers.LuminaNextDiT2DModel.forward.cross_attention_kwargs",description:`<strong>cross_attention_kwargs</strong> (<code>dict</code>, <em>optional</em>) &#x2014;
A kwargs dictionary that if specified is passed along to the <code>AttentionProcessor</code> as defined under
<code>self.processor</code> in
<a href="https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py" rel="nofollow">diffusers.models.attention_processor</a>.`,name:"cross_attention_kwargs"},{anchor:"diffusers.LuminaNextDiT2DModel.forward.return_dict",description:`<strong>return_dict</strong> (<code>bool</code>, <em>optional</em>, defaults to <code>True</code>) &#x2014;
Whether or not to return a <code>~models.transformer_2d.Transformer2DModelOutput</code> instead of a plain
tuple.`,name:"return_dict"}],source:"https://github.com/huggingface/diffusers/blob/vr_13751/src/diffusers/models/transformers/lumina_nextdit2d.py#L291",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


<p>If <code>return_dict</code> is True, a <code>~models.transformer_2d.Transformer2DModelOutput</code> is returned, otherwise
a plain <code>tuple</code> is returned.</p>
`,returnType:`<script context="module">export const metadata = 'undefined';<\/script>


<p><code>~models.transformer_2d.Transformer2DModelOutput</code> or <code>tuple</code></p>
`}}),T=new de({props:{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/models/lumina_nextdit2d.md"}}),{c(){r=d("meta"),C=i(),k=d("p"),P=i(),N(u.$$.fragment),A=i(),N(p.$$.fragment),E=i(),h=d("p"),h.innerHTML=J,I=i(),N(_.$$.fragment),q=i(),o=d("div"),N(g.$$.fragment),F=i(),D=d("p"),D.textContent=K,G=i(),b=d("p"),b.textContent=Q,R=i(),f=d("div"),N(x.$$.fragment),U=i(),v=d("p"),v.textContent=Y,H=i(),N(T.$$.fragment),S=i(),z=d("p"),this.h()},l(e){const t=se("svelte-u9bgzb",document.head);r=m(t,"META",{name:!0,content:!0}),t.forEach(n),C=a(e),k=m(e,"P",{}),W(k).forEach(n),P=a(e),$(u.$$.fragment,e),A=a(e),$(p.$$.fragment,e),E=a(e),h=m(e,"P",{"data-svelte-h":!0}),j(h)!=="svelte-13vz9fn"&&(h.innerHTML=J),I=a(e),$(_.$$.fragment,e),q=a(e),o=m(e,"DIV",{class:!0});var c=W(o);$(g.$$.fragment,c),F=a(c),D=m(c,"P",{"data-svelte-h":!0}),j(D)!=="svelte-8adszf"&&(D.textContent=K),G=a(c),b=m(c,"P",{"data-svelte-h":!0}),j(b)!=="svelte-wuyqug"&&(b.textContent=Q),R=a(c),f=m(c,"DIV",{class:!0});var V=W(f);$(x.$$.fragment,V),U=a(V),v=m(V,"P",{"data-svelte-h":!0}),j(v)!=="svelte-vv6okb"&&(v.textContent=Y),V.forEach(n),c.forEach(n),H=a(e),$(T.$$.fragment,e),S=a(e),z=m(e,"P",{}),W(z).forEach(n),this.h()},h(){X(r,"name","hf:doc:metadata"),X(r,"content",le),X(f,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),X(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,k,t),s(e,P,t),L(u,e,t),s(e,A,t),L(p,e,t),s(e,E,t),s(e,h,t),s(e,I,t),L(_,e,t),s(e,q,t),s(e,o,t),L(g,o,null),l(o,F),l(o,D),l(o,G),l(o,b),l(o,R),l(o,f),L(x,f,null),l(f,U),l(f,v),s(e,H,t),L(T,e,t),s(e,S,t),s(e,z,t),O=!0},p:ne,i(e){O||(M(u.$$.fragment,e),M(p.$$.fragment,e),M(_.$$.fragment,e),M(g.$$.fragment,e),M(x.$$.fragment,e),M(T.$$.fragment,e),O=!0)},o(e){w(u.$$.fragment,e),w(p.$$.fragment,e),w(_.$$.fragment,e),w(g.$$.fragment,e),w(x.$$.fragment,e),w(T.$$.fragment,e),O=!1},d(e){e&&(n(C),n(k),n(P),n(A),n(E),n(h),n(I),n(q),n(o),n(H),n(S),n(z)),n(r),y(u,e),y(p,e),y(_,e),y(g),y(x),y(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,me,te,{})}}export{ge as component};