diffusers/pr_13745/en/_app/immutable/nodes/66.4ca21b49.js

import{s as A,n as W,o as j}from"../chunks/scheduler.53228c21.js";import{S as R,i as U,e as g,s,c as p,h as V,a as $,d as n,b as r,f as L,g as b,j as X,k as S,l as B,m as o,n as T,t as x,o as v,p as D}from"../chunks/index.cac5d66a.js";import{C as F}from"../chunks/CopyLLMTxtMenu.4912207d.js";import{D as J}from"../chunks/Docstring.1e7ac4f3.js";import{H as q,E as K}from"../chunks/MermaidChart.svelte_svelte_type_style_lang.323ee77a.js";function N(k){let a,M,_,I,i,G,m,z,d,H="A Diffusion Transformer model for 2D data from [GlmImageTransformer2DModel] (TODO).",y,l,w,c,f,E,u,P,h,C;return i=new F({props:{containerStyle:"float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"}}),m=new q({props:{title:"GlmImageTransformer2DModel",local:"glmimagetransformer2dmodel",headingTag:"h1"}}),l=new q({props:{title:"GlmImageTransformer2DModel",local:"diffusers.GlmImageTransformer2DModel",headingTag:"h2"}}),f=new J({props:{name:"class diffusers.GlmImageTransformer2DModel",anchor:"diffusers.GlmImageTransformer2DModel",parameters:[{name:"patch_size",val:": int = 2"},{name:"in_channels",val:": int = 16"},{name:"out_channels",val:": int = 16"},{name:"num_layers",val:": int = 30"},{name:"attention_head_dim",val:": int = 40"},{name:"num_attention_heads",val:": int = 64"},{name:"text_embed_dim",val:": int = 1472"},{name:"time_embed_dim",val:": int = 512"},{name:"condition_dim",val:": int = 256"},{name:"prior_vq_quantizer_codebook_size",val:": int = 16384"}],parametersDescription:[{anchor:"diffusers.GlmImageTransformer2DModel.patch_size",description:`<strong>patch_size</strong> (<code>int</code>, defaults to <code>2</code>) &#x2014;
The size of the patches to use in the patch embedding layer.`,name:"patch_size"},{anchor:"diffusers.GlmImageTransformer2DModel.in_channels",description:`<strong>in_channels</strong> (<code>int</code>, defaults to <code>16</code>) &#x2014;
The number of channels in the input.`,name:"in_channels"},{anchor:"diffusers.GlmImageTransformer2DModel.num_layers",description:`<strong>num_layers</strong> (<code>int</code>, defaults to <code>30</code>) &#x2014;
The number of layers of Transformer blocks to use.`,name:"num_layers"},{anchor:"diffusers.GlmImageTransformer2DModel.attention_head_dim",description:`<strong>attention_head_dim</strong> (<code>int</code>, defaults to <code>40</code>) &#x2014;
The number of channels in each head.`,name:"attention_head_dim"},{anchor:"diffusers.GlmImageTransformer2DModel.num_attention_heads",description:`<strong>num_attention_heads</strong> (<code>int</code>, defaults to <code>64</code>) &#x2014;
The number of heads to use for multi-head attention.`,name:"num_attention_heads"},{anchor:"diffusers.GlmImageTransformer2DModel.out_channels",description:`<strong>out_channels</strong> (<code>int</code>, defaults to <code>16</code>) &#x2014;
The number of channels in the output.`,name:"out_channels"},{anchor:"diffusers.GlmImageTransformer2DModel.text_embed_dim",description:`<strong>text_embed_dim</strong> (<code>int</code>, defaults to <code>1472</code>) &#x2014;
Input dimension of text embeddings from the text encoder.`,name:"text_embed_dim"},{anchor:"diffusers.GlmImageTransformer2DModel.time_embed_dim",description:`<strong>time_embed_dim</strong> (<code>int</code>, defaults to <code>512</code>) &#x2014;
Output dimension of timestep embeddings.`,name:"time_embed_dim"},{anchor:"diffusers.GlmImageTransformer2DModel.condition_dim",description:`<strong>condition_dim</strong> (<code>int</code>, defaults to <code>256</code>) &#x2014;
The embedding dimension of the input SDXL-style resolution conditions (original_size, target_size,
crop_coords).`,name:"condition_dim"},{anchor:"diffusers.GlmImageTransformer2DModel.pos_embed_max_size",description:`<strong>pos_embed_max_size</strong> (<code>int</code>, defaults to <code>128</code>) &#x2014;
The maximum resolution of the positional embeddings, from which slices of shape <code>H x W</code> are taken and added
to input patched latents, where <code>H</code> and <code>W</code> are the latent height and width respectively. A value of 128
means that the maximum supported height and width for image generation is <code>128 * vae_scale_factor * patch_size =&gt; 128 * 8 * 2 =&gt; 2048</code>.`,name:"pos_embed_max_size"},{anchor:"diffusers.GlmImageTransformer2DModel.sample_size",description:`<strong>sample_size</strong> (<code>int</code>, defaults to <code>128</code>) &#x2014;
The base resolution of input latents. If height/width is not provided during generation, this value is used
to determine the resolution as <code>sample_size * vae_scale_factor =&gt; 128 * 8 =&gt; 1024</code>`,name:"sample_size"}],source:"https://github.com/huggingface/diffusers/blob/vr_13745/src/diffusers/models/transformers/transformer_glm_image.py#L503"}}),u=new K({props:{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/models/glm_image_transformer2d.md"}}),{c(){a=g("meta"),M=s(),_=g("p"),I=s(),p(i.$$.fragment),G=s(),p(m.$$.fragment),z=s(),d=g("p"),d.textContent=H,y=s(),p(l.$$.fragment),w=s(),c=g("div"),p(f.$$.fragment),E=s(),p(u.$$.fragment),P=s(),h=g("p"),this.h()},l(e){const t=V("svelte-u9bgzb",document.head);a=$(t,"META",{name:!0,content:!0}),t.forEach(n),M=r(e),_=$(e,"P",{}),L(_).forEach(n),I=r(e),b(i.$$.fragment,e),G=r(e),b(m.$$.fragment,e),z=r(e),d=$(e,"P",{"data-svelte-h":!0}),X(d)!=="svelte-f8vs2r"&&(d.textContent=H),y=r(e),b(l.$$.fragment,e),w=r(e),c=$(e,"DIV",{class:!0});var O=L(c);b(f.$$.fragment,O),O.forEach(n),E=r(e),b(u.$$.fragment,e),P=r(e),h=$(e,"P",{}),L(h).forEach(n),this.h()},h(){S(a,"name","hf:doc:metadata"),S(a,"content",Q),S(c,"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){B(document.head,a),o(e,M,t),o(e,_,t),o(e,I,t),T(i,e,t),o(e,G,t),T(m,e,t),o(e,z,t),o(e,d,t),o(e,y,t),T(l,e,t),o(e,w,t),o(e,c,t),T(f,c,null),o(e,E,t),T(u,e,t),o(e,P,t),o(e,h,t),C=!0},p:W,i(e){C||(x(i.$$.fragment,e),x(m.$$.fragment,e),x(l.$$.fragment,e),x(f.$$.fragment,e),x(u.$$.fragment,e),C=!0)},o(e){v(i.$$.fragment,e),v(m.$$.fragment,e),v(l.$$.fragment,e),v(f.$$.fragment,e),v(u.$$.fragment,e),C=!1},d(e){e&&(n(M),n(_),n(I),n(G),n(z),n(d),n(y),n(w),n(c),n(E),n(P),n(h)),n(a),D(i,e),D(m,e),D(l,e),D(f),D(u,e)}}}const Q='{"title":"GlmImageTransformer2DModel","local":"glmimagetransformer2dmodel","sections":[{"title":"GlmImageTransformer2DModel","local":"diffusers.GlmImageTransformer2DModel","sections":[],"depth":2}],"depth":1}';function Y(k){return j(()=>{new URLSearchParams(window.location.search).get("fw")}),[]}class ae extends R{constructor(a){super(),U(this,a,Y,N,A,{})}}export{ae as component};