diffusers/pr_13813/en/_app/immutable/nodes/68.12de78fc.js

import{s as _e,n as be,o as Te}from"../chunks/scheduler.53228c21.js";import{S as Me,i as $e,e as s,s as o,c as i,h as De,a as d,d as r,b as a,f as N,g as m,j as z,k as L,l as R,m as n,n as l,t as f,o as c,p}from"../chunks/index.cac5d66a.js";import{C as ye}from"../chunks/CopyLLMTxtMenu.5ac9ab94.js";import{D as me}from"../chunks/Docstring.8a316450.js";import{C as ge}from"../chunks/CodeBlock.606cbaf4.js";import{H as ae,E as ve}from"../chunks/MermaidChart.svelte_svelte_type_style_lang.92b2cd9d.js";function we(le){let u,F,G,V,b,j,T,E,M,fe='A Transformer model for image-like data from <a href="https://huggingface.co/HiDream-ai" rel="nofollow">HiDream-I1</a>.',Y,$,ce="The model can be loaded with the following code snippet.",Q,D,S,y,P,v,pe="GGUF checkpoints for the <code>HiDreamImageTransformer2DModel</code> can  be loaded using <code>~FromOriginalModelMixin.from_single_file</code>",q,w,O,H,X,h,I,se,_,x,ie,Z,ue='The <a href="/docs/diffusers/pr_13813/en/api/models/hidream_image_transformer#diffusers.HiDreamImageTransformer2DModel">HiDreamImageTransformer2DModel</a> forward method.',A,J,B,g,U,de,C,he='The output of <a href="/docs/diffusers/pr_13813/en/api/models/transformer2d#diffusers.Transformer2DModel">Transformer2DModel</a>.',K,k,ee,W,te;return b=new ye({props:{containerStyle:"float: right; margin-left: 10px; display: inline-flex; position: relative; z-index: 10;"}}),T=new ae({props:{title:"HiDreamImageTransformer2DModel",local:"hidreamimagetransformer2dmodel",headingTag:"h1"}}),D=new ge({props:{code:"ZnJvbSUyMGRpZmZ1c2VycyUyMGltcG9ydCUyMEhpRHJlYW1JbWFnZVRyYW5zZm9ybWVyMkRNb2RlbCUwQSUwQXRyYW5zZm9ybWVyJTIwJTNEJTIwSGlEcmVhbUltYWdlVHJhbnNmb3JtZXIyRE1vZGVsLmZyb21fcHJldHJhaW5lZCglMjJIaURyZWFtLWFpJTJGSGlEcmVhbS1JMS1GdWxsJTIyJTJDJTIwc3ViZm9sZGVyJTNEJTIydHJhbnNmb3JtZXIlMjIlMkMlMjB0b3JjaF9kdHlwZSUzRHRvcmNoLmJmbG9hdDE2KQ==",highlighted:`<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> HiDreamImageTransformer2DModel

transformer = HiDreamImageTransformer2DModel.from_pretrained(<span class="hljs-string">&quot;HiDream-ai/HiDream-I1-Full&quot;</span>, subfolder=<span class="hljs-string">&quot;transformer&quot;</span>, torch_dtype=torch.bfloat16)`,lang:"python",wrap:!1}}),y=new ae({props:{title:"Loading GGUF quantized checkpoints for HiDream-I1",local:"loading-gguf-quantized-checkpoints-for-hidream-i1",headingTag:"h2"}}),w=new ge({props:{code:"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",highlighted:`<span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> GGUFQuantizationConfig, HiDreamImageTransformer2DModel

ckpt_path = <span class="hljs-string">&quot;https://huggingface.co/city96/HiDream-I1-Dev-gguf/blob/main/hidream-i1-dev-Q2_K.gguf&quot;</span>
transformer = HiDreamImageTransformer2DModel.from_single_file(
    ckpt_path,
    quantization_config=GGUFQuantizationConfig(compute_dtype=torch.bfloat16),
    torch_dtype=torch.bfloat16
)`,lang:"python",wrap:!1}}),H=new ae({props:{title:"HiDreamImageTransformer2DModel",local:"diffusers.HiDreamImageTransformer2DModel",headingTag:"h2"}}),I=new me({props:{name:"class diffusers.HiDreamImageTransformer2DModel",anchor:"diffusers.HiDreamImageTransformer2DModel",parameters:[{name:"patch_size",val:": int | None = None"},{name:"in_channels",val:": int = 64"},{name:"out_channels",val:": int | None = None"},{name:"num_layers",val:": int = 16"},{name:"num_single_layers",val:": int = 32"},{name:"attention_head_dim",val:": int = 128"},{name:"num_attention_heads",val:": int = 20"},{name:"caption_channels",val:": list = None"},{name:"text_emb_dim",val:": int = 2048"},{name:"num_routed_experts",val:": int = 4"},{name:"num_activated_experts",val:": int = 2"},{name:"axes_dims_rope",val:": tuple = (32, 32)"},{name:"max_resolution",val:": tuple = (128, 128)"},{name:"llama_layers",val:": list = None"},{name:"force_inference_output",val:": bool = False"}],source:"https://github.com/huggingface/diffusers/blob/vr_13813/src/diffusers/models/transformers/transformer_hidream_image.py#L605"}}),x=new me({props:{name:"forward",anchor:"diffusers.HiDreamImageTransformer2DModel.forward",parameters:[{name:"hidden_states",val:": Tensor"},{name:"timesteps",val:": LongTensor = None"},{name:"encoder_hidden_states_t5",val:": Tensor = None"},{name:"encoder_hidden_states_llama3",val:": Tensor = None"},{name:"pooled_embeds",val:": Tensor = None"},{name:"img_ids",val:": torch.Tensor | None = None"},{name:"img_sizes",val:": list[tuple[int, int]] | None = None"},{name:"hidden_states_masks",val:": torch.Tensor | None = None"},{name:"attention_kwargs",val:": dict[str, typing.Any] | None = None"},{name:"return_dict",val:": bool = True"},{name:"**kwargs",val:""}],parametersDescription:[{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.hidden_states",description:`<strong>hidden_states</strong> (<code>torch.Tensor</code> of shape <code>(batch_size, in_channels, height, width)</code> or <code>(batch_size, patch_height * patch_width, patch_size * patch_size * channels)</code>) &#x2014;
Input <code>hidden_states</code>.`,name:"hidden_states"},{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.timesteps",description:`<strong>timesteps</strong> (<code>torch.LongTensor</code>) &#x2014;
Used to indicate denoising step.`,name:"timesteps"},{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.encoder_hidden_states_t5",description:`<strong>encoder_hidden_states_t5</strong> (<code>torch.Tensor</code>) &#x2014;
Conditional embeddings computed from the T5 text encoder.`,name:"encoder_hidden_states_t5"},{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.encoder_hidden_states_llama3",description:`<strong>encoder_hidden_states_llama3</strong> (<code>torch.Tensor</code>) &#x2014;
Conditional embeddings computed from the Llama3 text encoder.`,name:"encoder_hidden_states_llama3"},{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.pooled_embeds",description:`<strong>pooled_embeds</strong> (<code>torch.Tensor</code>) &#x2014;
Pooled text embeddings used for additional conditioning.`,name:"pooled_embeds"},{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.img_ids",description:`<strong>img_ids</strong> (<code>torch.Tensor</code>, <em>optional</em>) &#x2014;
Image position ids for the patched hidden states.`,name:"img_ids"},{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.img_sizes",description:`<strong>img_sizes</strong> (<code>list</code> of <code>tuple</code> of <code>int</code>, <em>optional</em>) &#x2014;
Per-sample patch grid sizes used to unpatchify the output.`,name:"img_sizes"},{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.hidden_states_masks",description:`<strong>hidden_states_masks</strong> (<code>torch.Tensor</code>, <em>optional</em>) &#x2014;
Mask over patched <code>hidden_states</code>.`,name:"hidden_states_masks"},{anchor:"diffusers.HiDreamImageTransformer2DModel.forward.attention_kwargs",description:`<strong>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:"attention_kwargs"},{anchor:"diffusers.HiDreamImageTransformer2DModel.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_13813/src/diffusers/models/transformers/transformer_hidream_image.py#L776",returnDescription:`<script context="module">export const metadata = 'undefined';<\/script>


<p>If <code>return_dict</code> is True, an <code>~models.transformer_2d.Transformer2DModelOutput</code> is returned, otherwise a
<code>tuple</code> where the first element is the sample tensor.</p>
`}}),J=new ae({props:{title:"Transformer2DModelOutput",local:"diffusers.models.modeling_outputs.Transformer2DModelOutput",headingTag:"h2"}}),U=new me({props:{name:"class diffusers.models.modeling_outputs.Transformer2DModelOutput",anchor:"diffusers.models.modeling_outputs.Transformer2DModelOutput",parameters:[{name:"sample",val:": torch.Tensor"}],parametersDescription:[{anchor:"diffusers.models.modeling_outputs.Transformer2DModelOutput.sample",description:`<strong>sample</strong> (<code>torch.Tensor</code> of shape <code>(batch_size, num_channels, height, width)</code> or <code>(batch size, num_vector_embeds - 1, num_latent_pixels)</code> if <a href="/docs/diffusers/pr_13813/en/api/models/transformer2d#diffusers.Transformer2DModel">Transformer2DModel</a> is discrete) &#x2014;
The hidden states output conditioned on the <code>encoder_hidden_states</code> input. If discrete, returns probability
distributions for the unnoised latent pixels.`,name:"sample"}],source:"https://github.com/huggingface/diffusers/blob/vr_13813/src/diffusers/models/modeling_outputs.py#L21"}}),k=new ve({props:{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/models/hidream_image_transformer.md"}}),{c(){u=s("meta"),F=o(),G=s("p"),V=o(),i(b.$$.fragment),j=o(),i(T.$$.fragment),E=o(),M=s("p"),M.innerHTML=fe,Y=o(),$=s("p"),$.textContent=ce,Q=o(),i(D.$$.fragment),S=o(),i(y.$$.fragment),P=o(),v=s("p"),v.innerHTML=pe,q=o(),i(w.$$.fragment),O=o(),i(H.$$.fragment),X=o(),h=s("div"),i(I.$$.fragment),se=o(),_=s("div"),i(x.$$.fragment),ie=o(),Z=s("p"),Z.innerHTML=ue,A=o(),i(J.$$.fragment),B=o(),g=s("div"),i(U.$$.fragment),de=o(),C=s("p"),C.innerHTML=he,K=o(),i(k.$$.fragment),ee=o(),W=s("p"),this.h()},l(e){const t=De("svelte-u9bgzb",document.head);u=d(t,"META",{name:!0,content:!0}),t.forEach(r),F=a(e),G=d(e,"P",{}),N(G).forEach(r),V=a(e),m(b.$$.fragment,e),j=a(e),m(T.$$.fragment,e),E=a(e),M=d(e,"P",{"data-svelte-h":!0}),z(M)!=="svelte-ta0ebe"&&(M.innerHTML=fe),Y=a(e),$=d(e,"P",{"data-svelte-h":!0}),z($)!=="svelte-1vuni30"&&($.textContent=ce),Q=a(e),m(D.$$.fragment,e),S=a(e),m(y.$$.fragment,e),P=a(e),v=d(e,"P",{"data-svelte-h":!0}),z(v)!=="svelte-p22jvc"&&(v.innerHTML=pe),q=a(e),m(w.$$.fragment,e),O=a(e),m(H.$$.fragment,e),X=a(e),h=d(e,"DIV",{class:!0});var re=N(h);m(I.$$.fragment,re),se=a(re),_=d(re,"DIV",{class:!0});var ne=N(_);m(x.$$.fragment,ne),ie=a(ne),Z=d(ne,"P",{"data-svelte-h":!0}),z(Z)!=="svelte-16w3qav"&&(Z.innerHTML=ue),ne.forEach(r),re.forEach(r),A=a(e),m(J.$$.fragment,e),B=a(e),g=d(e,"DIV",{class:!0});var oe=N(g);m(U.$$.fragment,oe),de=a(oe),C=d(oe,"P",{"data-svelte-h":!0}),z(C)!=="svelte-zeg0js"&&(C.innerHTML=he),oe.forEach(r),K=a(e),m(k.$$.fragment,e),ee=a(e),W=d(e,"P",{}),N(W).forEach(r),this.h()},h(){L(u,"name","hf:doc:metadata"),L(u,"content",He),L(_,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),L(h,"class","docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"),L(g,"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){R(document.head,u),n(e,F,t),n(e,G,t),n(e,V,t),l(b,e,t),n(e,j,t),l(T,e,t),n(e,E,t),n(e,M,t),n(e,Y,t),n(e,$,t),n(e,Q,t),l(D,e,t),n(e,S,t),l(y,e,t),n(e,P,t),n(e,v,t),n(e,q,t),l(w,e,t),n(e,O,t),l(H,e,t),n(e,X,t),n(e,h,t),l(I,h,null),R(h,se),R(h,_),l(x,_,null),R(_,ie),R(_,Z),n(e,A,t),l(J,e,t),n(e,B,t),n(e,g,t),l(U,g,null),R(g,de),R(g,C),n(e,K,t),l(k,e,t),n(e,ee,t),n(e,W,t),te=!0},p:be,i(e){te||(f(b.$$.fragment,e),f(T.$$.fragment,e),f(D.$$.fragment,e),f(y.$$.fragment,e),f(w.$$.fragment,e),f(H.$$.fragment,e),f(I.$$.fragment,e),f(x.$$.fragment,e),f(J.$$.fragment,e),f(U.$$.fragment,e),f(k.$$.fragment,e),te=!0)},o(e){c(b.$$.fragment,e),c(T.$$.fragment,e),c(D.$$.fragment,e),c(y.$$.fragment,e),c(w.$$.fragment,e),c(H.$$.fragment,e),c(I.$$.fragment,e),c(x.$$.fragment,e),c(J.$$.fragment,e),c(U.$$.fragment,e),c(k.$$.fragment,e),te=!1},d(e){e&&(r(F),r(G),r(V),r(j),r(E),r(M),r(Y),r($),r(Q),r(S),r(P),r(v),r(q),r(O),r(X),r(h),r(A),r(B),r(g),r(K),r(ee),r(W)),r(u),p(b,e),p(T,e),p(D,e),p(y,e),p(w,e),p(H,e),p(I),p(x),p(J,e),p(U),p(k,e)}}}const He='{"title":"HiDreamImageTransformer2DModel","local":"hidreamimagetransformer2dmodel","sections":[{"title":"Loading GGUF quantized checkpoints for HiDream-I1","local":"loading-gguf-quantized-checkpoints-for-hidream-i1","sections":[],"depth":2},{"title":"HiDreamImageTransformer2DModel","local":"diffusers.HiDreamImageTransformer2DModel","sections":[],"depth":2},{"title":"Transformer2DModelOutput","local":"diffusers.models.modeling_outputs.Transformer2DModelOutput","sections":[],"depth":2}],"depth":1}';function Ie(le){return Te(()=>{new URLSearchParams(window.location.search).get("fw")}),[]}class Ce extends Me{constructor(u){super(),$e(this,u,Ie,we,_e,{})}}export{Ce as component};