diffusers/pr_13966/en/_app/immutable/nodes/74.HfESBswL.js

import"../chunks/DsnmJJEf.js";import{i as T,h as D,H as l,D as m,E as L,s as w}from"../chunks/BtE7mKSK.js";import{p as M,o as N,s as e,f as x,a as c,b as y,c as f,d as _,n as p,r as h}from"../chunks/jDjavuwI.js";const k='{"title":"LatteTransformer3DModel","local":"lattetransformer3dmodel","sections":[],"depth":2}';var O=_('<meta name="hf:doc:metadata"/>'),z=_('<p></p> <!> <p>A Diffusion Transformer model for 3D data from <a href="https://github.com/Vchitect/Latte" rel="nofollow">Latte</a>.</p> <!> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!> <div class="docstring border-l-2 border-t-2 pl-4 pt-3.5 border-gray-100 rounded-tl-xl mb-6 mt-8"><!> <p>The <a href="/docs/diffusers/pr_13966/en/api/models/latte_transformer3d#diffusers.LatteTransformer3DModel">LatteTransformer3DModel</a> forward method.</p></div></div> <!> <p></p>',1);function F(u,g){M(g,!1),N(()=>{new URLSearchParams(window.location.search).get("fw")}),T();var o=z();D("1n8jchu",d=>{var i=O();w(i,"content",k),c(d,i)});var n=e(x(o),2);l(n,{title:"LatteTransformer3DModel",local:"lattetransformer3dmodel",headingTag:"h2"});var a=e(n,4);l(a,{title:"LatteTransformer3DModel",local:"diffusers.LatteTransformer3DModel",headingTag:"h2"});var t=e(a,2),r=f(t);m(r,{name:"class diffusers.LatteTransformer3DModel",anchor:"diffusers.LatteTransformer3DModel",source:"https://github.com/huggingface/diffusers/blob/vr_13966/src/diffusers/models/transformers/latte_transformer_3d.py#L27",parameters:[{name:"num_attention_heads",val:": int = 16"},{name:"attention_head_dim",val:": int = 88"},{name:"in_channels",val:": int | None = None"},{name:"out_channels",val:": int | None = None"},{name:"num_layers",val:": int = 1"},{name:"dropout",val:": float = 0.0"},{name:"cross_attention_dim",val:": int | None = None"},{name:"attention_bias",val:": bool = False"},{name:"sample_size",val:": int = 64"},{name:"patch_size",val:": int | None = None"},{name:"activation_fn",val:": str = 'geglu'"},{name:"num_embeds_ada_norm",val:": int | None = None"},{name:"norm_type",val:": str = 'layer_norm'"},{name:"norm_elementwise_affine",val:": bool = True"},{name:"norm_eps",val:": float = 1e-05"},{name:"caption_channels",val:": int = None"},{name:"video_length",val:": int = 16"}]});var s=e(r,2),b=f(s);m(b,{name:"forward",anchor:"diffusers.LatteTransformer3DModel.forward",source:"https://github.com/huggingface/diffusers/blob/vr_13966/src/diffusers/models/transformers/latte_transformer_3d.py#L166",parameters:[{name:"hidden_states",val:": Tensor"},{name:"timestep",val:": typing.Optional[torch.LongTensor] = None"},{name:"encoder_hidden_states",val:": typing.Optional[torch.Tensor] = None"},{name:"encoder_attention_mask",val:": typing.Optional[torch.Tensor] = None"},{name:"enable_temporal_attentions",val:": bool = True"},{name:"return_dict",val:": bool = True"}],parametersDescription:[{anchor:"diffusers.LatteTransformer3DModel.forward.hidden_states",description:`<strong>hidden_states</strong> (<code>torch.Tensor</code> of shape <code>(batch size, channel, num_frame, height, width)</code>) &#x2014;
Input <code>hidden_states</code>.`,name:"hidden_states"},{anchor:"diffusers.LatteTransformer3DModel.forward.timestep",description:`<strong>timestep</strong> ( <code>torch.LongTensor</code>, <em>optional</em>) &#x2014;
Used to indicate denoising step. Optional timestep to be applied as an embedding in <code>AdaLayerNorm</code>.`,name:"timestep"},{anchor:"diffusers.LatteTransformer3DModel.forward.encoder_hidden_states",description:`<strong>encoder_hidden_states</strong> ( <code>torch.FloatTensor</code> of shape <code>(batch size, sequence len, embed dims)</code>, <em>optional</em>) &#x2014;
Conditional embeddings for cross attention layer. If not given, cross-attention defaults to
self-attention.`,name:"encoder_hidden_states"},{anchor:"diffusers.LatteTransformer3DModel.forward.encoder_attention_mask",description:`<strong>encoder_attention_mask</strong> ( <code>torch.Tensor</code>, <em>optional</em>) &#x2014;
Cross-attention mask applied to <code>encoder_hidden_states</code>. Two formats supported:</p>
<ul>
<li>Mask <code>(batcheight, sequence_length)</code> True = keep, False = discard.</li>
<li>Bias <code>(batcheight, 1, sequence_length)</code> 0 = keep, -10000 = discard.</li>
</ul>
<p>If <code>ndim == 2</code>: will be interpreted as a mask, then converted into a bias consistent with the format
above. This bias will be added to the cross-attention scores.`,name:"encoder_attention_mask"},{anchor:"diffusers.LatteTransformer3DModel.forward.enable_temporal_attentions",description:`<strong>enable_temporal_attentions</strong> &#x2014;
(<code>bool</code>, <em>optional</em>, defaults to <code>True</code>): Whether to enable temporal attentions.`,name:"enable_temporal_attentions"},{anchor:"diffusers.LatteTransformer3DModel.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.unet_2d_condition.UNet2DConditionOutput</code> instead of a plain
tuple.`,name:"return_dict"}],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>
`}),p(2),h(s),h(t);var v=e(t,2);L(v,{source:"https://github.com/huggingface/diffusers/blob/main/docs/source/en/api/models/latte_transformer3d.md"}),p(2),c(u,o),y()}export{F as component};