kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-01-14 06:27:45 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
9b727f1c73
TensorRT-LLMs/python-api/tensorrt_llm.layers.html
石晓伟 b79ef8a080
update gh-pages (#2530)
2024-12-04 14:25:18 +08:00

859 lines
178 KiB
HTML
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<!DOCTYPE html>
<html class="writer-html5" lang="en" data-content_root="../">
<head>
<meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>Layers &mdash; tensorrt_llm documentation</title>
<link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=80d5e7a1" />
<link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=e59714d7" />
<link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
<script src="../_static/jquery.js?v=5d32c60e"></script>
<script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
<script src="../_static/documentation_options.js?v=5929fcd5"></script>
<script src="../_static/doctools.js?v=888ff710"></script>
<script src="../_static/sphinx_highlight.js?v=dc90522c"></script>
<script src="../_static/clipboard.min.js?v=a7894cd8"></script>
<script src="../_static/copybutton.js?v=65e89d2a"></script>
<script src="../_static/js/theme.js"></script>
<link rel="index" title="Index" href="../genindex.html" />
<link rel="search" title="Search" href="../search.html" />
<link rel="next" title="Functionals" href="tensorrt_llm.functional.html" />
<link rel="prev" title="Generate text with multiple LoRA adapters" href="../llm-api-examples/llm_multilora.html" />
</head>
<body class="wy-body-for-nav">
<div class="wy-grid-for-nav">
<nav data-toggle="wy-nav-shift" class="wy-nav-side">
<div class="wy-side-scroll">
<div class="wy-side-nav-search" >
<a href="../index.html" class="icon icon-home">
tensorrt_llm
</a>
<div role="search">
<form id="rtd-search-form" class="wy-form" action="../search.html" method="get">
<input type="text" name="q" placeholder="Search docs" aria-label="Search docs" />
<input type="hidden" name="check_keywords" value="yes" />
<input type="hidden" name="area" value="default" />
</form>
</div>
</div><div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="Navigation menu">
<p class="caption" role="heading"><span class="caption-text">Getting Started</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../overview.html">Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="../quick-start-guide.html">Quick Start Guide</a></li>
<li class="toctree-l1"><a class="reference internal" href="../key-features.html">Key Features</a></li>
<li class="toctree-l1"><a class="reference internal" href="../release-notes.html">Release Notes</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Installation</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../installation/linux.html">Installing on Linux</a></li>
<li class="toctree-l1"><a class="reference internal" href="../installation/build-from-source-linux.html">Building from Source Code on Linux</a></li>
<li class="toctree-l1"><a class="reference internal" href="../installation/windows.html">Installing on Windows</a></li>
<li class="toctree-l1"><a class="reference internal" href="../installation/build-from-source-windows.html">Building from Source Code on Windows</a></li>
<li class="toctree-l1"><a class="reference internal" href="../installation/grace-hopper.html">Installing on Grace Hopper</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">LLM API</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../llm-api/index.html">API Introduction</a></li>
<li class="toctree-l1"><a class="reference internal" href="../llm-api/reference.html">API Reference</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">LLM API Examples</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../llm-api-examples/index.html">LLM Examples Introduction</a></li>
<li class="toctree-l1"><a class="reference internal" href="../llm-api-examples/customization.html">Common Customizations</a></li>
<li class="toctree-l1"><a class="reference internal" href="../llm-api-examples/llm_api_examples.html">Examples</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Model Definition API</span></p>
<ul class="current">
<li class="toctree-l1 current"><a class="current reference internal" href="#">Layers</a><ul>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.activation">Activation</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.activation.Mish"><code class="docutils literal notranslate"><span class="pre">Mish</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.activation.Mish.forward"><code class="docutils literal notranslate"><span class="pre">Mish.forward()</span></code></a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.attention">Attention</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.Attention"><code class="docutils literal notranslate"><span class="pre">Attention</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.Attention.create_attention_const_params"><code class="docutils literal notranslate"><span class="pre">Attention.create_attention_const_params()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.Attention.fill_attention_params"><code class="docutils literal notranslate"><span class="pre">Attention.fill_attention_params()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.Attention.forward"><code class="docutils literal notranslate"><span class="pre">Attention.forward()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.Attention.postprocess"><code class="docutils literal notranslate"><span class="pre">Attention.postprocess()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.Attention.set_rel_attn_table"><code class="docutils literal notranslate"><span class="pre">Attention.set_rel_attn_table()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.AttentionMaskParams"><code class="docutils literal notranslate"><span class="pre">AttentionMaskParams</span></code></a></li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.AttentionParams"><code class="docutils literal notranslate"><span class="pre">AttentionParams</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.AttentionParams.fill_attention_const_params_for_long_rope"><code class="docutils literal notranslate"><span class="pre">AttentionParams.fill_attention_const_params_for_long_rope()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.AttentionParams.fill_attention_const_params_for_rope"><code class="docutils literal notranslate"><span class="pre">AttentionParams.fill_attention_const_params_for_rope()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.AttentionParams.is_valid"><code class="docutils literal notranslate"><span class="pre">AttentionParams.is_valid()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.AttentionParams.is_valid_cross_attn"><code class="docutils literal notranslate"><span class="pre">AttentionParams.is_valid_cross_attn()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.BertAttention"><code class="docutils literal notranslate"><span class="pre">BertAttention</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.BertAttention.forward"><code class="docutils literal notranslate"><span class="pre">BertAttention.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.BlockSparseAttnParams"><code class="docutils literal notranslate"><span class="pre">BlockSparseAttnParams</span></code></a></li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.CogVLMAttention"><code class="docutils literal notranslate"><span class="pre">CogVLMAttention</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.CogVLMAttention.forward"><code class="docutils literal notranslate"><span class="pre">CogVLMAttention.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.DeepseekV2Attention"><code class="docutils literal notranslate"><span class="pre">DeepseekV2Attention</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.DeepseekV2Attention.forward"><code class="docutils literal notranslate"><span class="pre">DeepseekV2Attention.forward()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.DeepseekV2Attention.weight_loader"><code class="docutils literal notranslate"><span class="pre">DeepseekV2Attention.weight_loader()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.KeyValueCacheParams"><code class="docutils literal notranslate"><span class="pre">KeyValueCacheParams</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.KeyValueCacheParams.fill_none_tensor_list"><code class="docutils literal notranslate"><span class="pre">KeyValueCacheParams.fill_none_tensor_list()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.KeyValueCacheParams.get_first_past_key_value"><code class="docutils literal notranslate"><span class="pre">KeyValueCacheParams.get_first_past_key_value()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.attention.KeyValueCacheParams.is_valid"><code class="docutils literal notranslate"><span class="pre">KeyValueCacheParams.is_valid()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.MropeParams"><code class="docutils literal notranslate"><span class="pre">MropeParams</span></code></a></li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.SpecDecodingParams"><code class="docutils literal notranslate"><span class="pre">SpecDecodingParams</span></code></a></li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.compute_relative_bias"><code class="docutils literal notranslate"><span class="pre">compute_relative_bias()</span></code></a></li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.attention.make_causal_mask"><code class="docutils literal notranslate"><span class="pre">make_causal_mask()</span></code></a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.cast">Cast</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.cast.Cast"><code class="docutils literal notranslate"><span class="pre">Cast</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.cast.Cast.forward"><code class="docutils literal notranslate"><span class="pre">Cast.forward()</span></code></a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.conv">Conv</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.conv.Conv1d"><code class="docutils literal notranslate"><span class="pre">Conv1d</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.conv.Conv1d.forward"><code class="docutils literal notranslate"><span class="pre">Conv1d.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.conv.Conv2d"><code class="docutils literal notranslate"><span class="pre">Conv2d</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.conv.Conv2d.forward"><code class="docutils literal notranslate"><span class="pre">Conv2d.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.conv.ConvTranspose2d"><code class="docutils literal notranslate"><span class="pre">ConvTranspose2d</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.conv.ConvTranspose2d.forward"><code class="docutils literal notranslate"><span class="pre">ConvTranspose2d.forward()</span></code></a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.embedding">Embedding</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.embedding.Embedding"><code class="docutils literal notranslate"><span class="pre">Embedding</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.embedding.Embedding.forward"><code class="docutils literal notranslate"><span class="pre">Embedding.forward()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.embedding.Embedding.postprocess"><code class="docutils literal notranslate"><span class="pre">Embedding.postprocess()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.embedding.Embedding.weight_loader"><code class="docutils literal notranslate"><span class="pre">Embedding.weight_loader()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.embedding.PromptTuningEmbedding"><code class="docutils literal notranslate"><span class="pre">PromptTuningEmbedding</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.embedding.PromptTuningEmbedding.forward"><code class="docutils literal notranslate"><span class="pre">PromptTuningEmbedding.forward()</span></code></a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.linear">Linear</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.linear.ColumnLinear"><code class="docutils literal notranslate"><span class="pre">ColumnLinear</span></code></a></li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.linear.Linear"><code class="docutils literal notranslate"><span class="pre">Linear</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.Linear.collect_and_bias"><code class="docutils literal notranslate"><span class="pre">Linear.collect_and_bias()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.Linear.postprocess"><code class="docutils literal notranslate"><span class="pre">Linear.postprocess()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.Linear.tp_split_dim"><code class="docutils literal notranslate"><span class="pre">Linear.tp_split_dim()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase"><code class="docutils literal notranslate"><span class="pre">LinearBase</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase.collect_and_bias"><code class="docutils literal notranslate"><span class="pre">LinearBase.collect_and_bias()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase.forward"><code class="docutils literal notranslate"><span class="pre">LinearBase.forward()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase.get_weight"><code class="docutils literal notranslate"><span class="pre">LinearBase.get_weight()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase.multiply_and_lora"><code class="docutils literal notranslate"><span class="pre">LinearBase.multiply_and_lora()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase.multiply_collect"><code class="docutils literal notranslate"><span class="pre">LinearBase.multiply_collect()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase.tp_split_dim"><code class="docutils literal notranslate"><span class="pre">LinearBase.tp_split_dim()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase.weight_is_kn"><code class="docutils literal notranslate"><span class="pre">LinearBase.weight_is_kn()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase.weight_loader"><code class="docutils literal notranslate"><span class="pre">LinearBase.weight_loader()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.linear.RowLinear"><code class="docutils literal notranslate"><span class="pre">RowLinear</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.RowLinear.collect_and_bias"><code class="docutils literal notranslate"><span class="pre">RowLinear.collect_and_bias()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.linear.RowLinear.tp_split_dim"><code class="docutils literal notranslate"><span class="pre">RowLinear.tp_split_dim()</span></code></a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.mlp">MLP</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.mlp.FusedGatedMLP"><code class="docutils literal notranslate"><span class="pre">FusedGatedMLP</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.mlp.FusedGatedMLP.fc_gate"><code class="docutils literal notranslate"><span class="pre">FusedGatedMLP.fc_gate()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.mlp.FusedGatedMLP.fc_gate_plugin"><code class="docutils literal notranslate"><span class="pre">FusedGatedMLP.fc_gate_plugin()</span></code></a></li>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.mlp.FusedGatedMLP.forward"><code class="docutils literal notranslate"><span class="pre">FusedGatedMLP.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.mlp.GatedMLP"><code class="docutils literal notranslate"><span class="pre">GatedMLP</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.mlp.GatedMLP.forward"><code class="docutils literal notranslate"><span class="pre">GatedMLP.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.mlp.MLP"><code class="docutils literal notranslate"><span class="pre">MLP</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.mlp.MLP.forward"><code class="docutils literal notranslate"><span class="pre">MLP.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.mlp.fc_gate_lora"><code class="docutils literal notranslate"><span class="pre">fc_gate_lora()</span></code></a></li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.normalization">Normalization</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.normalization.GroupNorm"><code class="docutils literal notranslate"><span class="pre">GroupNorm</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.normalization.GroupNorm.forward"><code class="docutils literal notranslate"><span class="pre">GroupNorm.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.normalization.LayerNorm"><code class="docutils literal notranslate"><span class="pre">LayerNorm</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.normalization.LayerNorm.forward"><code class="docutils literal notranslate"><span class="pre">LayerNorm.forward()</span></code></a></li>
</ul>
</li>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.normalization.RmsNorm"><code class="docutils literal notranslate"><span class="pre">RmsNorm</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.normalization.RmsNorm.forward"><code class="docutils literal notranslate"><span class="pre">RmsNorm.forward()</span></code></a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="#module-tensorrt_llm.layers.pooling">Pooling</a><ul>
<li class="toctree-l3"><a class="reference internal" href="#tensorrt_llm.layers.pooling.AvgPool2d"><code class="docutils literal notranslate"><span class="pre">AvgPool2d</span></code></a><ul>
<li class="toctree-l4"><a class="reference internal" href="#tensorrt_llm.layers.pooling.AvgPool2d.forward"><code class="docutils literal notranslate"><span class="pre">AvgPool2d.forward()</span></code></a></li>
</ul>
</li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="tensorrt_llm.functional.html">Functionals</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensorrt_llm.models.html">Models</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensorrt_llm.plugin.html">Plugin</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensorrt_llm.quantization.html">Quantization</a></li>
<li class="toctree-l1"><a class="reference internal" href="tensorrt_llm.runtime.html">Runtime</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">C++ API</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../_cpp_gen/executor.html">Executor</a></li>
<li class="toctree-l1"><a class="reference internal" href="../_cpp_gen/runtime.html">Runtime</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Command-Line Reference</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../commands/trtllm-build.html">trtllm-build</a></li>
<li class="toctree-l1"><a class="reference internal" href="../commands/trtllm-serve.html">trtllm-serve</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Architecture</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../architecture/overview.html">TensorRT-LLM Architecture</a></li>
<li class="toctree-l1"><a class="reference internal" href="../architecture/core-concepts.html">Model Definition</a></li>
<li class="toctree-l1"><a class="reference internal" href="../architecture/core-concepts.html#compilation">Compilation</a></li>
<li class="toctree-l1"><a class="reference internal" href="../architecture/core-concepts.html#runtime">Runtime</a></li>
<li class="toctree-l1"><a class="reference internal" href="../architecture/core-concepts.html#multi-gpu-and-multi-node-support">Multi-GPU and Multi-Node Support</a></li>
<li class="toctree-l1"><a class="reference internal" href="../architecture/checkpoint.html">TensorRT-LLM Checkpoint</a></li>
<li class="toctree-l1"><a class="reference internal" href="../architecture/workflow.html">TensorRT-LLM Build Workflow</a></li>
<li class="toctree-l1"><a class="reference internal" href="../architecture/add-model.html">Adding a Model</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Advanced</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../advanced/gpt-attention.html">Multi-Head, Multi-Query, and Group-Query Attention</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/gpt-runtime.html">C++ GPT Runtime</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/executor.html">Executor API</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/graph-rewriting.html">Graph Rewriting Module</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/inference-request.html">Inference Request</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/inference-request.html#responses">Responses</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/lora.html">Run gpt-2b + LoRA using GptManager / cpp runtime</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/expert-parallelism.html">Expert Parallelism in TensorRT-LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/kv-cache-reuse.html">KV cache reuse</a></li>
<li class="toctree-l1"><a class="reference internal" href="../advanced/speculative-decoding.html">Speculative Sampling</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Performance</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../performance/perf-overview.html">Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="../performance/perf-benchmarking.html">Benchmarking</a></li>
<li class="toctree-l1"><a class="reference internal" href="../performance/perf-best-practices.html">Best Practices</a></li>
<li class="toctree-l1"><a class="reference internal" href="../performance/perf-analysis.html">Performance Analysis</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Reference</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../reference/troubleshooting.html">Troubleshooting</a></li>
<li class="toctree-l1"><a class="reference internal" href="../reference/support-matrix.html">Support Matrix</a></li>
<li class="toctree-l1"><a class="reference internal" href="../reference/precision.html">Numerical Precision</a></li>
<li class="toctree-l1"><a class="reference internal" href="../reference/memory.html">Memory Usage of TensorRT-LLM</a></li>
</ul>
<p class="caption" role="heading"><span class="caption-text">Blogs</span></p>
<ul>
<li class="toctree-l1"><a class="reference internal" href="../blogs/H100vsA100.html">H100 has 4.6x A100 Performance in TensorRT-LLM, achieving 10,000 tok/s at 100ms to first token</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/H200launch.html">H200 achieves nearly 12,000 tokens/sec on Llama2-13B with TensorRT-LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/Falcon180B-H200.html">Falcon-180B on a single H200 GPU with INT4 AWQ, and 6.7x faster Llama-70B over A100</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/quantization-in-TRT-LLM.html">Speed up inference with SOTA quantization techniques in TRT-LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/XQA-kernel.html">New XQA-kernel provides 2.4x more Llama-70B throughput within the same latency budget</a></li>
</ul>
</div>
</div>
</nav>
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap"><nav class="wy-nav-top" aria-label="Mobile navigation menu" >
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
<a href="../index.html">tensorrt_llm</a>
</nav>
<div class="wy-nav-content">
<div class="rst-content">
<div role="navigation" aria-label="Page navigation">
<ul class="wy-breadcrumbs">
<li><a href="../index.html" class="icon icon-home" aria-label="Home"></a></li>
<li class="breadcrumb-item active">Layers</li>
<li class="wy-breadcrumbs-aside">
<a href="../_sources/python-api/tensorrt_llm.layers.rst.txt" rel="nofollow"> View page source</a>
</li>
</ul>
<hr/>
</div>
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
<div itemprop="articleBody">
<section id="module-tensorrt_llm">
<span id="layers"></span><h1>Layers<a class="headerlink" href="#module-tensorrt_llm" title="Link to this heading"></a></h1>
<section id="module-tensorrt_llm.layers.activation">
<span id="activation"></span><h2>Activation<a class="headerlink" href="#module-tensorrt_llm.layers.activation" title="Link to this heading"></a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.activation.Mish">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.activation.</span></span><span class="sig-name descname"><span class="pre">Mish</span></span><a class="reference internal" href="../_modules/tensorrt_llm/layers/activation.html#Mish"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.activation.Mish" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.activation.Mish.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">input</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/activation.html#Mish.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.activation.Mish.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
</section>
<section id="module-tensorrt_llm.layers.attention">
<span id="attention"></span><h2>Attention<a class="headerlink" href="#module-tensorrt_llm.layers.attention" title="Link to this heading"></a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.Attention">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">Attention</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">*</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">local_layer_idx</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_attention_heads</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_kv_heads=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_position_embeddings=1024</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_layers=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">apply_query_key_layer_scaling=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_head_size=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">qk_layernorm=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">layernorm_type=LayerNormType.LayerNorm</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">layernorm_share=True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">inner_layernorm=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">eps=1e-05</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_mask_type=AttentionMaskType.padding</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias=True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">position_embedding_type=PositionEmbeddingType.learned_absolute</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_base=10000.0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_scaling=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_percentage=1.0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rope_scaling_short_factors=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rope_scaling_long_factors=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rope_scaling_short_mscale=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rope_scaling_long_mscale=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">original_max_position_embeddings=1024</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_rank=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">quant_mode:</span> <span class="pre">~tensorrt_llm.quantization.mode.QuantMode</span> <span class="pre">=</span> <span class="pre">QuantMode.None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">q_scaling=1.0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cross_attention=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">relative_attention=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_distance=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_buckets=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dense_bias=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">clip_qkv=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">alibi_bias_max=8</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">skip_cross_kv=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_attn_value=0.0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">block_sparse_params=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_implicit_relative_attention=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">reorder=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">layer_idx_in_cache_pool=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">enable_qkv=True</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#Attention"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.Attention" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.Attention.create_attention_const_params">
<em class="property"><span class="pre">static</span><span class="w"> </span></em><span class="sig-name descname"><span class="pre">create_attention_const_params</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">model_cls</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">config</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#Attention.create_attention_const_params"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.Attention.create_attention_const_params" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.Attention.fill_attention_params">
<em class="property"><span class="pre">static</span><span class="w"> </span></em><span class="sig-name descname"><span class="pre">fill_attention_params</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">model_cls</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_params</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#Attention.fill_attention_params"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.Attention.fill_attention_params" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.Attention.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_mask</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_packed_mask</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_cache</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">spec_decoding_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">mrope_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">kv_cache_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">encoder_output</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">position_embedding</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">norm_before_bmm1</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_layer_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cross_kv_cache_gen</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cross_kv_reuse</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">reduce_fusion_params</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.AllReduceFusionParams" title="tensorrt_llm.functional.AllReduceFusionParams"><span class="pre">AllReduceFusionParams</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">skip_attn</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#Attention.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.Attention.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.Attention.postprocess">
<span class="sig-name descname"><span class="pre">postprocess</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">tllm_key</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">weights</span></span></em>, <em class="sig-param"><span class="o"><span class="pre">**</span></span><span class="n"><span class="pre">kwargs</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#Attention.postprocess"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.Attention.postprocess" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.Attention.set_rel_attn_table">
<span class="sig-name descname"><span class="pre">set_rel_attn_table</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">max_seq_len</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">precomputed_relative_attention</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#Attention.set_rel_attn_table"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.Attention.set_rel_attn_table" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.AttentionMaskParams">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">AttentionMaskParams</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">self_attention_mask</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">self_attention_packed_mask</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cross_attention_mask</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cross_attention_packed_mask</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#AttentionMaskParams"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.AttentionMaskParams" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">object</span></code></p>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.AttentionParams">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">AttentionParams</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">sequence_length</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">context_lengths</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_context_lengths</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_context_length</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_request_types</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">encoder_input_lengths</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">encoder_max_input_length</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_runtime_perf_knobs</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_context_progress</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#AttentionParams"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.AttentionParams" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">object</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.AttentionParams.fill_attention_const_params_for_long_rope">
<span class="sig-name descname"><span class="pre">fill_attention_const_params_for_long_rope</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">embed_positions_short_factors</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">embed_positions_long_factors</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">embed_positions_short_factors_for_attention_plugin</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">embed_positions_long_factors_for_attention_plugin</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">short_mscale</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">long_mscale</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">short_inv_freq</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">long_inv_freq</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#AttentionParams.fill_attention_const_params_for_long_rope"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.AttentionParams.fill_attention_const_params_for_long_rope" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.AttentionParams.fill_attention_const_params_for_rope">
<span class="sig-name descname"><span class="pre">fill_attention_const_params_for_rope</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">embed_positions</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_inv_freq</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">embed_positions_for_gpt_attention</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#AttentionParams.fill_attention_const_params_for_rope"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.AttentionParams.fill_attention_const_params_for_rope" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.AttentionParams.is_valid">
<span class="sig-name descname"><span class="pre">is_valid</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">gpt_attention_plugin</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">remove_input_padding</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_kv_cache</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#AttentionParams.is_valid"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.AttentionParams.is_valid" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.AttentionParams.is_valid_cross_attn">
<span class="sig-name descname"><span class="pre">is_valid_cross_attn</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">do_cross_attention</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#AttentionParams.is_valid_cross_attn"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.AttentionParams.is_valid_cross_attn" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.BertAttention">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">BertAttention</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_attention_heads</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_position_embeddings=1024</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_layers=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_head_size=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_kv_heads=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">q_scaling=1.0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">apply_query_key_layer_scaling=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias=True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_rank=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cp_group=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cp_size=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">relative_attention=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_distance=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_buckets=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">quant_mode=QuantMode.None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#BertAttention"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.BertAttention" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.BertAttention.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_mask</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">input_lengths</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_input_length</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_layer_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#BertAttention.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.BertAttention.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.BlockSparseAttnParams">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">BlockSparseAttnParams</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">block_size</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">64</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">homo_head_pattern</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_local_blocks</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">16</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">vertical_stride</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">8</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#BlockSparseAttnParams"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.BlockSparseAttnParams" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">object</span></code></p>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.CogVLMAttention">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">CogVLMAttention</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">*</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">local_layer_idx</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_attention_heads</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_kv_heads=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_position_embeddings=1024</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_mask_type=AttentionMaskType.causal</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias=True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">position_embedding_type=PositionEmbeddingType.learned_absolute</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_base=10000.0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_scaling=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_rank=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">quant_mode:</span> <span class="pre">~tensorrt_llm.quantization.mode.QuantMode</span> <span class="pre">=</span> <span class="pre">QuantMode.None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dense_bias=None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#CogVLMAttention"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.CogVLMAttention" title="Link to this definition"></a></dt>
<dd><p>Bases: <a class="reference internal" href="#tensorrt_llm.layers.attention.Attention" title="tensorrt_llm.layers.attention.Attention"><code class="xref py py-class docutils literal notranslate"><span class="pre">Attention</span></code></a></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.CogVLMAttention.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_cache</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">kv_cache_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">vision_token_mask</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">position_embedding</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#CogVLMAttention.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.CogVLMAttention.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.DeepseekV2Attention">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">DeepseekV2Attention</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">*</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">local_layer_idx</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_attention_heads</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">q_lora_rank</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">kv_lora_rank</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">qk_nope_head_dim=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">qk_rope_head_dim=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">v_head_dim=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">eps=1e-06</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_mask_type=AttentionMaskType.causal</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">position_embedding_type=PositionEmbeddingType.learned_absolute</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_position_embeddings=1024</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_base=10000.0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_scaling=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_beta_fast=32</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_beta_slow=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_mscale=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_mscale_all_dim=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_embedding_origin_max_position=4096</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rotary_scaling=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_rank=0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">quant_mode:</span> <span class="pre">~tensorrt_llm.quantization.mode.QuantMode</span> <span class="pre">=</span> <span class="pre">QuantMode.None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#DeepseekV2Attention"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.DeepseekV2Attention" title="Link to this definition"></a></dt>
<dd><p>Bases: <a class="reference internal" href="#tensorrt_llm.layers.attention.Attention" title="tensorrt_llm.layers.attention.Attention"><code class="xref py py-class docutils literal notranslate"><span class="pre">Attention</span></code></a></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.DeepseekV2Attention.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_cache</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">spec_decoding_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">kv_cache_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">attention_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#DeepseekV2Attention.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.DeepseekV2Attention.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.DeepseekV2Attention.weight_loader">
<span class="sig-name descname"><span class="pre">weight_loader</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">mapping</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Mapping</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">param</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Parameter</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">loaded_weight</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tensor</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#DeepseekV2Attention.weight_loader"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.DeepseekV2Attention.weight_loader" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.KeyValueCacheParams">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">KeyValueCacheParams</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">past_key_value</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">List</span><span class="p"><span class="pre">[</span></span><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a><span class="p"><span class="pre">]</span></span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_past_key_value_lengths</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_max_attention_window_sizes</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_sink_token_length</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">kv_cache_block_offsets</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_kv_cache_block_offsets</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_kv_cache_pool_pointers</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_kv_cache_pool_mapping</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cache_indirection</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">past_key_value_length</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">cross_kv_cache_block_offsets</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_cross_kv_cache_block_offsets</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_cross_kv_cache_pool_pointers</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">host_cross_kv_cache_pool_mapping</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#KeyValueCacheParams"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.KeyValueCacheParams" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">object</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.KeyValueCacheParams.fill_none_tensor_list">
<span class="sig-name descname"><span class="pre">fill_none_tensor_list</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">list_size</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#KeyValueCacheParams.fill_none_tensor_list"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.KeyValueCacheParams.fill_none_tensor_list" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.KeyValueCacheParams.get_first_past_key_value">
<span class="sig-name descname"><span class="pre">get_first_past_key_value</span></span><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#KeyValueCacheParams.get_first_past_key_value"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.KeyValueCacheParams.get_first_past_key_value" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.KeyValueCacheParams.is_valid">
<span class="sig-name descname"><span class="pre">is_valid</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">gpt_attention_plugin</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#KeyValueCacheParams.is_valid"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.KeyValueCacheParams.is_valid" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.MropeParams">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">MropeParams</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">mrope_rotary_sin_cos</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">mrope_position_deltas</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#MropeParams"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.MropeParams" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">object</span></code></p>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.SpecDecodingParams">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">SpecDecodingParams</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">spec_decoding_is_generation_length_variable</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">spec_decoding_max_generation_length</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">spec_decoding_generation_lengths</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">spec_decoding_position_offsets</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">spec_decoding_packed_mask</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#SpecDecodingParams"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.SpecDecodingParams" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">object</span></code></p>
</dd></dl>
<dl class="py function">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.compute_relative_bias">
<span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">compute_relative_bias</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">query_length</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">key_length</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_buckets</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">max_distance</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bidirectional</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">rel_attn_table</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_rank</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#compute_relative_bias"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.compute_relative_bias" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py function">
<dt class="sig sig-object py" id="tensorrt_llm.layers.attention.make_causal_mask">
<span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.attention.</span></span><span class="sig-name descname"><span class="pre">make_causal_mask</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">bsz</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tgt_len</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">past_key_values_length</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/attention.html#make_causal_mask"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.attention.make_causal_mask" title="Link to this definition"></a></dt>
<dd></dd></dl>
</section>
<section id="module-tensorrt_llm.layers.cast">
<span id="cast"></span><h2>Cast<a class="headerlink" href="#module-tensorrt_llm.layers.cast" title="Link to this heading"></a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.cast.Cast">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.cast.</span></span><span class="sig-name descname"><span class="pre">Cast</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">output_dtype</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">'float32'</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/cast.html#Cast"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.cast.Cast" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.cast.Cast.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/cast.html#Cast.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.cast.Cast.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
</section>
<section id="module-tensorrt_llm.layers.conv">
<span id="conv"></span><h2>Conv<a class="headerlink" href="#module-tensorrt_llm.layers.conv" title="Link to this heading"></a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.conv.Conv1d">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.conv.</span></span><span class="sig-name descname"><span class="pre">Conv1d</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">in_channels</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">out_channels</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">kernel_size</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">stride</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">padding</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dilation</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">groups</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">padding_mode</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">'zeros'</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/conv.html#Conv1d"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.conv.Conv1d" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.conv.Conv1d.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">input</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/conv.html#Conv1d.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.conv.Conv1d.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.conv.Conv2d">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.conv.</span></span><span class="sig-name descname"><span class="pre">Conv2d</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">in_channels</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">out_channels</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">kernel_size</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span></em>, <em class="sig-param"><span class="n"><span class="pre">stride</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">(1,</span> <span class="pre">1)</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">padding</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">(0,</span> <span class="pre">0)</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dilation</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">(1,</span> <span class="pre">1)</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">groups</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">padding_mode</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">'zeros'</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/conv.html#Conv2d"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.conv.Conv2d" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.conv.Conv2d.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">input</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/conv.html#Conv2d.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.conv.Conv2d.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.conv.ConvTranspose2d">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.conv.</span></span><span class="sig-name descname"><span class="pre">ConvTranspose2d</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">in_channels</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">out_channels</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">kernel_size</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span></em>, <em class="sig-param"><span class="n"><span class="pre">stride</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">(1,</span> <span class="pre">1)</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">padding</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">(0,</span> <span class="pre">0)</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">output_padding</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">(0,</span> <span class="pre">0)</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dilation</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">,</span></span><span class="w"> </span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">(1,</span> <span class="pre">1)</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">groups</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">padding_mode</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">'zeros'</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/conv.html#ConvTranspose2d"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.conv.ConvTranspose2d" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.conv.ConvTranspose2d.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">input</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">output_size</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/conv.html#ConvTranspose2d.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.conv.ConvTranspose2d.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
</section>
<section id="module-tensorrt_llm.layers.embedding">
<span id="embedding"></span><h2>Embedding<a class="headerlink" href="#module-tensorrt_llm.layers.embedding" title="Link to this heading"></a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.embedding.Embedding">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.embedding.</span></span><span class="sig-name descname"><span class="pre">Embedding</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">num_embeddings</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">embedding_dim</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">list</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">sharding_dim</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_rank</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">int</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/embedding.html#Embedding"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.embedding.Embedding" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<p>The embedding layer takes input indices (x) and the embedding lookup table (weight) as input.
And output the corresponding embeddings according to input indices.
The size of weight is [num_embeddings, embedding_dim]</p>
<p>Four parameters (tp_size, tp_group, sharding_dim, tp_rank) are involved in tensor parallelism.
Only when “tp_size &gt; 1 and tp_group is not None”, tensor parallelism is enabled.</p>
<blockquote>
<div><dl class="simple">
<dt>When “sharding_dim == 0”, the weight is shared in the vocabulary dimension.</dt><dd><p>tp_rank must be set when sharding_dim == 0.</p>
</dd>
</dl>
<p>When “sharding_dim == 1”, the weight is shard in the hidden dimension.</p>
</div></blockquote>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.embedding.Embedding.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/embedding.html#Embedding.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.embedding.Embedding.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.embedding.Embedding.postprocess">
<span class="sig-name descname"><span class="pre">postprocess</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">tllm_key</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">weights</span></span></em>, <em class="sig-param"><span class="o"><span class="pre">**</span></span><span class="n"><span class="pre">kwargs</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/embedding.html#Embedding.postprocess"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.embedding.Embedding.postprocess" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.embedding.Embedding.weight_loader">
<span class="sig-name descname"><span class="pre">weight_loader</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">mapping</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Mapping</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">param</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Parameter</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">loaded_weight</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tensor</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/embedding.html#Embedding.weight_loader"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.embedding.Embedding.weight_loader" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.embedding.PromptTuningEmbedding">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.embedding.</span></span><span class="sig-name descname"><span class="pre">PromptTuningEmbedding</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">num_embeddings</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">embedding_dim</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">vocab_size</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">sharding_dim</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_rank</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">0</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/embedding.html#PromptTuningEmbedding"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.embedding.PromptTuningEmbedding" title="Link to this definition"></a></dt>
<dd><p>Bases: <a class="reference internal" href="#tensorrt_llm.layers.embedding.Embedding" title="tensorrt_llm.layers.embedding.Embedding"><code class="xref py py-class docutils literal notranslate"><span class="pre">Embedding</span></code></a></p>
<p>PromptTuningEmbedding handles fine-tuned prompts with virtual tokens. At runtime,
a supplementary embedding dictionary is passed. Tokens whose ids are &gt;= vocab_size are embedded
with that additional dictionary.
The prompt tuning dictionary holds multiple tasks, and each sequence is assigned a given task.
Prompt-tuned tokens from a given sequence use the adequate task dictionary, as defined by the <cite>tasks</cite> input.</p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.embedding.PromptTuningEmbedding.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">tokens</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">prompt_embedding_table</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tasks</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">task_vocab_size</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/embedding.html#PromptTuningEmbedding.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.embedding.PromptTuningEmbedding.forward" title="Link to this definition"></a></dt>
<dd><blockquote>
<div><p>Pass all tokens through both normal and prompt embedding tables.
Tokens are masked so that “normal” embedding only see “normal” tokens. Same logic for “prompt” embedding.
After those two embedding, combine results based on whether the token was “normal” or “prompt-tuned”.</p>
</div></blockquote>
<dl class="field-list simple">
<dt class="field-odd">Parameters<span class="colon">:</span></dt>
<dd class="field-odd"><ul class="simple">
<li><p><strong>tokens</strong> – Tensor
the ids to embed, size [batch_size, seq_len]</p></li>
<li><p><strong>prompt_embedding_table</strong> – Tensor
the additional embedding table for prompt-tuned tokens, size [num_tasks * num_tokens_per_task, hidden_size]</p></li>
<li><p><strong>tasks</strong> – Tensor
the task required by each token, size [batch_size, seq_len]</p></li>
<li><p><strong>task_vocab_size</strong> – Tensor
the number of tokens used for each task, should be equal to prompt_embedding_table’s num_tokens_per_task, size [1]</p></li>
</ul>
</dd>
<dt class="field-even">Returns<span class="colon">:</span></dt>
<dd class="field-even"><p>Tokens’ embedding</p>
</dd>
</dl>
</dd></dl>
</dd></dl>
</section>
<section id="module-tensorrt_llm.layers.linear">
<span id="linear"></span><h2>Linear<a class="headerlink" href="#module-tensorrt_llm.layers.linear" title="Link to this heading"></a></h2>
<dl class="py attribute">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.ColumnLinear">
<span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.linear.</span></span><span class="sig-name descname"><span class="pre">ColumnLinear</span></span><a class="headerlink" href="#tensorrt_llm.layers.linear.ColumnLinear" title="Link to this definition"></a></dt>
<dd><p>alias of <a class="reference internal" href="#tensorrt_llm.layers.linear.Linear" title="tensorrt_llm.layers.linear.Linear"><code class="xref py py-class docutils literal notranslate"><span class="pre">Linear</span></code></a></p>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.Linear">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.linear.</span></span><span class="sig-name descname"><span class="pre">Linear</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">in_features</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">out_features</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">gather_output</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">share_weight</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">strict_dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">pad_lda</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">prefer_managed_weight</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">is_qkv</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">False</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#Linear"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.Linear" title="Link to this definition"></a></dt>
<dd><p>Bases: <a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase" title="tensorrt_llm.layers.linear.LinearBase"><code class="xref py py-class docutils literal notranslate"><span class="pre">LinearBase</span></code></a></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.Linear.collect_and_bias">
<span class="sig-name descname"><span class="pre">collect_and_bias</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em>, <em class="sig-param"><span class="o"><span class="pre">**</span></span><span class="n"><span class="pre">kwargs</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#Linear.collect_and_bias"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.Linear.collect_and_bias" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.Linear.postprocess">
<span class="sig-name descname"><span class="pre">postprocess</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">tllm_key</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">weights</span></span></em>, <em class="sig-param"><span class="o"><span class="pre">**</span></span><span class="n"><span class="pre">kwargs</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#Linear.postprocess"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.Linear.postprocess" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.Linear.tp_split_dim">
<em class="property"><span class="pre">classmethod</span><span class="w"> </span></em><span class="sig-name descname"><span class="pre">tp_split_dim</span></span><span class="sig-paren">(</span><span class="sig-paren">)</span> <span class="sig-return"><span class="sig-return-icon">&#x2192;</span> <span class="sig-return-typehint"><span class="pre">int</span></span></span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#Linear.tp_split_dim"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.Linear.tp_split_dim" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.linear.</span></span><span class="sig-name descname"><span class="pre">LinearBase</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">local_in_features</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">local_out_features</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">share_weight</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">strict_dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">pad_lda</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">prefer_managed_weight</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase.collect_and_bias">
<em class="property"><span class="pre">abstract</span><span class="w"> </span></em><span class="sig-name descname"><span class="pre">collect_and_bias</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span></em><span class="sig-paren">)</span> <span class="sig-return"><span class="sig-return-icon">&#x2192;</span> <span class="sig-return-typehint"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span></span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase.collect_and_bias"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase.collect_and_bias" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_runtime_params</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">LoraRuntimeParams</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_hidden_state</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="o"><span class="pre">**</span></span><span class="n"><span class="pre">kwargs</span></span></em><span class="sig-paren">)</span> <span class="sig-return"><span class="sig-return-icon">&#x2192;</span> <span class="sig-return-typehint"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span></span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase.get_weight">
<span class="sig-name descname"><span class="pre">get_weight</span></span><span class="sig-paren">(</span><span class="sig-paren">)</span> <span class="sig-return"><span class="sig-return-icon">&#x2192;</span> <span class="sig-return-typehint"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a></span></span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase.get_weight"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase.get_weight" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase.multiply_and_lora">
<span class="sig-name descname"><span class="pre">multiply_and_lora</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">weight</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">gemm_plugin</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">low_latency_gemm_plugin</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_fp8</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">alpha</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">ndarray</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_runtime_params</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">LoraRuntimeParams</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_hidden_state</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase.multiply_and_lora"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase.multiply_and_lora" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase.multiply_collect">
<span class="sig-name descname"><span class="pre">multiply_collect</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">weight</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">gemm_plugin</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">low_latency_gemm_plugin</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">str</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">use_fp8</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">alpha</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">ndarray</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_runtime_params</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">LoraRuntimeParams</span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_hidden_state</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.Tensor" title="tensorrt_llm.functional.Tensor"><span class="pre">Tensor</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="o"><span class="pre">**</span></span><span class="n"><span class="pre">kwargs</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase.multiply_collect"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase.multiply_collect" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase.tp_split_dim">
<em class="property"><span class="pre">abstract</span><span class="w"> </span><span class="pre">classmethod</span><span class="w"> </span></em><span class="sig-name descname"><span class="pre">tp_split_dim</span></span><span class="sig-paren">(</span><span class="sig-paren">)</span> <span class="sig-return"><span class="sig-return-icon">&#x2192;</span> <span class="sig-return-typehint"><span class="pre">int</span></span></span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase.tp_split_dim"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase.tp_split_dim" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase.weight_is_kn">
<span class="sig-name descname"><span class="pre">weight_is_kn</span></span><span class="sig-paren">(</span><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase.weight_is_kn"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase.weight_is_kn" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.LinearBase.weight_loader">
<span class="sig-name descname"><span class="pre">weight_loader</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">mapping</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Mapping</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">param</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Parameter</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">loaded_weight</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tensor</span></span></em><span class="sig-paren">)</span> <span class="sig-return"><span class="sig-return-icon">&#x2192;</span> <span class="sig-return-typehint"><span class="pre">None</span></span></span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#LinearBase.weight_loader"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.LinearBase.weight_loader" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.RowLinear">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.linear.</span></span><span class="sig-name descname"><span class="pre">RowLinear</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">in_features</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">out_features</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">strict_dtype</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">pad_lda</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">0</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">prefer_managed_weight</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">is_expert</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">False</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#RowLinear"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.RowLinear" title="Link to this definition"></a></dt>
<dd><p>Bases: <a class="reference internal" href="#tensorrt_llm.layers.linear.LinearBase" title="tensorrt_llm.layers.linear.LinearBase"><code class="xref py py-class docutils literal notranslate"><span class="pre">LinearBase</span></code></a></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.RowLinear.collect_and_bias">
<span class="sig-name descname"><span class="pre">collect_and_bias</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em>, <em class="sig-param"><span class="o"><span class="pre">**</span></span><span class="n"><span class="pre">kwargs</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#RowLinear.collect_and_bias"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.RowLinear.collect_and_bias" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.linear.RowLinear.tp_split_dim">
<em class="property"><span class="pre">classmethod</span><span class="w"> </span></em><span class="sig-name descname"><span class="pre">tp_split_dim</span></span><span class="sig-paren">(</span><span class="sig-paren">)</span> <span class="sig-return"><span class="sig-return-icon">&#x2192;</span> <span class="sig-return-typehint"><span class="pre">int</span></span></span><a class="reference internal" href="../_modules/tensorrt_llm/layers/linear.html#RowLinear.tp_split_dim"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.linear.RowLinear.tp_split_dim" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
</section>
<section id="module-tensorrt_llm.layers.mlp">
<span id="mlp"></span><h2>MLP<a class="headerlink" href="#module-tensorrt_llm.layers.mlp" title="Link to this heading"></a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.FusedGatedMLP">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.mlp.</span></span><span class="sig-name descname"><span class="pre">FusedGatedMLP</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">ffn_hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">hidden_act</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias=True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">quant_mode=QuantMode.None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">inner_layernorm=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">eps=1e-05</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">is_expert=False</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#FusedGatedMLP"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.FusedGatedMLP" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.FusedGatedMLP.fc_gate">
<span class="sig-name descname"><span class="pre">fc_gate</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_layer_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#FusedGatedMLP.fc_gate"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.FusedGatedMLP.fc_gate" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.FusedGatedMLP.fc_gate_plugin">
<span class="sig-name descname"><span class="pre">fc_gate_plugin</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_layer_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#FusedGatedMLP.fc_gate_plugin"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.FusedGatedMLP.fc_gate_plugin" title="Link to this definition"></a></dt>
<dd></dd></dl>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.FusedGatedMLP.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_layer_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">reduce_fusion_params</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.AllReduceFusionParams" title="tensorrt_llm.functional.AllReduceFusionParams"><span class="pre">AllReduceFusionParams</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#FusedGatedMLP.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.FusedGatedMLP.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.GatedMLP">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.mlp.</span></span><span class="sig-name descname"><span class="pre">GatedMLP</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">ffn_hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">hidden_act</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias=True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">quant_mode=QuantMode.None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">inner_layernorm=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">eps=1e-05</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">is_expert=False</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#GatedMLP"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.GatedMLP" title="Link to this definition"></a></dt>
<dd><p>Bases: <a class="reference internal" href="#tensorrt_llm.layers.mlp.MLP" title="tensorrt_llm.layers.mlp.MLP"><code class="xref py py-class docutils literal notranslate"><span class="pre">MLP</span></code></a></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.GatedMLP.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_layer_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">reduce_fusion_params</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><a class="reference internal" href="tensorrt_llm.functional.html#tensorrt_llm.functional.AllReduceFusionParams" title="tensorrt_llm.functional.AllReduceFusionParams"><span class="pre">AllReduceFusionParams</span></a><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#GatedMLP.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.GatedMLP.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.MLP">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.mlp.</span></span><span class="sig-name descname"><span class="pre">MLP</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">ffn_hidden_size</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">hidden_act</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias=True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_group=None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size=1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">quant_mode=QuantMode.None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">inner_layernorm=False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">eps=1e-05</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">is_expert=False</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#MLP"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.MLP" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.MLP.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_layer_params</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">gegelu_limit</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#MLP.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.MLP.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py function">
<dt class="sig sig-object py" id="tensorrt_llm.layers.mlp.fc_gate_lora">
<span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.mlp.</span></span><span class="sig-name descname"><span class="pre">fc_gate_lora</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">hidden_states</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">lora_layer_params</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/mlp.html#fc_gate_lora"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.mlp.fc_gate_lora" title="Link to this definition"></a></dt>
<dd></dd></dl>
</section>
<section id="module-tensorrt_llm.layers.normalization">
<span id="normalization"></span><h2>Normalization<a class="headerlink" href="#module-tensorrt_llm.layers.normalization" title="Link to this heading"></a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.normalization.GroupNorm">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.normalization.</span></span><span class="sig-name descname"><span class="pre">GroupNorm</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">num_groups</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_channels</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">eps</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1e-05</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">affine</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/normalization.html#GroupNorm"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.normalization.GroupNorm" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.normalization.GroupNorm.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/normalization.html#GroupNorm.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.normalization.GroupNorm.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.normalization.LayerNorm">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.normalization.</span></span><span class="sig-name descname"><span class="pre">LayerNorm</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">normalized_shape</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">eps</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1e-05</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">elementwise_affine</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">bias</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_size</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">tp_dim</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">-1</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/normalization.html#LayerNorm"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.normalization.LayerNorm" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.normalization.LayerNorm.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">normalized_shape</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/normalization.html#LayerNorm.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.normalization.LayerNorm.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.normalization.RmsNorm">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.normalization.</span></span><span class="sig-name descname"><span class="pre">RmsNorm</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">normalized_shape</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">num_groups</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">eps</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">1e-06</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">elementwise_affine</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">True</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">dtype</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/normalization.html#RmsNorm"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.normalization.RmsNorm" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.normalization.RmsNorm.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">x</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">normalized_shape</span></span><span class="o"><span class="pre">=</span></span><span class="default_value"><span class="pre">None</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/normalization.html#RmsNorm.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.normalization.RmsNorm.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
</section>
<section id="module-tensorrt_llm.layers.pooling">
<span id="pooling"></span><h2>Pooling<a class="headerlink" href="#module-tensorrt_llm.layers.pooling" title="Link to this heading"></a></h2>
<dl class="py class">
<dt class="sig sig-object py" id="tensorrt_llm.layers.pooling.AvgPool2d">
<em class="property"><span class="pre">class</span><span class="w"> </span></em><span class="sig-prename descclassname"><span class="pre">tensorrt_llm.layers.pooling.</span></span><span class="sig-name descname"><span class="pre">AvgPool2d</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">kernel_size</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">]</span></span></span></em>, <em class="sig-param"><span class="n"><span class="pre">stride</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">]</span></span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">None</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">padding</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">Tuple</span><span class="p"><span class="pre">[</span></span><span class="pre">int</span><span class="p"><span class="pre">]</span></span><span class="w"> </span><span class="p"><span class="pre">|</span></span><span class="w"> </span><span class="pre">None</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">(0,</span> <span class="pre">0)</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">ceil_mode</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">False</span></span></em>, <em class="sig-param"><span class="n"><span class="pre">count_include_pad</span></span><span class="p"><span class="pre">:</span></span><span class="w"> </span><span class="n"><span class="pre">bool</span></span><span class="w"> </span><span class="o"><span class="pre">=</span></span><span class="w"> </span><span class="default_value"><span class="pre">True</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/pooling.html#AvgPool2d"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.pooling.AvgPool2d" title="Link to this definition"></a></dt>
<dd><p>Bases: <code class="xref py py-class docutils literal notranslate"><span class="pre">Module</span></code></p>
<dl class="py method">
<dt class="sig sig-object py" id="tensorrt_llm.layers.pooling.AvgPool2d.forward">
<span class="sig-name descname"><span class="pre">forward</span></span><span class="sig-paren">(</span><em class="sig-param"><span class="n"><span class="pre">input</span></span></em><span class="sig-paren">)</span><a class="reference internal" href="../_modules/tensorrt_llm/layers/pooling.html#AvgPool2d.forward"><span class="viewcode-link"><span class="pre">[source]</span></span></a><a class="headerlink" href="#tensorrt_llm.layers.pooling.AvgPool2d.forward" title="Link to this definition"></a></dt>
<dd></dd></dl>
</dd></dl>
</section>
</section>
</div>
</div>
<footer><div class="rst-footer-buttons" role="navigation" aria-label="Footer">
<a href="../llm-api-examples/llm_multilora.html" class="btn btn-neutral float-left" title="Generate text with multiple LoRA adapters" accesskey="p" rel="prev"><span class="fa fa-arrow-circle-left" aria-hidden="true"></span> Previous</a>
<a href="tensorrt_llm.functional.html" class="btn btn-neutral float-right" title="Functionals" accesskey="n" rel="next">Next <span class="fa fa-arrow-circle-right" aria-hidden="true"></span></a>
</div>
<hr/>
<div role="contentinfo">
<jinja2.runtime.BlockReference object at 0x7f9458605c10>
<div class="footer">
<p>
Copyright © 2024 NVIDIA Corporation
</p>
<p>
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/privacy-policy/" target="_blank" rel="noopener"
data-cms-ai="0">Privacy Policy</a> |
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/privacy-center/" target="_blank" rel="noopener"
data-cms-ai="0">Manage My Privacy</a> |
<a class="Link" href="https://www.nvidia.com/en-us/preferences/start/" target="_blank" rel="noopener"
data-cms-ai="0">Do Not Sell or Share My Data</a> |
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/terms-of-service/" target="_blank"
rel="noopener" data-cms-ai="0">Terms of Service</a> |
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/accessibility/" target="_blank" rel="noopener"
data-cms-ai="0">Accessibility</a> |
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/company-policies/" target="_blank"
rel="noopener" data-cms-ai="0">Corporate Policies</a> |
<a class="Link" href="https://www.nvidia.com/en-us/product-security/" target="_blank" rel="noopener"
data-cms-ai="0">Product Security</a> |
<a class="Link" href="https://www.nvidia.com/en-us/contact/" target="_blank" rel="noopener"
data-cms-ai="0">Contact</a>
</p>
</div>
</div>
</footer>
</div>
</div>
</section>
</div>
<script>
jQuery(function () {
SphinxRtdTheme.Navigation.enable(true);
});
</script>
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink