kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-01-28 06:33:15 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
7c3dbd3928
TensorRT-LLMs/1.0.0/advanced/lora.html
Kaiyu Xie 090dc64986 Update GitHub pages to v1.0.0
2025-09-24 02:05:54 +00:00

830 lines
54 KiB
HTML
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<!DOCTYPE html>
<html lang="en" data-content_root="../" >
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" /><meta name="viewport" content="width=device-width, initial-scale=1" />
<title>Run gpt-2b + LoRA using Executor / cpp runtime &#8212; TensorRT LLM</title>
<script data-cfasync="false">
document.documentElement.dataset.mode = localStorage.getItem("mode") || "";
document.documentElement.dataset.theme = localStorage.getItem("theme") || "";
</script>
<!--
this give us a css class that will be invisible only if js is disabled
-->
<noscript>
<style>
.pst-js-only { display: none !important; }
</style>
</noscript>
<!-- Loaded before other Sphinx assets -->
<link href="../_static/styles/theme.css?digest=8878045cc6db502f8baf" rel="stylesheet" />
<link href="../_static/styles/pydata-sphinx-theme.css?digest=8878045cc6db502f8baf" rel="stylesheet" />
<link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=8f2a1f02" />
<link rel="stylesheet" type="text/css" href="../_static/styles/nvidia-sphinx-theme.css?v=df3ac72c" />
<link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
<link rel="stylesheet" type="text/css" href="../_static/autodoc_pydantic.css" />
<link rel="stylesheet" type="text/css" href="../_static/togglebutton.css?v=13237357" />
<link rel="stylesheet" type="text/css" href="../_static/custom.css?v=19d20f17" />
<!-- So that users can add custom icons -->
<script src="../_static/scripts/fontawesome.js?digest=8878045cc6db502f8baf"></script>
<!-- Pre-loaded scripts that we'll load fully later -->
<link rel="preload" as="script" href="../_static/scripts/bootstrap.js?digest=8878045cc6db502f8baf" />
<link rel="preload" as="script" href="../_static/scripts/pydata-sphinx-theme.js?digest=8878045cc6db502f8baf" />
<script src="../_static/documentation_options.js?v=5929fcd5"></script>
<script src="../_static/doctools.js?v=9a2dae69"></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>let toggleHintShow = 'Click to show';</script>
<script>let toggleHintHide = 'Click to hide';</script>
<script>let toggleOpenOnPrint = 'true';</script>
<script src="../_static/togglebutton.js?v=4a39c7ea"></script>
<script>var togglebuttonSelector = '.toggle, .admonition.dropdown';</script>
<script>var togglebuttonSelector = '.toggle, .admonition.dropdown';</script>
<script>DOCUMENTATION_OPTIONS.pagename = 'advanced/lora';</script>
<script>
DOCUMENTATION_OPTIONS.theme_version = '0.16.1';
DOCUMENTATION_OPTIONS.theme_switcher_json_url = './_static/switcher.json';
DOCUMENTATION_OPTIONS.theme_switcher_version_match = '1.0.0';
DOCUMENTATION_OPTIONS.show_version_warning_banner =
false;
</script>
<link rel="icon" href="../_static/favicon.png"/>
<link rel="index" title="Index" href="../genindex.html" />
<link rel="search" title="Search" href="../search.html" />
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<meta name="docsearch:language" content="en"/>
<meta name="docsearch:version" content="1.0.0" />
</head>
<body data-bs-spy="scroll" data-bs-target=".bd-toc-nav" data-offset="180" data-bs-root-margin="0px 0px -60%" data-default-mode="">
<div id="pst-skip-link" class="skip-link d-print-none"><a href="#main-content">Skip to main content</a></div>
<div id="pst-scroll-pixel-helper"></div>
<button type="button" class="btn rounded-pill" id="pst-back-to-top">
<i class="fa-solid fa-arrow-up"></i>Back to top</button>
<dialog id="pst-search-dialog">
<form class="bd-search d-flex align-items-center"
action="../search.html"
method="get">
<i class="fa-solid fa-magnifying-glass"></i>
<input type="search"
class="form-control"
name="q"
placeholder="Search the docs ..."
aria-label="Search the docs ..."
autocomplete="off"
autocorrect="off"
autocapitalize="off"
spellcheck="false"/>
<span class="search-button__kbd-shortcut"><kbd class="kbd-shortcut__modifier">Ctrl</kbd>+<kbd>K</kbd></span>
</form>
</dialog>
<div class="pst-async-banner-revealer d-none">
<aside id="bd-header-version-warning" class="d-none d-print-none" aria-label="Version warning"></aside>
</div>
<header class="bd-header navbar navbar-expand-lg bd-navbar d-print-none">
<div class="bd-header__inner bd-page-width">
<button class="pst-navbar-icon sidebar-toggle primary-toggle" aria-label="Site navigation">
<span class="fa-solid fa-bars"></span>
</button>
<div class="col-lg-3 navbar-header-items__start">
<div class="navbar-item">
<a class="navbar-brand logo" href="../index.html">
<img src="../_static/nvidia-logo-horiz-rgb-blk-for-screen.svg" class="logo__image only-light" alt="TensorRT LLM - Home"/>
<img src="../_static/nvidia-logo-horiz-rgb-wht-for-screen.svg" class="logo__image only-dark pst-js-only" alt="TensorRT LLM - Home"/>
<p class="title logo__title">TensorRT LLM</p>
</a></div>
</div>
<div class="col-lg-9 navbar-header-items">
<div class="me-auto navbar-header-items__center">
<div class="navbar-item">
<div class="version-switcher__container dropdown pst-js-only">
<button id="pst-version-switcher-button-2"
type="button"
class="version-switcher__button btn btn-sm dropdown-toggle"
data-bs-toggle="dropdown"
aria-haspopup="listbox"
aria-controls="pst-version-switcher-list-2"
aria-label="Version switcher list"
>
Choose version <!-- this text may get changed later by javascript -->
<span class="caret"></span>
</button>
<div id="pst-version-switcher-list-2"
class="version-switcher__menu dropdown-menu list-group-flush py-0"
role="listbox" aria-labelledby="pst-version-switcher-button-2">
<!-- dropdown will be populated by javascript on page load -->
</div>
</div></div>
</div>
<div class="navbar-header-items__end">
<div class="navbar-item navbar-persistent--container">
<button class="btn search-button-field search-button__button pst-js-only" title="Search" aria-label="Search" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="fa-solid fa-magnifying-glass"></i>
<span class="search-button__default-text">Search</span>
<span class="search-button__kbd-shortcut"><kbd class="kbd-shortcut__modifier">Ctrl</kbd>+<kbd class="kbd-shortcut__modifier">K</kbd></span>
</button>
</div>
<div class="navbar-item">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button pst-js-only" aria-label="Color mode" data-bs-title="Color mode" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="theme-switch fa-solid fa-sun fa-lg" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg" data-mode="auto" title="System Settings"></i>
</button></div>
</div>
</div>
<div class="navbar-persistent--mobile">
<button class="btn search-button-field search-button__button pst-js-only" title="Search" aria-label="Search" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="fa-solid fa-magnifying-glass"></i>
<span class="search-button__default-text">Search</span>
<span class="search-button__kbd-shortcut"><kbd class="kbd-shortcut__modifier">Ctrl</kbd>+<kbd class="kbd-shortcut__modifier">K</kbd></span>
</button>
</div>
<button class="pst-navbar-icon sidebar-toggle secondary-toggle" aria-label="On this page">
<span class="fa-solid fa-outdent"></span>
</button>
</div>
</header>
<div class="bd-container">
<div class="bd-container__inner bd-page-width">
<dialog id="pst-primary-sidebar-modal"></dialog>
<div id="pst-primary-sidebar" class="bd-sidebar-primary bd-sidebar">
<a class="navbar-brand logo" href="../index.html">
<img src="../_static/nvidia-logo-horiz-rgb-blk-for-screen.svg" class="logo__image only-light" alt="TensorRT LLM - Home"/>
<img src="../_static/nvidia-logo-horiz-rgb-wht-for-screen.svg" class="logo__image only-dark pst-js-only" alt="TensorRT LLM - Home"/>
<p class="title logo__title">TensorRT LLM</p>
</a>
<div class="sidebar-header-items sidebar-primary__section">
<div class="sidebar-header-items__center">
<div class="navbar-item">
<div class="version-switcher__container dropdown pst-js-only">
<button id="pst-version-switcher-button-3"
type="button"
class="version-switcher__button btn btn-sm dropdown-toggle"
data-bs-toggle="dropdown"
aria-haspopup="listbox"
aria-controls="pst-version-switcher-list-3"
aria-label="Version switcher list"
>
Choose version <!-- this text may get changed later by javascript -->
<span class="caret"></span>
</button>
<div id="pst-version-switcher-list-3"
class="version-switcher__menu dropdown-menu list-group-flush py-0"
role="listbox" aria-labelledby="pst-version-switcher-button-3">
<!-- dropdown will be populated by javascript on page load -->
</div>
</div></div>
</div>
<div class="sidebar-header-items__end">
<div class="navbar-item">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button pst-js-only" aria-label="Color mode" data-bs-title="Color mode" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="theme-switch fa-solid fa-sun fa-lg" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg" data-mode="auto" title="System Settings"></i>
</button></div>
</div>
</div>
<div class="sidebar-primary-items__start sidebar-primary__section">
<div class="sidebar-primary-item">
<nav class="bd-docs-nav bd-links"
aria-label="Table of Contents">
<p class="bd-links__title" role="heading" aria-level="1">Table of Contents</p>
<div class="bd-toc-item navbar-nav"><p aria-level="2" class="caption" role="heading"><span class="caption-text">Getting Started</span></p>
<ul class="nav bd-sidenav">
<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 has-children"><a class="reference internal" href="../installation/index.html">Installation</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="../installation/containers.html">Pre-built release container images on NGC</a></li>
<li class="toctree-l2"><a class="reference internal" href="../installation/linux.html">Installing on Linux via <code class="docutils literal notranslate"><span class="pre">pip</span></code></a></li>
<li class="toctree-l2"><a class="reference internal" href="../installation/build-from-source-linux.html">Building from Source Code on Linux</a></li>
</ul>
</details></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Deployment Guide</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1 has-children"><a class="reference internal" href="../examples/llm_api_examples.html">LLM Examples</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_inference.html">Generate text</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_inference_async.html">Generate text asynchronously</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_inference_async_streaming.html">Generate text in streaming</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_inference_distributed.html">Distributed LLM Generation</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_guided_decoding.html">Generate text with guided decoding</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_logits_processor.html">Control generated text using logits processor</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_multilora.html">Generate text with multiple LoRA adapters</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_speculative_decoding.html">Speculative Decoding</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_runtime.html">Runtime Configuration Examples</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_sampling.html">Sampling Techniques Showcase</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_mgmn_llm_distributed.html">Run LLM-API with pytorch backend on Slurm</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_mgmn_trtllm_bench.html">Run trtllm-bench with pytorch backend on Slurm</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/llm_mgmn_trtllm_serve.html">Run trtllm-serve with pytorch backend on Slurm</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../examples/trtllm_serve_examples.html">Online Serving Examples</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="../examples/curl_chat_client.html">Curl Chat Client</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/curl_chat_client_for_multimodal.html">Curl Chat Client For Multimodal</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/curl_completion_client.html">Curl Completion Client</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/deepseek_r1_reasoning_parser.html">Deepseek R1 Reasoning Parser</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/genai_perf_client.html">Genai Perf Client</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/genai_perf_client_for_multimodal.html">Genai Perf Client For Multimodal</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/openai_chat_client.html">OpenAI Chat Client</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/openai_chat_client_for_multimodal.html">OpenAI Chat Client for Multimodal</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/openai_completion_client.html">OpenAI Completion Client</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/openai_completion_client_for_lora.html">Openai Completion Client For Lora</a></li>
<li class="toctree-l2"><a class="reference internal" href="../examples/openai_completion_client_json_schema.html">OpenAI Completion Client with JSON Schema</a></li>
</ul>
</details></li>
<li class="toctree-l1"><a class="reference internal" href="../examples/dynamo_k8s_example.html">Dynamo K8s Example</a></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../deployment-guide/index.html">Model Recipes</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="../deployment-guide/quick-start-recipe-for-deepseek-r1-on-trtllm.html">Quick Start Recipe for DeepSeek R1 on TensorRT LLM - Blackwell &amp; Hopper Hardware</a></li>
<li class="toctree-l2"><a class="reference internal" href="../deployment-guide/quick-start-recipe-for-llama3.3-70b-on-trtllm.html">Quick Start Recipe for Llama3.3 70B on TensorRT LLM - Blackwell &amp; Hopper Hardware</a></li>
<li class="toctree-l2"><a class="reference internal" href="../deployment-guide/quick-start-recipe-for-llama4-scout-on-trtllm.html">Quick Start Recipe for Llama4 Scout 17B on TensorRT LLM - Blackwell &amp; Hopper Hardware</a></li>
</ul>
</details></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Models</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../models/supported-models.html">Supported Models</a></li>
<li class="toctree-l1"><a class="reference internal" href="../models/adding-new-model.html">Adding a New Model</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">CLI Reference</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../commands/trtllm-bench.html">trtllm-bench</a></li>
<li class="toctree-l1"><a class="reference internal" href="../commands/trtllm-eval.html">trtllm-eval</a></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../commands/trtllm-serve/index.html">trtllm-serve</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="../commands/trtllm-serve/trtllm-serve.html">trtllm-serve</a></li>
<li class="toctree-l2"><a class="reference internal" href="../commands/trtllm-serve/run-benchmark-with-trtllm-serve.html">Run benchmarking with <code class="docutils literal notranslate"><span class="pre">trtllm-serve</span></code></a></li>
</ul>
</details></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">API Reference</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../llm-api/index.html">LLM API Introduction</a></li>
<li class="toctree-l1"><a class="reference internal" href="../llm-api/reference.html">API Reference</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Features</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../features/feature-combination-matrix.html">Feature Combination Matrix</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/attention.html">Multi-Head, Multi-Query, and Group-Query Attention</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/disagg-serving.html">Disaggregated Serving (Beta)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/kvcache.html">KV Cache System</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/long-sequence.html">Long Sequences</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/lora.html">LoRA (Low-Rank Adaptation)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/multi-modality.html">Multimodal Support in TensorRT LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/overlap-scheduler.html">Overlap Scheduler</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/paged-attention-ifb-scheduler.html">Paged Attention, IFB, and Request Scheduling</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/parallel-strategy.html">Parallelism in TensorRT LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/quantization.html">Quantization</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/sampling.html">Sampling</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/speculative-decoding.html">Speculative Decoding</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/checkpoint-loading.html">Checkpoint Loading</a></li>
<li class="toctree-l1"><a class="reference internal" href="../features/auto_deploy/auto-deploy.html">AutoDeploy (Prototype)</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Developer Guide</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../architecture/overview.html">Architecture Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="../developer-guide/perf-analysis.html">Performance Analysis</a></li>
<li class="toctree-l1"><a class="reference internal" href="../developer-guide/perf-benchmarking.html">TensorRT LLM Benchmarking</a></li>
<li class="toctree-l1"><a class="reference internal" href="../developer-guide/ci-overview.html">Continuous Integration Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="../developer-guide/dev-containers.html">Using Dev Containers</a></li>
<li class="toctree-l1"><a class="reference internal" href="../developer-guide/api-change.html">LLM API Change Guide</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Blogs</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog1_Pushing_Latency_Boundaries_Optimizing_DeepSeek-R1_Performance_on_NVIDIA_B200_GPUs.html">Pushing Latency Boundaries: Optimizing DeepSeek-R1 Performance on NVIDIA B200 GPUs</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog2_DeepSeek_R1_MTP_Implementation_and_Optimization.html">DeepSeek R1 MTP Implementation and Optimization</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog3_Optimizing_DeepSeek_R1_Throughput_on_NVIDIA_Blackwell_GPUs.html">Optimizing DeepSeek R1 Throughput on NVIDIA Blackwell GPUs: A Deep Dive for Developers</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog4_Scaling_Expert_Parallelism_in_TensorRT-LLM.html">Scaling Expert Parallelism in TensorRT LLM (Part 1: Design and Implementation of Large-scale EP)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog5_Disaggregated_Serving_in_TensorRT-LLM.html">Disaggregated Serving in TensorRT LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog6_Llama4_maverick_eagle_guide.html">How to launch Llama4 Maverick + Eagle3 TensorRT LLM server</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog7_NGram_performance_Analysis_And_Auto_Enablement.html">N-Gram Speculative Decoding in TensorRT LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog8_Scaling_Expert_Parallelism_in_TensorRT-LLM_part2.html">Scaling Expert Parallelism in TensorRT LLM (Part 2: Performance Status and Optimization)</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog9_Deploying_GPT_OSS_on_TRTLLM.html">Running a High Performance GPT-OSS-120B Inference Server with TensorRT LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/Best_perf_practice_on_DeepSeek-R1_in_TensorRT-LLM.html">How to get best performance on DeepSeek-R1 in TensorRT LLM</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/XQA-kernel.html">New XQA-kernel provides 2.4x more Llama-70B throughput within the same latency budget</a></li>
<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>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Quick Links</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference external" href="https://github.com/NVIDIA/TensorRT-LLM/releases">Releases</a></li>
<li class="toctree-l1"><a class="reference external" href="https://github.com/NVIDIA/TensorRT-LLM">Github Code</a></li>
<li class="toctree-l1"><a class="reference external" href="https://github.com/NVIDIA/TensorRT-LLM/issues?q=is%3Aissue%20state%3Aopen%20label%3Aroadmap">Roadmap</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Use TensorRT Engine</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../legacy/tensorrt_quickstart.html">LLM API with TensorRT Engine</a></li>
</ul>
</div>
</nav></div>
</div>
<div class="sidebar-primary-items__end sidebar-primary__section">
</div>
</div>
<main id="main-content" class="bd-main" role="main">
<div class="bd-content">
<div class="bd-article-container">
<div class="bd-header-article d-print-none">
<div class="header-article-items header-article__inner">
<div class="header-article-items__start">
<div class="header-article-item">
<nav aria-label="Breadcrumb" class="d-print-none">
<ul class="bd-breadcrumbs">
<li class="breadcrumb-item breadcrumb-home">
<a href="../index.html" class="nav-link" aria-label="Home">
<i class="fa-solid fa-home"></i>
</a>
</li>
<li class="breadcrumb-item active" aria-current="page"><span class="ellipsis">Run gpt-2b + LoRA using Executor / cpp runtime</span></li>
</ul>
</nav>
</div>
</div>
</div>
</div>
<div id="searchbox"></div>
<article class="bd-article">
<section id="run-gpt-2b-lora-using-executor-cpp-runtime">
<span id="lora"></span><h1>Run gpt-2b + LoRA using Executor / cpp runtime<a class="headerlink" href="#run-gpt-2b-lora-using-executor-cpp-runtime" title="Link to this heading">#</a></h1>
<p>First build a model with LoRA and inflight-batching enabled.</p>
<div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>git-lfs<span class="w"> </span>clone<span class="w"> </span>https://huggingface.co/qychen/luotuo-lora-7b-0.1
git-lfs<span class="w"> </span>clone<span class="w"> </span>https://huggingface.co/kunishou/Japanese-Alpaca-LoRA-7b-v0
<span class="nv">BASE_MODEL</span><span class="o">=</span>llama-7b-hf
python<span class="w"> </span>examples/models/core/llama/convert_checkpoint.py<span class="w"> </span>--model_dir<span class="w"> </span><span class="si">${</span><span class="nv">BASE_MODEL</span><span class="si">}</span><span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--output_dir<span class="w"> </span>/tmp/llama_7b/trt_ckpt/fp16/1-gpu/<span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--dtype<span class="w"> </span>float16
trtllm-build<span class="w"> </span>--checkpoint_dir<span class="w"> </span>/tmp/llama_7b/trt_ckpt/fp16/1-gpu/<span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--output_dir<span class="w"> </span>/tmp/llama_7b_with_lora_qkv/trt_engines/fp16/1-gpu/<span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--remove_input_padding<span class="w"> </span><span class="nb">enable</span><span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--gpt_attention_plugin<span class="w"> </span>float16<span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--context_fmha<span class="w"> </span><span class="nb">enable</span><span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--paged_kv_cache<span class="w"> </span><span class="nb">enable</span><span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--gemm_plugin<span class="w"> </span>float16<span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--lora_plugin<span class="w"> </span>float16<span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--max_batch_size<span class="w"> </span><span class="m">128</span><span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--max_input_len<span class="w"> </span><span class="m">512</span><span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--max_seq_len<span class="w"> </span><span class="m">562</span><span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--lora_dir<span class="w"> </span>Japanese-Alpaca-LoRA-7b-v0<span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--max_lora_rank<span class="w"> </span><span class="m">8</span><span class="w"> </span><span class="se">\</span>
<span class="w"> </span>--lora_target_modules<span class="w"> </span><span class="s2">&quot;attn_q&quot;</span><span class="w"> </span><span class="s2">&quot;attn_k&quot;</span><span class="w"> </span><span class="s2">&quot;attn_v&quot;</span>
</pre></div>
</div>
<p>To pass LoRAs into the cpp runtime they must be converted to the format below.
The script below will convert a Hugging Face LoRA model to the correct NumPy tensor.</p>
<div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>python3<span class="w"> </span>tensorrt_llm/examples/hf_lora_convert.py<span class="w"> </span>-i<span class="w"> </span>Japanese-Alpaca-LoRA-7b-v0<span class="w"> </span>-o<span class="w"> </span>Japanese-Alpaca-LoRA-7b-v0-weights<span class="w"> </span>--storage-type<span class="w"> </span>float16
python3<span class="w"> </span>tensorrt_llm/examples/hf_lora_convert.py<span class="w"> </span>-i<span class="w"> </span>luotuo-lora-7b-0.1<span class="w"> </span>-o<span class="w"> </span>luotuo-lora-7b-0.1-weights<span class="w"> </span>--storage-type<span class="w"> </span>float16
</pre></div>
</div>
<p>Refer to the <a class="reference external" href="https://github.com/triton-inference-server/tensorrtllm_backend/blob/main/docs/lora.md">tensorrtllm_backend documentation</a> for a Multi-LoRA example using Triton.</p>
<section id="lora-tensor-format-details">
<h2>LoRA tensor format details<a class="headerlink" href="#lora-tensor-format-details" title="Link to this heading">#</a></h2>
<p>To run inference using <code class="docutils literal notranslate"><span class="pre">Executor</span></code>, a <code class="docutils literal notranslate"><span class="pre">Request</span></code> must have a <code class="docutils literal notranslate"><span class="pre">LoraConfig</span></code> that contains a <code class="docutils literal notranslate"><span class="pre">task_id</span></code>, <code class="docutils literal notranslate"><span class="pre">weights</span></code> and <code class="docutils literal notranslate"><span class="pre">config</span></code> parameters.</p>
<p><code class="docutils literal notranslate"><span class="pre">task_id</span></code> the unique task ID for the given LoRA.</p>
<p>To perform inference with a specific LoRA for the first time, <code class="docutils literal notranslate"><span class="pre">task_id</span></code>, <code class="docutils literal notranslate"><span class="pre">weights</span></code>, and <code class="docutils literal notranslate"><span class="pre">config</span></code> must all be given. The LoRA will be cached, so that subsequent requests for the same task only require <code class="docutils literal notranslate"><span class="pre">task_id</span></code>.
If the cache is full, the oldest LoRA will be evicted to make space for new ones. An error is returned if <code class="docutils literal notranslate"><span class="pre">task_id</span></code> is not cached.</p>
<p><code class="docutils literal notranslate"><span class="pre">weights</span></code> contains the weights for all the LoRAs. Currently, this should include weights for all TP and PP ranks.
The weights tensor has the shape <code class="docutils literal notranslate"><span class="pre">[num_lora_modules_layers,</span> <span class="pre">D</span> <span class="pre">x</span> <span class="pre">Hi</span> <span class="pre">+</span> <span class="pre">Ho</span> <span class="pre">x</span> <span class="pre">D</span> <span class="pre">]</span></code>. The last dimension holds the in / out adapter weights for the associated module (for example, <code class="docutils literal notranslate"><span class="pre">attn_qkv</span></code>) and model layer.</p>
<p>Each of the in / out tensors are first flattened and then concatenated together in the format above.
The first dimension (of size <code class="docutils literal notranslate"><span class="pre">num_lora_module_layers</span></code>) has an entry for each module-layer (that is, there is an entry for <code class="docutils literal notranslate"><span class="pre">attn_q</span> <span class="pre">layer1</span></code> and another for <code class="docutils literal notranslate"><span class="pre">attn_k</span> <span class="pre">layer1</span></code>).</p>
<p><code class="docutils literal notranslate"><span class="pre">D=adapter_size</span> <span class="pre">(i.e.</span> <span class="pre">R</span> <span class="pre">value),</span> <span class="pre">Hi=hidden_size_in,</span> <span class="pre">Ho=hidden_size_out.</span></code></p>
<p><code class="docutils literal notranslate"><span class="pre">config</span></code> is a configuration tensor which identifies the moduleId, layerId, and adapter size of each element of <code class="docutils literal notranslate"><span class="pre">LoraWeights</span></code>. It has the shape <code class="docutils literal notranslate"><span class="pre">[num_lora_modules_layers,</span> <span class="pre">3]</span></code>. The last dimension holds <code class="docutils literal notranslate"><span class="pre">[module_id,</span> <span class="pre">layer_idx,</span> <span class="pre">adapter_size</span> <span class="pre">D</span> <span class="pre">(i.e.</span> <span class="pre">R</span> <span class="pre">value)]</span></code>.</p>
<p>This feature supports LoRAs as described in https://arxiv.org/pdf/2106.09685.pdf</p>
<section id="example-lora-tensors">
<h3>Example LoRA tensors<a class="headerlink" href="#example-lora-tensors" title="Link to this heading">#</a></h3>
<p>Here is an example of <code class="docutils literal notranslate"><span class="pre">LoraWeights</span></code> and <code class="docutils literal notranslate"><span class="pre">LoraConfig</span></code> tensors for a model with tp=1, pp=1, 4 layers, and a hidden size of 4.
The following tensors are for a LoRA which has a <code class="docutils literal notranslate"><span class="pre">q</span></code> and <code class="docutils literal notranslate"><span class="pre">k</span></code> adapter.</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># loraConfig</span>
<span class="p">[</span>
<span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">]</span>
<span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">4</span><span class="p">]</span>
<span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">]</span>
<span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">4</span><span class="p">]</span>
<span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">2</span><span class="p">]</span> <span class="c1"># Note that the final 2 layers only adapt `q`</span>
<span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">8</span><span class="p">]</span>
<span class="p">]</span>
<span class="c1"># Note: The loraConfig tensor configures the loraWeights tensor.</span>
<span class="c1"># The contents of each row of loraWeights is specified be the corresponding row in loraConfig</span>
<span class="c1"># loraWeights</span>
<span class="c1"># Note: that &#39;in weights&#39; and &#39;out weights&#39; are &#39;A&#39; and &#39;B&#39; in the LoRA paper.</span>
<span class="p">[</span>
<span class="p">[</span> <span class="o">&lt;</span><span class="mi">2</span> <span class="n">x</span> <span class="mi">4</span> <span class="ow">in</span> <span class="n">weights</span><span class="o">&gt;</span><span class="p">,</span> <span class="o">&lt;</span><span class="mi">4</span> <span class="n">x</span> <span class="mi">2</span> <span class="n">out</span> <span class="n">weights</span><span class="o">&gt;</span> <span class="o">&lt;</span><span class="n">padding</span><span class="o">&gt;</span> <span class="p">]</span> <span class="c1"># `q` adapter for layer 0</span>
<span class="p">[</span> <span class="o">&lt;</span><span class="mi">4</span> <span class="n">x</span> <span class="mi">4</span> <span class="ow">in</span> <span class="n">weights</span><span class="o">&gt;</span><span class="p">,</span> <span class="o">&lt;</span><span class="mi">4</span> <span class="n">x</span> <span class="mi">4</span> <span class="n">out</span> <span class="n">weights</span><span class="o">&gt;</span> <span class="o">&lt;</span><span class="n">padding</span><span class="o">&gt;</span> <span class="p">]</span> <span class="c1"># `k` adapter for layer 0</span>
<span class="p">[</span> <span class="o">&lt;</span><span class="mi">2</span> <span class="n">x</span> <span class="mi">4</span> <span class="ow">in</span> <span class="n">weights</span><span class="o">&gt;</span><span class="p">,</span> <span class="o">&lt;</span><span class="mi">4</span> <span class="n">x</span> <span class="mi">2</span> <span class="n">out</span> <span class="n">weights</span><span class="o">&gt;</span> <span class="o">&lt;</span><span class="n">padding</span><span class="o">&gt;</span> <span class="p">]</span> <span class="c1"># `q` adapter for layer 1</span>
<span class="p">[</span> <span class="o">&lt;</span><span class="mi">4</span> <span class="n">x</span> <span class="mi">4</span> <span class="ow">in</span> <span class="n">weights</span><span class="o">&gt;</span><span class="p">,</span> <span class="o">&lt;</span><span class="mi">4</span> <span class="n">x</span> <span class="mi">4</span> <span class="n">out</span> <span class="n">weights</span><span class="o">&gt;</span> <span class="o">&lt;</span><span class="n">padding</span><span class="o">&gt;</span> <span class="p">]</span> <span class="c1"># `k` adapter for layer 1</span>
<span class="p">[</span> <span class="o">&lt;</span><span class="mi">2</span> <span class="n">x</span> <span class="mi">4</span> <span class="ow">in</span> <span class="n">weights</span><span class="o">&gt;</span><span class="p">,</span> <span class="o">&lt;</span><span class="mi">4</span> <span class="n">x</span> <span class="mi">2</span> <span class="n">out</span> <span class="n">weights</span><span class="o">&gt;</span> <span class="o">&lt;</span><span class="n">padding</span><span class="o">&gt;</span> <span class="p">]</span> <span class="c1"># `q` adapter for layer 2</span>
<span class="p">[</span> <span class="o">&lt;</span><span class="mi">8</span> <span class="n">x</span> <span class="mi">4</span> <span class="ow">in</span> <span class="n">weights</span><span class="o">&gt;</span><span class="p">,</span> <span class="o">&lt;</span><span class="mi">4</span> <span class="n">x</span> <span class="mi">8</span> <span class="n">out</span> <span class="n">weights</span><span class="o">&gt;</span> <span class="p">]</span> <span class="c1"># `q` adapter for layer 3. Note the final layer has a adapter size of 8</span>
<span class="p">]</span>
</pre></div>
</div>
</section>
<section id="lora-module-id-mapping">
<h3>LoRA Module id mapping<a class="headerlink" href="#lora-module-id-mapping" title="Link to this heading">#</a></h3>
<div class="pst-scrollable-table-container"><table class="table">
<thead>
<tr class="row-odd"><th class="head"><p>module name (as specified in <code class="docutils literal notranslate"><span class="pre">convert_checkpoint.py</span></code> scripts)</p></th>
<th class="head"><p>module id</p></th>
<th class="head"><p>description</p></th>
</tr>
</thead>
<tbody>
<tr class="row-even"><td><p>attn_qkv</p></td>
<td><p>0</p></td>
<td><p>compbined qkv adapter</p></td>
</tr>
<tr class="row-odd"><td><p>attn_q</p></td>
<td><p>1</p></td>
<td><p>q adapter</p></td>
</tr>
<tr class="row-even"><td><p>attn_k</p></td>
<td><p>2</p></td>
<td><p>k adapter</p></td>
</tr>
<tr class="row-odd"><td><p>attn_v</p></td>
<td><p>3</p></td>
<td><p>v adapter</p></td>
</tr>
<tr class="row-even"><td><p>attn_dense</p></td>
<td><p>4</p></td>
<td><p>adapter for the dense layer in attention</p></td>
</tr>
<tr class="row-odd"><td><p>mlp_h_to_4h</p></td>
<td><p>5</p></td>
<td><p>for llama2 adapter for gated mlp layer after attention / RMSNorm: up projection</p></td>
</tr>
<tr class="row-even"><td><p>mlp_4h_to_h</p></td>
<td><p>6</p></td>
<td><p>for llama2 adapter for gated mlp layer after attention / RMSNorm: down projection</p></td>
</tr>
<tr class="row-odd"><td><p>mlp_gate</p></td>
<td><p>7</p></td>
<td><p>for llama2 adapter for gated mlp later after attention / RMSNorm: gate</p></td>
</tr>
<tr class="row-even"><td><p>cross_attn_qkv</p></td>
<td><p>8</p></td>
<td><p>compbined qkv adapter for cross attention</p></td>
</tr>
<tr class="row-odd"><td><p>cross_attn_q</p></td>
<td><p>9</p></td>
<td><p>q adapter for cross attention</p></td>
</tr>
<tr class="row-even"><td><p>cross_attn_k</p></td>
<td><p>10</p></td>
<td><p>k adapter for cross attention</p></td>
</tr>
<tr class="row-odd"><td><p>cross_attn_v</p></td>
<td><p>11</p></td>
<td><p>v adapter for cross attention</p></td>
</tr>
<tr class="row-even"><td><p>cross_attn_dense</p></td>
<td><p>12</p></td>
<td><p>adapter for the dense layer in cross attention</p></td>
</tr>
<tr class="row-odd"><td><p>moe_h_to_4h</p></td>
<td><p>13</p></td>
<td><p>for mixtral adapter for expert mlp layer: up projection</p></td>
</tr>
<tr class="row-even"><td><p>moe_4h_to_h</p></td>
<td><p>14</p></td>
<td><p>for mixtral adapter for expert mlp layer: down projection</p></td>
</tr>
<tr class="row-odd"><td><p>moe_gate</p></td>
<td><p>15</p></td>
<td><p>for mixtral adapter for expert mlp layer: gate</p></td>
</tr>
<tr class="row-even"><td><p>moe_router</p></td>
<td><p>16</p></td>
<td><p>for mixtral adapter for expert router layer</p></td>
</tr>
<tr class="row-odd"><td><p>mlp_router</p></td>
<td><p>17</p></td>
<td><p>for qwen2-moe adapter for shared expert gate layer</p></td>
</tr>
<tr class="row-even"><td><p>mlp_gate_up</p></td>
<td><p>18</p></td>
<td><p>adapter for gated mlp layer after attention / RMSNorm: gate + up projection</p></td>
</tr>
</tbody>
</table>
</div>
</section>
<section id="loracache-configuration">
<h3>LoraCache configuration<a class="headerlink" href="#loracache-configuration" title="Link to this heading">#</a></h3>
<p>The core idea is that we will have a fixed size, 2-level LoRA cache in TRT-LLM. The higher level cache resides on the host and the lower level is on GPU (distinct from the existing KV cache). Sizes of both are user configurable.</p>
<p>The CPU cache is configured to be a max size. The GPU cache is configured to a percentage of free GPU memory after engine load. As requests come in LoRAs are stored in the host cache.</p>
<p>As requests are scheduled for execution LoRAs are loaded into the GPU cache.</p>
</section>
<section id="lora-with-tensor-parallel">
<h3>LoRA with tensor parallel<a class="headerlink" href="#lora-with-tensor-parallel" title="Link to this heading">#</a></h3>
<p>The partition of tensor parallel for LoRA is special. There are two cases: <code class="docutils literal notranslate"><span class="pre">RowLinear</span></code> and <code class="docutils literal notranslate"><span class="pre">ColumnLinear</span></code>. Assume we have a linear layer and the input feature size is <code class="docutils literal notranslate"><span class="pre">K</span></code> and the output feature size is <code class="docutils literal notranslate"><span class="pre">N</span></code>. Then, the shape of the weight is <code class="docutils literal notranslate"><span class="pre">[K,</span> <span class="pre">N]</span></code>.</p>
<p>First, consider this linear layer is a <code class="docutils literal notranslate"><span class="pre">ColumnLinear</span></code> layer. When we partition the weight, we split the weight by column with <code class="docutils literal notranslate"><span class="pre">tp_size</span></code>. Then, there are <code class="docutils literal notranslate"><span class="pre">tp_size</span></code> split weights and the shapes of these weights are <code class="docutils literal notranslate"><span class="pre">[K,</span> <span class="pre">N</span> <span class="pre">//</span> <span class="pre">tp_size]</span></code>. When we apply LoRA adapter on such <code class="docutils literal notranslate"><span class="pre">ColumnLinear</span></code> layer, the shapes of original two weights are <code class="docutils literal notranslate"><span class="pre">[K,</span> <span class="pre">lora_rank]</span></code> and <code class="docutils literal notranslate"><span class="pre">[lora_rank,</span> <span class="pre">N]</span></code>. So, we only partition the second weight and get <code class="docutils literal notranslate"><span class="pre">tp_size</span></code> split weights with shapes <code class="docutils literal notranslate"><span class="pre">[lora_rank,</span> <span class="pre">N</span> <span class="pre">//</span> <span class="pre">tp_size]</span></code>. For the first weight, each GPU maintains the same entire weight (with shape <code class="docutils literal notranslate"><span class="pre">[K,</span> <span class="pre">lora_rank]</span></code>).</p>
<p>Next, consider this linear layer is a <code class="docutils literal notranslate"><span class="pre">RowLinear</span></code> layer. When we partition the weight, we split the weight by row with <code class="docutils literal notranslate"><span class="pre">tp_size</span></code>. Then, there are <code class="docutils literal notranslate"><span class="pre">tp_size</span></code> split weights and the shapes of these weights are <code class="docutils literal notranslate"><span class="pre">[K</span> <span class="pre">//</span> <span class="pre">tp_size,</span> <span class="pre">N]</span></code>. When we apply LoRA adapter on such <code class="docutils literal notranslate"><span class="pre">RowLinear</span></code> layer, the shapes of original two weights are <code class="docutils literal notranslate"><span class="pre">[K,</span> <span class="pre">lora_rank]</span></code> and <code class="docutils literal notranslate"><span class="pre">[lora_rank,</span> <span class="pre">N]</span></code>. So, we only partition the first weight and get <code class="docutils literal notranslate"><span class="pre">tp_size</span></code> split weights with shapes <code class="docutils literal notranslate"><span class="pre">[K</span> <span class="pre">//</span> <span class="pre">tp_size,</span> <span class="pre">lora_rank]</span></code>. For the second weight, each GPU maintains the same entire weight (with shape <code class="docutils literal notranslate"><span class="pre">[lora_rank,</span> <span class="pre">N]</span></code>).</p>
</section>
<section id="dora">
<h3>DoRA<a class="headerlink" href="#dora" title="Link to this heading">#</a></h3>
<p>TensorRT-LLM supports DoRA as described in https://arxiv.org/abs/2402.09353 . To enable DoRA, you must add the additional <code class="docutils literal notranslate"><span class="pre">--dora_plugin</span> <span class="pre">enable</span></code> flag to the <code class="docutils literal notranslate"><span class="pre">trtllm-build</span></code> command.</p>
<p>The DoRA scales must be normalized before they are submitted to TensorRT-LLM in an inference request. The normalization requires the base model weights. To normalize your adapter you may use the script provided in <code class="docutils literal notranslate"><span class="pre">tensorrt_llm/examples/dora/normalize_weights.py</span></code>.</p>
<p>When using DoRA, the format of <code class="docutils literal notranslate"><span class="pre">LoraWeights</span></code> and <code class="docutils literal notranslate"><span class="pre">LoraConfig</span></code> changes slightly.
The shape of <code class="docutils literal notranslate"><span class="pre">LoraConfig</span></code> becomes <code class="docutils literal notranslate"><span class="pre">[module_id,</span> <span class="pre">layer_idx,</span> <span class="pre">adapter_size</span> <span class="pre">D</span> <span class="pre">(i.e.</span> <span class="pre">R</span> <span class="pre">value),</span> <span class="pre">is_dora]</span></code>, with <code class="docutils literal notranslate"><span class="pre">is_dora</span></code> a boolean flag that determines whether the supplied adapter contains DoRA scales or not. If the old config shape is used, it is assumed the adapter does not have DoRA scales.
The shape of <code class="docutils literal notranslate"><span class="pre">LoraWeights</span></code> becomes <code class="docutils literal notranslate"><span class="pre">[num_lora_modules_layers,</span> <span class="pre">D</span> <span class="pre">x</span> <span class="pre">Hi</span> <span class="pre">+</span> <span class="pre">Ho</span> <span class="pre">x</span> <span class="pre">D</span> <span class="pre">+</span> <span class="pre">Ho]</span></code>, and the last <code class="docutils literal notranslate"><span class="pre">Ho</span></code> values are the DoRA scale vector.</p>
</section>
</section>
</section>
</article>
<footer class="prev-next-footer d-print-none">
<div class="prev-next-area">
</div>
</footer>
</div>
<dialog id="pst-secondary-sidebar-modal"></dialog>
<div id="pst-secondary-sidebar" class="bd-sidebar-secondary bd-toc"><div class="sidebar-secondary-items sidebar-secondary__inner">
<div class="sidebar-secondary-item">
<div
id="pst-page-navigation-heading-2"
class="page-toc tocsection onthispage">
<i class="fa-solid fa-list"></i> On this page
</div>
<nav class="bd-toc-nav page-toc" aria-labelledby="pst-page-navigation-heading-2">
<ul class="visible nav section-nav flex-column">
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#lora-tensor-format-details">LoRA tensor format details</a><ul class="nav section-nav flex-column">
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#example-lora-tensors">Example LoRA tensors</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#lora-module-id-mapping">LoRA Module id mapping</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#loracache-configuration">LoraCache configuration</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#lora-with-tensor-parallel">LoRA with tensor parallel</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#dora">DoRA</a></li>
</ul>
</li>
</ul>
</nav></div>
</div></div>
</div>
<footer class="bd-footer-content">
</footer>
</main>
</div>
</div>
<!-- Scripts loaded after <body> so the DOM is not blocked -->
<script defer src="../_static/scripts/bootstrap.js?digest=8878045cc6db502f8baf"></script>
<script defer src="../_static/scripts/pydata-sphinx-theme.js?digest=8878045cc6db502f8baf"></script>
<footer class="bd-footer">
<div class="bd-footer__inner bd-page-width">
<div class="footer-items__start">
<div class="footer-item">
<a class="footer-brand logo" href="https://www.nvidia.com">
<img src="../_static/nvidia-logo-horiz-rgb-1c-blk-for-screen.svg" class="logo__image only-light" alt="NVIDIA"/>
<img src="../_static/nvidia-logo-horiz-rgb-1c-wht-for-screen.svg" class="logo__image only-dark" alt="NVIDIA"/>
</a></div>
<div class="footer-item">
<div class="footer-links">
<a class="external" href="https://www.nvidia.com/en-us/about-nvidia/privacy-policy/">Privacy Policy</a>
|
<a class="external" href="https://www.nvidia.com/en-us/about-nvidia/privacy-center/">Manage My Privacy</a>
|
<a class="external" href="https://www.nvidia.com/en-us/preferences/start/">Do Not Sell or Share My Data</a>
|
<a class="external" href="https://www.nvidia.com/en-us/about-nvidia/terms-of-service/">Terms of Service</a>
|
<a class="external" href="https://www.nvidia.com/en-us/about-nvidia/accessibility/">Accessibility</a>
|
<a class="external" href="https://www.nvidia.com/en-us/about-nvidia/company-policies/">Corporate Policies</a>
|
<a class="external" href="https://www.nvidia.com/en-us/product-security/">Product Security</a>
|
<a class="external" href="https://www.nvidia.com/en-us/contact/">Contact</a>
</div>
</div>
<div class="footer-item">
<p class="copyright">
Copyright © 2025, NVidia.
<br/>
</p>
</div>
<div class="footer-item">
<div class="extra_footer">
<p>Last updated on September 23, 2025.</p>
<p>This page is generated by TensorRT-LLM commit <a href="https://github.com/NVIDIA/TensorRT-LLM/tree/ae8270b">ae8270b</a>.</p>
</div></div>
</div>
</div>
</footer>
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink