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
464700a50a
TensorRT-LLMs/features/checkpoint-loading.html
Kaiyu Xie 464700a50a Update GitHub pages in root to v1.1.0rc4
2025-09-10 06:17:44 +00:00

1002 lines
68 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.

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 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>Checkpoint Loading &#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=95073da6" />
<!-- 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 = 'features/checkpoint-loading';</script>
<script>
DOCUMENTATION_OPTIONS.theme_version = '0.16.1';
DOCUMENTATION_OPTIONS.theme_switcher_json_url = './_static/switcher.json';
DOCUMENTATION_OPTIONS.theme_switcher_version_match = '1.1.0rc4';
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" />
<link rel="next" title="AutoDeploy (Prototype)" href="auto_deploy/auto-deploy.html" />
<link rel="prev" title="Speculative Decoding" href="speculative-decoding.html" />
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<meta name="docsearch:language" content="en"/>
<meta name="docsearch:version" content="1.1.0rc4" />
</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_kv_cache_connector.html">KV Cache Connector</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>
<li class="toctree-l2"><a class="reference internal" href="../deployment-guide/quick-start-recipe-for-gpt-oss-on-trtllm.html">Quick Start Recipe for GPT-OSS on TensorRT-LLM - Blackwell 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="current nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="feature-combination-matrix.html">Feature Combination Matrix</a></li>
<li class="toctree-l1"><a class="reference internal" href="attention.html">Multi-Head, Multi-Query, and Group-Query Attention</a></li>
<li class="toctree-l1"><a class="reference internal" href="disagg-serving.html">Disaggregated Serving (Beta)</a></li>
<li class="toctree-l1"><a class="reference internal" href="kvcache.html">KV Cache System</a></li>
<li class="toctree-l1"><a class="reference internal" href="long-sequence.html">Long Sequences</a></li>
<li class="toctree-l1"><a class="reference internal" href="lora.html">LoRA (Low-Rank Adaptation)</a></li>
<li class="toctree-l1"><a class="reference internal" href="multi-modality.html">Multimodal Support in TensorRT LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="overlap-scheduler.html">Overlap Scheduler</a></li>
<li class="toctree-l1"><a class="reference internal" href="paged-attention-ifb-scheduler.html">Paged Attention, IFB, and Request Scheduling</a></li>
<li class="toctree-l1"><a class="reference internal" href="parallel-strategy.html">Parallelism in TensorRT LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="quantization.html">Quantization</a></li>
<li class="toctree-l1"><a class="reference internal" href="sampling.html">Sampling</a></li>
<li class="toctree-l1"><a class="reference internal" href="speculative-decoding.html">Speculative Decoding</a></li>
<li class="toctree-l1 current active"><a class="current reference internal" href="#">Checkpoint Loading</a></li>
<li class="toctree-l1"><a class="reference internal" href="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>
</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/blog10_ADP_Balance_Strategy.html">ADP Balance Strategy</a></li>
<li class="toctree-l1"><a class="reference internal" href="../blogs/tech_blog/blog11_GPT_OSS_Eagle3.html">Running GPT-OSS-120B with Eagle3 Speculative Decoding on GB200/B200 (TensorRT LLM)</a></li>
<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">Checkpoint Loading</span></li>
</ul>
</nav>
</div>
</div>
</div>
</div>
<div id="searchbox"></div>
<article class="bd-article">
<section id="checkpoint-loading">
<h1>Checkpoint Loading<a class="headerlink" href="#checkpoint-loading" title="Link to this heading">#</a></h1>
<p>The PyTorch backend provides a flexible and extensible infrastructure for loading model checkpoints from different formats, such as HuggingFace (HF). This system allows you to load models from various sources (e.g., HuggingFace or custom formats) by implementing the required components, such as the checkpoint’s weight loader, mapper, and configuration parser.</p>
<section id="table-of-contents">
<h2>Table of Contents<a class="headerlink" href="#table-of-contents" title="Link to this heading">#</a></h2>
<ol class="arabic simple">
<li><p><a class="reference internal" href="#overview">Overview</a></p></li>
<li><p><a class="reference internal" href="#core-components">Core Components</a></p></li>
<li><p><a class="reference internal" href="#built-in-checkpoint-formats">Built-in Checkpoint Formats</a></p></li>
<li><p><a class="reference internal" href="#using-checkpoint-loaders">Using Checkpoint Loaders</a></p></li>
<li><p><a class="reference internal" href="#creating-custom-checkpoint-loaders">Creating Custom Checkpoint Loaders</a></p></li>
</ol>
</section>
<section id="overview">
<h2>Overview<a class="headerlink" href="#overview" title="Link to this heading">#</a></h2>
<p>The checkpoint loading design is built around a plugin-like architecture that is separated into four distinct components:</p>
<ul class="simple">
<li><p><strong>Checkpoint Loaders</strong>: Orchestrate the loading process for specific formats</p></li>
<li><p><strong>Config Loaders</strong>: Handle model configuration parsing and validation</p></li>
<li><p><strong>Weight Loaders</strong>: Manage the actual loading of model weights from storage into memory</p></li>
<li><p><strong>Weight Mappers</strong>: Map and transform loaded weights to TensorRT LLM model’s definition</p></li>
</ul>
<p>This modular design allows for easy extension to support new checkpoint formats while maintaining backward compatibility and performance optimizations. By separating the checkpoint loading components into four different subcomponents, any user can employ any relevant previous work while also introducing their own custom checkpoint-specific components.</p>
<p>If one wishes to support a new checkpoint format, they must implement all four components.
Likewise, if the format shares some components with an already supported framework (e.g., HF), only the custom-specific components need to be implemented.</p>
</section>
<section id="core-components">
<h2>Core Components<a class="headerlink" href="#core-components" title="Link to this heading">#</a></h2>
<section id="basecheckpointloader">
<h3>BaseCheckpointLoader<a class="headerlink" href="#basecheckpointloader" title="Link to this heading">#</a></h3>
<p>The <code class="docutils literal notranslate"><span class="pre">BaseCheckpointLoader</span></code> is the central base interface for all checkpoint loading required operators. It provides a unified API regardless of the underlying checkpoint format. This interface is responsible for holding and exposing all objects required for the loading and parsing process.</p>
<p><strong>Key Methods:</strong></p>
<ul class="simple">
<li><p><code class="docutils literal notranslate"><span class="pre">load_config(checkpoint_dir,</span> <span class="pre">**kwargs)</span></code>: Loads and returns a <code class="docutils literal notranslate"><span class="pre">ModelConfig</span></code> object</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">load_weights(checkpoint_dir,</span> <span class="pre">**kwargs)</span></code>: Loads and returns a dictionary of weights</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">get_initialized_weight_mapper(model,</span> <span class="pre">config)</span></code>: Returns a runtime initialized weight mapper for the model</p></li>
<li><p><code class="docutils literal notranslate"><span class="pre">cleanup()</span></code>: Releases resources and cleans up internal state</p></li>
</ul>
</section>
<section id="baseconfigloader">
<h3>BaseConfigLoader<a class="headerlink" href="#baseconfigloader" title="Link to this heading">#</a></h3>
<p>Responsible for loading model configurations from checkpoint directories and parsing them into TRTLLM <code class="docutils literal notranslate"><span class="pre">ModelConfig</span></code>:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_config_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseConfigLoader</span>
<span class="k">class</span><span class="w"> </span><span class="nc">CustomConfigLoader</span><span class="p">(</span><span class="n">BaseConfigLoader</span><span class="p">):</span>
<span class="k">def</span><span class="w"> </span><span class="nf">load</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">checkpoint_dir</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">ModelConfig</span><span class="p">:</span>
<span class="c1"># Load and parse configuration from your custom format</span>
<span class="n">pretrained_config</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_get_pretrained_config</span><span class="p">(</span><span class="n">checkpoint_dir</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">)</span>
<span class="k">return</span> <span class="n">ModelConfig</span><span class="p">(</span><span class="n">pretrained_config</span><span class="o">=</span><span class="n">pretrained_config</span><span class="p">,</span>
<span class="o">...</span><span class="p">)</span>
<span class="k">def</span><span class="w"> </span><span class="nf">_get_pretrained_config</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">checkpoint_dir</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">):</span>
<span class="o">...</span>
</pre></div>
</div>
</section>
<section id="baseweightloader">
<h3>BaseWeightLoader<a class="headerlink" href="#baseweightloader" title="Link to this heading">#</a></h3>
<p>Handles the loading of model weights from storage:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_weight_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseWeightLoader</span>
<span class="k">class</span><span class="w"> </span><span class="nc">CustomWeightLoader</span><span class="p">(</span><span class="n">BaseWeightLoader</span><span class="p">):</span>
<span class="k">def</span><span class="w"> </span><span class="nf">load_weights</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">checkpoint_dir</span><span class="p">:</span> <span class="nb">str</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">Any</span><span class="p">]:</span>
<span class="c1"># Load weights from your custom format</span>
<span class="c1"># Return a dictionary mapping parameter names to tensors</span>
<span class="k">return</span> <span class="n">weights_dict</span>
</pre></div>
</div>
</section>
<section id="baseweightmapper">
<h3>BaseWeightMapper<a class="headerlink" href="#baseweightmapper" title="Link to this heading">#</a></h3>
<p>Transforms weights between different naming conventions and applies model-specific transformations into TRTLLM model’s object.</p>
</section>
</section>
<section id="built-in-checkpoint-formats">
<h2>Built-in Checkpoint Formats<a class="headerlink" href="#built-in-checkpoint-formats" title="Link to this heading">#</a></h2>
<section id="huggingface-format">
<h3>HuggingFace Format<a class="headerlink" href="#huggingface-format" title="Link to this heading">#</a></h3>
<p>Currently, HF checkpoint loader is the primary built-in format, supporting:</p>
<ul class="simple">
<li><p><strong>Weights loading</strong> (<code class="docutils literal notranslate"><span class="pre">.safetensors/.bin/.pth</span></code>) - Loading HF compatible weights from disk</p></li>
<li><p><strong>Configuration parser</strong> - Parsing HF stored configuration information to TRTLLM <code class="docutils literal notranslate"><span class="pre">ModelConfig</span></code> object</p></li>
<li><p><strong>Weights Mapping</strong> - Converting HF weights into TRTLLM compatible representation</p></li>
</ul>
</section>
</section>
<section id="using-checkpoint-loaders">
<h2>Using Checkpoint Loaders<a class="headerlink" href="#using-checkpoint-loaders" title="Link to this heading">#</a></h2>
<section id="basic-usage">
<h3>Basic Usage<a class="headerlink" href="#basic-usage" title="Link to this heading">#</a></h3>
<p>There are two main approaches to trigger the use of checkpoint loading objects.</p>
<p>The first approach, through llm-api, as shown in the following example:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm</span><span class="w"> </span><span class="kn">import</span> <span class="n">LLM</span>
<span class="n">hf_model_dir</span> <span class="o">=</span> <span class="s2">&quot;llama-models-v2/llama-v2-13b-hf&quot;</span>
<span class="n">llm</span> <span class="o">=</span> <span class="n">LLM</span><span class="p">(</span><span class="n">model</span><span class="o">=</span><span class="n">hf_model_dir</span><span class="p">)</span>
</pre></div>
</div>
<p>In this example, <code class="docutils literal notranslate"><span class="pre">HfCheckpointLoader</span></code> will be selected by default.</p>
<p>To explicitly set the checkpoint loader, you need to call the required checkpoint-specific loader</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm</span><span class="w"> </span><span class="kn">import</span> <span class="n">LLM</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.hf.checkpoint_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">HfCheckpointLoader</span>
<span class="n">hf_model_dir</span> <span class="o">=</span> <span class="s2">&quot;llama-models-v2/llama-v2-13b-hf&quot;</span>
<span class="n">llm</span> <span class="o">=</span> <span class="n">LLM</span><span class="p">(</span><span class="n">model</span><span class="o">=</span><span class="n">hf_model_dir</span><span class="p">,</span>
<span class="n">checkpoint_loader</span><span class="o">=</span><span class="n">HfCheckpointLoader</span><span class="p">())</span>
</pre></div>
</div>
<p>Similarly, if one wants to use a basic implemented checkpoint loader, but with a specific subcomponent, they can provide any specific subcomponent upon need</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm</span><span class="w"> </span><span class="kn">import</span> <span class="n">LLM</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.hf.checkpoint_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">HfCheckpointLoader</span>
<span class="n">hf_model_dir</span> <span class="o">=</span> <span class="s2">&quot;llama-models-v2/llama-v2-13b-hf&quot;</span>
<span class="n">llm</span> <span class="o">=</span> <span class="n">LLM</span><span class="p">(</span><span class="n">model</span><span class="o">=</span><span class="n">hf_model_dir</span><span class="p">,</span>
<span class="n">checkpoint_loader</span><span class="o">=</span><span class="n">HfCheckpointLoader</span><span class="p">(</span><span class="n">weight_loader</span><span class="o">=</span><span class="n">MyCustomWeightLoader</span><span class="p">()))</span>
</pre></div>
</div>
<p>In the second approach, one can directly use the components of the checkpoint loading.</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.hf.gemma3_weight_mapper</span><span class="w"> </span><span class="kn">import</span> \
<span class="n">Gemma3HfWeightMapper</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.modeling_gemma3</span><span class="w"> </span><span class="kn">import</span> <span class="n">Gemma3ForCausalLM</span>
<span class="n">gemma3</span> <span class="o">=</span> <span class="n">Gemma3ForCausalLM</span><span class="p">(</span><span class="n">model_config</span><span class="p">)</span>
<span class="n">weight_mapper</span> <span class="o">=</span> <span class="n">Gemma3HfWeightMapper</span><span class="p">()</span>
<span class="n">weight_mapper</span><span class="o">.</span><span class="n">init_model_and_config</span><span class="p">(</span><span class="n">gemma3</span><span class="p">,</span> <span class="n">model_config</span><span class="p">)</span>
<span class="n">gemma3</span><span class="o">.</span><span class="n">load_weights</span><span class="p">(</span><span class="n">hf_gemma3</span><span class="o">.</span><span class="n">state_dict</span><span class="p">(),</span> <span class="n">weight_mapper</span><span class="p">)</span>
</pre></div>
</div>
</section>
</section>
<section id="creating-custom-checkpoint-loaders">
<h2>Creating Custom Checkpoint Loaders<a class="headerlink" href="#creating-custom-checkpoint-loaders" title="Link to this heading">#</a></h2>
<p>To support a new checkpoint format, you need to implement all four components. This section provides minimal templates for each component.</p>
<section id="when-to-create-custom-components">
<h3>When to Create Custom Components<a class="headerlink" href="#when-to-create-custom-components" title="Link to this heading">#</a></h3>
<ul class="simple">
<li><p><strong>Complete New Format</strong>: Implement all four components when supporting a completely new checkpoint format</p></li>
<li><p><strong>Custom Weight Storage</strong>: Only implement a custom weight loader if you have a unique weight storage format (e.g., custom binary format, database storage, etc.)</p></li>
<li><p><strong>Custom Configuration</strong>: Only implement a custom config loader if your configuration format cannot be parsed by existing parsers.</p></li>
<li><p><strong>Custom Weight Mapping</strong>: Only implement a custom weight mapper if your model has unique weight naming or transformation requirements that are checkpoint-specific.</p></li>
</ul>
</section>
<section id="step-1-create-the-checkpoint-loader">
<h3>Step 1: Create the Checkpoint Loader<a class="headerlink" href="#step-1-create-the-checkpoint-loader" title="Link to this heading">#</a></h3>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">typing</span><span class="w"> </span><span class="kn">import</span> <span class="n">Optional</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_checkpoint_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseCheckpointLoader</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_config_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseConfigLoader</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_weight_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseWeightLoader</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_weight_mapper</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseWeightMapper</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.modeling_utils</span><span class="w"> </span><span class="kn">import</span> <span class="n">register_checkpoint_loader</span>
<span class="nd">@register_checkpoint_loader</span><span class="p">(</span><span class="s2">&quot;CUSTOM_FORMAT&quot;</span><span class="p">)</span>
<span class="k">class</span><span class="w"> </span><span class="nc">CustomCheckpointLoader</span><span class="p">(</span><span class="n">BaseCheckpointLoader</span><span class="p">):</span>
<span class="k">def</span><span class="w"> </span><span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span>
<span class="o">*</span><span class="p">,</span>
<span class="n">weight_loader</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">BaseWeightLoader</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">weight_mapper</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">BaseWeightMapper</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">config_loader</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">BaseConfigLoader</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">):</span>
<span class="bp">self</span><span class="o">.</span><span class="n">_weight_loader</span> <span class="o">=</span> <span class="n">weight_loader</span> <span class="ow">or</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_default_weight_loader</span><span class="p">()</span>
<span class="bp">self</span><span class="o">.</span><span class="n">_config_loader</span> <span class="o">=</span> <span class="n">config_loader</span> <span class="ow">or</span> <span class="bp">self</span><span class="o">.</span><span class="n">get_default_config_loader</span><span class="p">()</span>
<span class="bp">self</span><span class="o">.</span><span class="n">_weight_mapper</span> <span class="o">=</span> <span class="n">weight_mapper</span>
<span class="bp">self</span><span class="o">.</span><span class="n">_checkpoint_format</span> <span class="o">=</span> <span class="s2">&quot;CUSTOM_FORMAT&quot;</span>
<span class="k">def</span><span class="w"> </span><span class="nf">get_default_weight_loader</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">BaseWeightLoader</span><span class="p">:</span>
<span class="k">return</span> <span class="n">CustomWeightLoader</span><span class="p">()</span>
<span class="k">def</span><span class="w"> </span><span class="nf">get_default_config_loader</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">BaseConfigLoader</span><span class="p">:</span>
<span class="k">return</span> <span class="n">CustomConfigLoader</span><span class="p">()</span>
</pre></div>
</div>
</section>
<section id="step-2-create-the-checkpoint-weight-loader">
<h3>Step 2: Create the Checkpoint Weight Loader<a class="headerlink" href="#step-2-create-the-checkpoint-weight-loader" title="Link to this heading">#</a></h3>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">typing</span><span class="w"> </span><span class="kn">import</span> <span class="n">Any</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_weight_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseWeightLoader</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.modeling_utils</span><span class="w"> </span><span class="kn">import</span> <span class="n">register_checkpoint_weight_loader</span>
<span class="nd">@register_checkpoint_weight_loader</span><span class="p">(</span><span class="s2">&quot;CUSTOM_FORMAT&quot;</span><span class="p">)</span>
<span class="k">class</span><span class="w"> </span><span class="nc">CustomWeightLoader</span><span class="p">(</span><span class="n">BaseWeightLoader</span><span class="p">):</span>
<span class="k">def</span><span class="w"> </span><span class="nf">load_weights</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">checkpoint_dir</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">dict</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">Any</span><span class="p">]:</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> Load weights from your custom format.</span>
<span class="sd"> Args:</span>
<span class="sd"> checkpoint_dir: Directory containing checkpoint files</span>
<span class="sd"> **kwargs: Additional loading parameters</span>
<span class="sd"> Returns:</span>
<span class="sd"> Dictionary mapping parameter names to tensors</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="n">weights</span> <span class="o">=</span> <span class="p">{}</span>
<span class="c1"># Implement your custom weight loading logic here</span>
<span class="c1"># Examples:</span>
<span class="c1"># - Load from custom binary files</span>
<span class="c1"># - Load from databases</span>
<span class="c1"># - Load from compressed archives</span>
<span class="c1"># - Apply custom preprocessing</span>
<span class="k">return</span> <span class="n">weights</span>
</pre></div>
</div>
</section>
<section id="step-3-create-the-checkpoint-config-loader">
<h3>Step 3: Create the Checkpoint Config Loader<a class="headerlink" href="#step-3-create-the-checkpoint-config-loader" title="Link to this heading">#</a></h3>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.model_config</span><span class="w"> </span><span class="kn">import</span> <span class="n">ModelConfig</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_config_loader</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseConfigLoader</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.modeling_utils</span><span class="w"> </span><span class="kn">import</span> <span class="n">register_config_loader</span>
<span class="nd">@register_config_loader</span><span class="p">(</span><span class="s2">&quot;CUSTOM_FORMAT&quot;</span><span class="p">)</span>
<span class="k">class</span><span class="w"> </span><span class="nc">CustomConfigLoader</span><span class="p">(</span><span class="n">BaseConfigLoader</span><span class="p">):</span>
<span class="k">def</span><span class="w"> </span><span class="nf">load</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">checkpoint_dir</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">ModelConfig</span><span class="p">:</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> Load and parse configuration from your custom format.</span>
<span class="sd"> Args:</span>
<span class="sd"> checkpoint_dir: Directory containing configuration files</span>
<span class="sd"> **kwargs: Additional loading parameters</span>
<span class="sd"> Returns:</span>
<span class="sd"> ModelConfig object containing parsed configuration</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="c1"># Load your custom configuration format</span>
<span class="c1"># Examples:</span>
<span class="c1"># - Parse YAML/TOML files</span>
<span class="c1"># - Convert from proprietary formats</span>
<span class="n">pretrained_config</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_load_pretrained_config</span><span class="p">(</span><span class="n">checkpoint_dir</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">)</span>
<span class="k">return</span> <span class="n">ModelConfig</span><span class="p">(</span>
<span class="n">pretrained_config</span><span class="o">=</span><span class="n">pretrained_config</span><span class="p">,</span>
<span class="c1"># Add other ModelConfig parameters as needed</span>
<span class="p">)</span>
<span class="k">def</span><span class="w"> </span><span class="nf">_load_pretrained_config</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">checkpoint_dir</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">):</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;Load the raw configuration from your custom format.&quot;&quot;&quot;</span>
<span class="k">pass</span>
</pre></div>
</div>
</section>
<section id="step-4-create-the-checkpoint-weight-mapper">
<h3>Step 4: Create the Checkpoint Weight Mapper<a class="headerlink" href="#step-4-create-the-checkpoint-weight-mapper" title="Link to this heading">#</a></h3>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span><span class="w"> </span><span class="nn">torch</span><span class="w"> </span><span class="kn">import</span> <span class="n">nn</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.checkpoints.base_weight_mapper</span><span class="w"> </span><span class="kn">import</span> <span class="n">BaseWeightMapper</span>
<span class="kn">from</span><span class="w"> </span><span class="nn">tensorrt_llm._torch.models.modeling_utils</span><span class="w"> </span><span class="kn">import</span> <span class="n">register_mapper</span>
<span class="nd">@register_mapper</span><span class="p">(</span><span class="s2">&quot;CUSTOM_FORMAT&quot;</span><span class="p">)</span>
<span class="k">class</span><span class="w"> </span><span class="nc">CustomWeightMapper</span><span class="p">(</span><span class="n">BaseWeightMapper</span><span class="p">):</span>
<span class="k">def</span><span class="w"> </span><span class="fm">__init__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
<span class="nb">super</span><span class="p">()</span><span class="o">.</span><span class="fm">__init__</span><span class="p">()</span>
<span class="c1"># Define any weight transformation callbacks</span>
<span class="bp">self</span><span class="o">.</span><span class="n">_callbacks</span> <span class="o">=</span> <span class="p">[</span>
<span class="c1"># Add your custom weight transformation functions</span>
<span class="c1"># self._custom_transform_function,</span>
<span class="p">]</span>
<span class="k">def</span><span class="w"> </span><span class="nf">map_weights</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="kc">None</span><span class="p">:</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> Define mappings between source and target weight names.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">update</span><span class="p">({</span>
<span class="c1"># Map source names to target names</span>
<span class="c1"># &#39;target_module_name&#39;: [&#39;source_param1&#39;, &#39;source_param2&#39;],</span>
<span class="c1"># Example: &#39;qkv_proj&#39;: [&#39;q_proj&#39;, &#39;k_proj&#39;, &#39;v_proj&#39;]</span>
<span class="p">})</span>
<span class="k">def</span><span class="w"> </span><span class="nf">apply_callbacks</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">module</span><span class="p">:</span> <span class="n">nn</span><span class="o">.</span><span class="n">Module</span><span class="p">,</span> <span class="n">module_name</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span>
<span class="n">module_names_breakdown</span><span class="p">:</span> <span class="nb">list</span><span class="p">[</span><span class="nb">str</span><span class="p">],</span>
<span class="n">weights</span><span class="p">:</span> <span class="nb">dict</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">list</span><span class="p">[</span><span class="nb">dict</span><span class="p">]:</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> Apply weight transformations for modules that require special handling.</span>
<span class="sd"> Args:</span>
<span class="sd"> module: The target module</span>
<span class="sd"> module_name: The specific module name being processed</span>
<span class="sd"> module_names_breakdown: Module path components</span>
<span class="sd"> weights: Source weights dictionary</span>
<span class="sd"> Returns:</span>
<span class="sd"> List of transformed weight dictionaries</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="n">module_weights</span> <span class="o">=</span> <span class="p">[]</span>
<span class="k">for</span> <span class="n">new_name</span> <span class="ow">in</span> <span class="bp">self</span><span class="o">.</span><span class="n">_mapping</span><span class="p">[</span><span class="n">module_name</span><span class="p">]:</span>
<span class="c1"># Filter weights for this specific parameter</span>
<span class="n">fw</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">filter_weights</span><span class="p">(</span>
<span class="s1">&#39;.&#39;</span><span class="o">.</span><span class="n">join</span><span class="p">(</span><span class="n">module_names_breakdown</span> <span class="o">+</span> <span class="p">[</span><span class="n">new_name</span><span class="p">]),</span> <span class="n">weights</span><span class="p">)</span>
<span class="c1"># Apply transformation callbacks</span>
<span class="k">for</span> <span class="n">callback</span> <span class="ow">in</span> <span class="bp">self</span><span class="o">.</span><span class="n">_callbacks</span><span class="p">:</span>
<span class="n">fw</span> <span class="o">=</span> <span class="n">callback</span><span class="p">(</span><span class="n">module</span><span class="p">,</span> <span class="n">new_name</span><span class="p">,</span> <span class="n">fw</span><span class="p">)</span>
<span class="n">module_weights</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">fw</span><span class="p">)</span>
<span class="k">return</span> <span class="n">module_weights</span>
<span class="k">def</span><span class="w"> </span><span class="nf">should_skip_module</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">module_name</span><span class="p">:</span> <span class="nb">str</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">bool</span><span class="p">:</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> Define which modules should be skipped during loading.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="c1"># Add logic to skip specific modules based on your requirements</span>
<span class="c1"># Examples:</span>
<span class="c1"># - Skip LoRA-specific modules</span>
<span class="c1"># - Skip temporary/auxiliary modules</span>
<span class="k">return</span> <span class="nb">super</span><span class="p">()</span><span class="o">.</span><span class="n">should_skip_module</span><span class="p">(</span><span class="n">module_name</span><span class="p">)</span>
</pre></div>
</div>
<p>Note: when creating a custom mapper, you can either define a checkpoint-format-specific mapper. For example:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="nd">@register_mapper</span><span class="p">(</span><span class="s2">&quot;CUSTOM_FORMAT&quot;</span><span class="p">)</span>
<span class="k">class</span><span class="w"> </span><span class="nc">CustomWeightMapper</span><span class="p">(</span><span class="n">BaseWeightMapper</span><span class="p">)</span>
</pre></div>
</div>
<p>Alternatively, you can define a checkpoint-model-specific mapper. For example:</p>
<div class="highlight-python notranslate"><div class="highlight"><pre><span></span><span class="nd">@register_mapper</span><span class="p">(</span><span class="s2">&quot;CUSTOM_FORMAT&quot;</span><span class="p">,</span> <span class="s2">&quot;Gemma3ForCausalLM&quot;</span><span class="p">)</span>
<span class="k">class</span><span class="w"> </span><span class="nc">CustomWeightMapper</span><span class="p">(</span><span class="n">BaseWeightMapper</span><span class="p">)</span>
</pre></div>
</div>
<p>By setting the model name, the registered mapper will be asscoiated with the specific model.</p>
</section>
</section>
</section>
</article>
<footer class="prev-next-footer d-print-none">
<div class="prev-next-area">
<a class="left-prev"
href="speculative-decoding.html"
title="previous page">
<i class="fa-solid fa-angle-left"></i>
<div class="prev-next-info">
<p class="prev-next-subtitle">previous</p>
<p class="prev-next-title">Speculative Decoding</p>
</div>
</a>
<a class="right-next"
href="auto_deploy/auto-deploy.html"
title="next page">
<div class="prev-next-info">
<p class="prev-next-subtitle">next</p>
<p class="prev-next-title">AutoDeploy (Prototype)</p>
</div>
<i class="fa-solid fa-angle-right"></i>
</a>
</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="#table-of-contents">Table of Contents</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#overview">Overview</a></li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#core-components">Core Components</a><ul class="nav section-nav flex-column">
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#basecheckpointloader">BaseCheckpointLoader</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#baseconfigloader">BaseConfigLoader</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#baseweightloader">BaseWeightLoader</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#baseweightmapper">BaseWeightMapper</a></li>
</ul>
</li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#built-in-checkpoint-formats">Built-in Checkpoint Formats</a><ul class="nav section-nav flex-column">
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#huggingface-format">HuggingFace Format</a></li>
</ul>
</li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#using-checkpoint-loaders">Using Checkpoint Loaders</a><ul class="nav section-nav flex-column">
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#basic-usage">Basic Usage</a></li>
</ul>
</li>
<li class="toc-h2 nav-item toc-entry"><a class="reference internal nav-link" href="#creating-custom-checkpoint-loaders">Creating Custom Checkpoint Loaders</a><ul class="nav section-nav flex-column">
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#when-to-create-custom-components">When to Create Custom Components</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#step-1-create-the-checkpoint-loader">Step 1: Create the Checkpoint Loader</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#step-2-create-the-checkpoint-weight-loader">Step 2: Create the Checkpoint Weight Loader</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#step-3-create-the-checkpoint-config-loader">Step 3: Create the Checkpoint Config Loader</a></li>
<li class="toc-h3 nav-item toc-entry"><a class="reference internal nav-link" href="#step-4-create-the-checkpoint-weight-mapper">Step 4: Create the Checkpoint Weight Mapper</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 09, 2025.</p>
<p>This page is generated by TensorRT-LLM commit <a href="https://github.com/NVIDIA/TensorRT-LLM/tree/62b564a">62b564a</a>.</p>
</div></div>
</div>
</div>
</footer>
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink