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
4b3728c4f6
TensorRT-LLMs/_modules/tensorrt_llm/runtime/model_runner.html
石晓伟 4b3728c4f6
update gh-pages (#3400) 
Signed-off-by: Shixiaowei02 <39303645+Shixiaowei02@users.noreply.github.com>
2025-04-09 13:16:40 +08:00

1520 lines
171 KiB
HTML
Raw Blame History

<!DOCTYPE html>
<html lang="en" data-content_root="../../../" >
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
<title>tensorrt_llm.runtime.model_runner &#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=a746c00c" />
<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" />
<!-- 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=888ff710"></script>
<script src="../../../_static/sphinx_highlight.js?v=dc90522c"></script>
<script src="../../../_static/clipboard.min.js?v=a7894cd8"></script>
<script src="../../../_static/copybutton.js?v=65e89d2a"></script>
<script>DOCUMENTATION_OPTIONS.pagename = '_modules/tensorrt_llm/runtime/model_runner';</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="" />
</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="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>
</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__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"><a class="reference internal" href="../../../key-features.html">Key Features</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../torch.html">PyTorch Backend</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../release-notes.html">Release Notes</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Installation</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../installation/linux.html">Installing on Linux</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../installation/build-from-source-linux.html">Building from Source Code on Linux</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../installation/grace-hopper.html">Installing on Grace Hopper</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">LLM API</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../llm-api/index.html">API Introduction</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../llm-api/reference.html">API Reference</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">LLM API Examples</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1 has-children"><a class="reference internal" href="../../../llm-api-examples/index.html">LLM Examples Introduction</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="../../../llm-api-examples/llm_lookahead_decoding.html">Generate Text Using Lookahead Decoding</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_guided_decoding.html">Generate text with guided decoding</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_medusa_decoding.html">Generate Text Using Medusa Decoding</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_multilora.html">Generate text with multiple LoRA adapters</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference_async_streaming.html">Generate Text in Streaming</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference.html">Generate text</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_quantization.html">Generation with Quantization</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference_distributed.html">Distributed LLM Generation</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference_async.html">Generate Text Asynchronously</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_logits_processor.html">Control generated text using logits post processor</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference_customize.html">Generate text with customization</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_auto_parallel.html">Automatic Parallelism with LLM</a></li>
</ul>
</details></li>
<li class="toctree-l1"><a class="reference internal" href="../../../llm-api-examples/customization.html">Common Customizations</a></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../../../llm-api-examples/llm_api_examples.html">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="../../../llm-api-examples/llm_lookahead_decoding.html">Generate Text Using Lookahead Decoding</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_guided_decoding.html">Generate text with guided decoding</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_medusa_decoding.html">Generate Text Using Medusa Decoding</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_multilora.html">Generate text with multiple LoRA adapters</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference_async_streaming.html">Generate Text in Streaming</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference.html">Generate text</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_quantization.html">Generation with Quantization</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference_distributed.html">Distributed LLM Generation</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference_async.html">Generate Text Asynchronously</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_logits_processor.html">Control generated text using logits post processor</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_inference_customize.html">Generate text with customization</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../llm-api-examples/llm_auto_parallel.html">Automatic Parallelism with LLM</a></li>
</ul>
</details></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Model Definition API</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../python-api/tensorrt_llm.layers.html">Layers</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../python-api/tensorrt_llm.functional.html">Functionals</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../python-api/tensorrt_llm.models.html">Models</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../python-api/tensorrt_llm.plugin.html">Plugin</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../python-api/tensorrt_llm.quantization.html">Quantization</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../python-api/tensorrt_llm.runtime.html">Runtime</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">C++ API</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../_cpp_gen/executor.html">Executor</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../_cpp_gen/runtime.html">Runtime</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Command-Line Reference</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../commands/trtllm-build.html">trtllm-build</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../commands/trtllm-serve.html">trtllm-serve</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Architecture</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/overview.html">TensorRT-LLM Architecture</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/core-concepts.html">Model Definition</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/checkpoint.html">TensorRT-LLM Checkpoint</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/workflow.html">TensorRT-LLM Build Workflow</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/add-model.html">Adding a Model</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Advanced</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/gpt-attention.html">Multi-Head, Multi-Query, and Group-Query Attention</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/gpt-runtime.html">C++ GPT Runtime</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/executor.html">Executor API</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/graph-rewriting.html">Graph Rewriting Module</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/inference-request.html">Inference Request</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/lora.html">Run gpt-2b + LoRA using GptManager / cpp runtime</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/expert-parallelism.html">Expert Parallelism in TensorRT-LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/kv-cache-reuse.html">KV cache reuse</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/speculative-decoding.html">Speculative Sampling</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/disaggregated-service.html">Disaggregated-Service (experimental)</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Performance</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../performance/perf-overview.html">Overview</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../performance/perf-benchmarking.html">Benchmarking</a></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="../../../performance/performance-tuning-guide/index.html">Performance Tuning Guide</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="../../../performance/performance-tuning-guide/benchmarking-default-performance.html">Benchmarking Default Performance</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../performance/performance-tuning-guide/useful-build-time-flags.html">Useful Build-Time Flags</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../performance/performance-tuning-guide/tuning-max-batch-size-and-max-num-tokens.html">Tuning Max Batch Size and Max Num Tokens</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../performance/performance-tuning-guide/deciding-model-sharding-strategy.html">Deciding Model Sharding Strategy</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../performance/performance-tuning-guide/fp8-quantization.html">FP8 Quantization</a></li>
<li class="toctree-l2"><a class="reference internal" href="../../../performance/performance-tuning-guide/useful-runtime-flags.html">Useful Runtime Options</a></li>
</ul>
</details></li>
<li class="toctree-l1"><a class="reference internal" href="../../../performance/perf-analysis.html">Performance Analysis</a></li>
</ul>
<p aria-level="2" class="caption" role="heading"><span class="caption-text">Reference</span></p>
<ul class="nav bd-sidenav">
<li class="toctree-l1"><a class="reference internal" href="../../../reference/troubleshooting.html">Troubleshooting</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../reference/support-matrix.html">Support Matrix</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../reference/precision.html">Numerical Precision</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../reference/memory.html">Memory Usage of TensorRT-LLM</a></li>
</ul>
<p 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/H100vsA100.html">H100 has 4.6x A100 Performance in TensorRT-LLM, achieving 10,000 tok/s at 100ms to first token</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/H200launch.html">H200 achieves nearly 12,000 tokens/sec on Llama2-13B with TensorRT-LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/Falcon180B-H200.html">Falcon-180B on a single H200 GPU with INT4 AWQ, and 6.7x faster Llama-70B over A100</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/quantization-in-TRT-LLM.html">Speed up inference with SOTA quantization techniques in TRT-LLM</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/XQA-kernel.html">New XQA-kernel provides 2.4x more Llama-70B throughput within the same latency budget</a></li>
</ul>
</div>
</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"><a href="../../index.html" class="nav-link">Module code</a></li>
<li class="breadcrumb-item active" aria-current="page"><span class="ellipsis">tensorrt_llm.runtime.model_runner</span></li>
</ul>
</nav>
</div>
</div>
</div>
</div>
<div id="searchbox"></div>
<article class="bd-article">
<h1>Source code for tensorrt_llm.runtime.model_runner</h1><div class="highlight"><pre>
<span></span><span class="c1"># SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION &amp; AFFILIATES. All rights reserved.</span>
<span class="c1"># SPDX-License-Identifier: Apache-2.0</span>
<span class="c1">#</span>
<span class="c1"># Licensed under the Apache License, Version 2.0 (the &quot;License&quot;);</span>
<span class="c1"># you may not use this file except in compliance with the License.</span>
<span class="c1"># You may obtain a copy of the License at</span>
<span class="c1">#</span>
<span class="c1"># http://www.apache.org/licenses/LICENSE-2.0</span>
<span class="c1">#</span>
<span class="c1"># Unless required by applicable law or agreed to in writing, software</span>
<span class="c1"># distributed under the License is distributed on an &quot;AS IS&quot; BASIS,</span>
<span class="c1"># WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
<span class="c1"># See the License for the specific language governing permissions and</span>
<span class="c1"># limitations under the License.</span>
<span class="kn">import</span> <span class="nn">copy</span>
<span class="kn">import</span> <span class="nn">json</span>
<span class="kn">import</span> <span class="nn">math</span>
<span class="kn">from</span> <span class="nn">pathlib</span> <span class="kn">import</span> <span class="n">Path</span>
<span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">List</span><span class="p">,</span> <span class="n">Optional</span><span class="p">,</span> <span class="n">Tuple</span><span class="p">,</span> <span class="n">Union</span>
<span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
<span class="kn">import</span> <span class="nn">tensorrt</span> <span class="k">as</span> <span class="nn">trt</span>
<span class="kn">import</span> <span class="nn">torch</span>
<span class="kn">from</span> <span class="nn">..</span> <span class="kn">import</span> <span class="n">profiler</span>
<span class="kn">from</span> <span class="nn">.._utils</span> <span class="kn">import</span> <span class="n">mpi_comm</span><span class="p">,</span> <span class="n">mpi_world_size</span><span class="p">,</span> <span class="n">numpy_to_torch</span>
<span class="kn">from</span> <span class="nn">..bindings</span> <span class="kn">import</span> <span class="n">KVCacheType</span><span class="p">,</span> <span class="n">MpiComm</span>
<span class="kn">from</span> <span class="nn">..bindings.executor</span> <span class="kn">import</span> <span class="n">Executor</span>
<span class="kn">from</span> <span class="nn">..builder</span> <span class="kn">import</span> <span class="n">Engine</span><span class="p">,</span> <span class="n">EngineConfig</span><span class="p">,</span> <span class="n">get_engine_version</span>
<span class="kn">from</span> <span class="nn">..logger</span> <span class="kn">import</span> <span class="n">logger</span>
<span class="kn">from</span> <span class="nn">..mapping</span> <span class="kn">import</span> <span class="n">Mapping</span>
<span class="kn">from</span> <span class="nn">..quantization</span> <span class="kn">import</span> <span class="n">QuantMode</span>
<span class="kn">from</span> <span class="nn">.generation</span> <span class="kn">import</span> <span class="p">(</span><span class="n">DISABLE_TORCH_DEVICE_SET</span><span class="p">,</span> <span class="n">ChatGLMGenerationSession</span><span class="p">,</span>
<span class="n">GenerationSession</span><span class="p">,</span> <span class="n">LogitsProcessor</span><span class="p">,</span> <span class="n">LoraManager</span><span class="p">,</span>
<span class="n">ModelConfig</span><span class="p">,</span> <span class="n">QWenForCausalLMGenerationSession</span><span class="p">,</span>
<span class="n">SamplingConfig</span><span class="p">,</span> <span class="n">StoppingCriteria</span><span class="p">,</span> <span class="n">to_word_list_format</span><span class="p">)</span>
<span class="k">def</span> <span class="nf">get_engine_name</span><span class="p">(</span><span class="n">model</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="n">dtype</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="n">tp_size</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span> <span class="n">pp_size</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
<span class="n">rank</span><span class="p">:</span> <span class="nb">int</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">str</span><span class="p">:</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> Get the serialized engine file name.</span>
<span class="sd"> Args:</span>
<span class="sd"> model (str):</span>
<span class="sd"> Model name, e.g., bloom, gpt.</span>
<span class="sd"> dtype (str):</span>
<span class="sd"> Data type, e.g., float32, float16, bfloat16,</span>
<span class="sd"> tp_size (int):</span>
<span class="sd"> The size of tensor parallel.</span>
<span class="sd"> pp_size (int):</span>
<span class="sd"> The size of pipeline parallel.</span>
<span class="sd"> rank (int):</span>
<span class="sd"> The rank id.</span>
<span class="sd"> Returns:</span>
<span class="sd"> str: The serialized engine file name.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="k">if</span> <span class="n">pp_size</span> <span class="o">==</span> <span class="mi">1</span><span class="p">:</span>
<span class="k">return</span> <span class="s1">&#39;</span><span class="si">{}</span><span class="s1">_</span><span class="si">{}</span><span class="s1">_tp</span><span class="si">{}</span><span class="s1">_rank</span><span class="si">{}</span><span class="s1">.engine&#39;</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">dtype</span><span class="p">,</span> <span class="n">tp_size</span><span class="p">,</span> <span class="n">rank</span><span class="p">)</span>
<span class="k">return</span> <span class="s1">&#39;</span><span class="si">{}</span><span class="s1">_</span><span class="si">{}</span><span class="s1">_tp</span><span class="si">{}</span><span class="s1">_pp</span><span class="si">{}</span><span class="s1">_rank</span><span class="si">{}</span><span class="s1">.engine&#39;</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">dtype</span><span class="p">,</span> <span class="n">tp_size</span><span class="p">,</span>
<span class="n">pp_size</span><span class="p">,</span> <span class="n">rank</span><span class="p">)</span>
<span class="k">def</span> <span class="nf">read_config</span><span class="p">(</span><span class="n">config_path</span><span class="p">:</span> <span class="n">Path</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">Tuple</span><span class="p">[</span><span class="n">ModelConfig</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"> Read the engine config file and create a ModelConfig instance, return the ModelConfig instance</span>
<span class="sd"> and other config fields in a dict.</span>
<span class="sd"> Args:</span>
<span class="sd"> config_path (Path):</span>
<span class="sd"> The path of engine config file.</span>
<span class="sd"> Returns:</span>
<span class="sd"> Tuple[ModelConfig, dict]: A ModelConfig instance and other config fields.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="n">config_path</span><span class="p">,</span> <span class="s1">&#39;r&#39;</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
<span class="n">config</span> <span class="o">=</span> <span class="n">json</span><span class="o">.</span><span class="n">load</span><span class="p">(</span><span class="n">f</span><span class="p">)</span>
<span class="k">return</span> <span class="n">_builder_to_model_config</span><span class="p">(</span><span class="n">config</span><span class="p">)</span>
<span class="k">def</span> <span class="nf">_builder_to_model_config</span><span class="p">(</span><span class="n">config</span><span class="p">:</span> <span class="nb">dict</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">Tuple</span><span class="p">[</span><span class="n">ModelConfig</span><span class="p">,</span> <span class="nb">dict</span><span class="p">]:</span>
<span class="n">builder_config</span> <span class="o">=</span> <span class="n">config</span><span class="p">[</span><span class="s1">&#39;builder_config&#39;</span><span class="p">]</span>
<span class="n">model_name</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;name&#39;</span><span class="p">]</span>
<span class="n">dtype</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;precision&#39;</span><span class="p">]</span>
<span class="n">tp_size</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;tensor_parallel&#39;</span><span class="p">]</span>
<span class="n">pp_size</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;pipeline_parallel&#39;</span><span class="p">,</span> <span class="mi">1</span><span class="p">)</span>
<span class="n">kv_cache_type</span> <span class="o">=</span> <span class="n">KVCacheType</span><span class="p">(</span><span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;kv_cache_type&#39;</span><span class="p">))</span>
<span class="n">world_size</span> <span class="o">=</span> <span class="n">tp_size</span> <span class="o">*</span> <span class="n">pp_size</span>
<span class="k">assert</span> <span class="n">world_size</span> <span class="o">==</span> <span class="n">mpi_world_size</span><span class="p">(),</span> \
<span class="sa">f</span><span class="s1">&#39;Engine world size (</span><span class="si">{</span><span class="n">tp_size</span><span class="si">}</span><span class="s1"> * </span><span class="si">{</span><span class="n">pp_size</span><span class="si">}</span><span class="s1">) != Runtime world size (</span><span class="si">{</span><span class="n">mpi_world_size</span><span class="p">()</span><span class="si">}</span><span class="s1">)&#39;</span>
<span class="n">num_heads</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;num_heads&#39;</span><span class="p">]</span>
<span class="k">assert</span> <span class="n">num_heads</span> <span class="o">%</span> <span class="n">tp_size</span> <span class="o">==</span> <span class="mi">0</span><span class="p">,</span> \
<span class="sa">f</span><span class="s2">&quot;The number of heads (</span><span class="si">{</span><span class="n">num_heads</span><span class="si">}</span><span class="s2">) is not a multiple of tp_size (</span><span class="si">{</span><span class="n">tp_size</span><span class="si">}</span><span class="s2">)&quot;</span>
<span class="n">num_kv_heads</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;num_kv_heads&#39;</span><span class="p">,</span> <span class="n">num_heads</span><span class="p">)</span>
<span class="c1"># TODO: multi_query_mode should be removed</span>
<span class="n">multi_query_mode</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;multi_query_mode&#39;</span><span class="p">,</span> <span class="kc">False</span><span class="p">)</span>
<span class="k">if</span> <span class="n">multi_query_mode</span><span class="p">:</span>
<span class="n">logger</span><span class="o">.</span><span class="n">warning</span><span class="p">(</span>
<span class="s2">&quot;`multi_query_mode` config is deprecated. Please rebuild the engine.&quot;</span>
<span class="p">)</span>
<span class="c1"># num_kv_heads, if exists in config, should override multi_query_mode</span>
<span class="k">if</span> <span class="n">multi_query_mode</span> <span class="ow">and</span> <span class="p">(</span><span class="s1">&#39;num_kv_heads&#39;</span> <span class="ow">not</span> <span class="ow">in</span> <span class="n">builder_config</span><span class="p">):</span>
<span class="n">num_kv_heads</span> <span class="o">=</span> <span class="mi">1</span>
<span class="n">num_heads</span> <span class="o">=</span> <span class="n">num_heads</span> <span class="o">//</span> <span class="n">tp_size</span>
<span class="n">num_kv_heads</span> <span class="o">=</span> <span class="p">(</span><span class="n">num_kv_heads</span> <span class="o">+</span> <span class="n">tp_size</span> <span class="o">-</span> <span class="mi">1</span><span class="p">)</span> <span class="o">//</span> <span class="n">tp_size</span>
<span class="n">head_size</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;head_size&#39;</span><span class="p">,</span> <span class="kc">None</span><span class="p">)</span>
<span class="n">hidden_size</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;hidden_size&#39;</span><span class="p">]</span> <span class="o">//</span> <span class="n">tp_size</span>
<span class="n">vocab_size</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;vocab_size&#39;</span><span class="p">]</span>
<span class="n">num_layers</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;num_layers&#39;</span><span class="p">]</span>
<span class="n">max_batch_size</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;max_batch_size&#39;</span><span class="p">]</span>
<span class="n">max_beam_width</span> <span class="o">=</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;max_beam_width&#39;</span><span class="p">]</span>
<span class="n">cross_attention</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;cross_attention&#39;</span><span class="p">,</span> <span class="kc">False</span><span class="p">)</span>
<span class="n">has_position_embedding</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;has_position_embedding&#39;</span><span class="p">,</span> <span class="kc">True</span><span class="p">)</span>
<span class="n">has_token_type_embedding</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;has_token_type_embedding&#39;</span><span class="p">,</span>
<span class="kc">False</span><span class="p">)</span>
<span class="n">gather_context_logits</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;gather_context_logits&#39;</span><span class="p">,</span> <span class="kc">False</span><span class="p">)</span>
<span class="n">gather_generation_logits</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;gather_generation_logits&#39;</span><span class="p">,</span>
<span class="kc">False</span><span class="p">)</span>
<span class="n">max_prompt_embedding_table_size</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span>
<span class="s1">&#39;max_prompt_embedding_table_size&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span>
<span class="n">quant_mode</span> <span class="o">=</span> <span class="n">QuantMode</span><span class="p">(</span><span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;quant_mode&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">))</span>
<span class="n">lora_target_modules</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;lora_target_modules&#39;</span><span class="p">)</span>
<span class="n">lora_trtllm_modules_to_hf_modules</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span>
<span class="s1">&#39;trtllm_modules_to_hf_modules&#39;</span><span class="p">)</span>
<span class="n">max_medusa_token_len</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;max_draft_len&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span>
<span class="n">num_medusa_heads</span> <span class="o">=</span> <span class="n">builder_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;num_medusa_heads&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span>
<span class="n">skip_cross_attn_blocks</span> <span class="o">=</span> <span class="nb">bool</span><span class="p">(</span><span class="n">config</span><span class="p">[</span><span class="s1">&#39;pretrained_config&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">get</span><span class="p">(</span>
<span class="s1">&#39;skip_cross_attn_blocks&#39;</span><span class="p">,</span> <span class="kc">False</span><span class="p">))</span>
<span class="c1"># ReDrafter</span>
<span class="n">redrafter_num_beams</span> <span class="o">=</span> <span class="n">config</span><span class="p">[</span><span class="s1">&#39;pretrained_config&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">get</span><span class="p">(</span>
<span class="s1">&#39;redrafter_num_beams&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span>
<span class="n">redrafter_draft_len_per_beam</span> <span class="o">=</span> <span class="n">config</span><span class="p">[</span><span class="s1">&#39;pretrained_config&#39;</span><span class="p">]</span><span class="o">.</span><span class="n">get</span><span class="p">(</span>
<span class="s1">&#39;redrafter_draft_len_per_beam&#39;</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span>
<span class="n">plugin_config</span> <span class="o">=</span> <span class="n">config</span><span class="p">[</span><span class="s1">&#39;plugin_config&#39;</span><span class="p">]</span>
<span class="n">use_gpt_attention_plugin</span> <span class="o">=</span> <span class="nb">bool</span><span class="p">(</span><span class="n">plugin_config</span><span class="p">[</span><span class="s1">&#39;gpt_attention_plugin&#39;</span><span class="p">])</span>
<span class="n">mamba_conv1d_plugin</span> <span class="o">=</span> <span class="nb">bool</span><span class="p">(</span><span class="n">plugin_config</span><span class="p">[</span><span class="s1">&#39;mamba_conv1d_plugin&#39;</span><span class="p">])</span>
<span class="n">remove_input_padding</span> <span class="o">=</span> <span class="n">plugin_config</span><span class="p">[</span><span class="s1">&#39;remove_input_padding&#39;</span><span class="p">]</span>
<span class="n">paged_state</span> <span class="o">=</span> <span class="n">plugin_config</span><span class="p">[</span><span class="s1">&#39;paged_state&#39;</span><span class="p">]</span>
<span class="n">tokens_per_block</span> <span class="o">=</span> <span class="n">plugin_config</span><span class="p">[</span><span class="s1">&#39;tokens_per_block&#39;</span><span class="p">]</span>
<span class="n">lora_plugin</span> <span class="o">=</span> <span class="n">plugin_config</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;lora_plugin&#39;</span><span class="p">)</span>
<span class="n">model_config</span> <span class="o">=</span> <span class="n">ModelConfig</span><span class="p">(</span>
<span class="n">max_batch_size</span><span class="o">=</span><span class="n">max_batch_size</span><span class="p">,</span>
<span class="n">max_beam_width</span><span class="o">=</span><span class="n">max_beam_width</span><span class="p">,</span>
<span class="n">vocab_size</span><span class="o">=</span><span class="n">vocab_size</span><span class="p">,</span>
<span class="n">num_layers</span><span class="o">=</span><span class="n">num_layers</span><span class="p">,</span>
<span class="n">num_heads</span><span class="o">=</span><span class="n">num_heads</span><span class="p">,</span>
<span class="n">num_kv_heads</span><span class="o">=</span><span class="n">num_kv_heads</span><span class="p">,</span>
<span class="n">hidden_size</span><span class="o">=</span><span class="n">hidden_size</span><span class="p">,</span>
<span class="n">head_size</span><span class="o">=</span><span class="n">head_size</span><span class="p">,</span>
<span class="n">gpt_attention_plugin</span><span class="o">=</span><span class="n">use_gpt_attention_plugin</span><span class="p">,</span>
<span class="n">mamba_conv1d_plugin</span><span class="o">=</span><span class="n">mamba_conv1d_plugin</span><span class="p">,</span>
<span class="n">remove_input_padding</span><span class="o">=</span><span class="n">remove_input_padding</span><span class="p">,</span>
<span class="n">model_name</span><span class="o">=</span><span class="n">model_name</span><span class="p">,</span>
<span class="n">kv_cache_type</span><span class="o">=</span><span class="n">kv_cache_type</span><span class="p">,</span>
<span class="n">paged_state</span><span class="o">=</span><span class="n">paged_state</span><span class="p">,</span>
<span class="n">cross_attention</span><span class="o">=</span><span class="n">cross_attention</span><span class="p">,</span>
<span class="n">has_position_embedding</span><span class="o">=</span><span class="n">has_position_embedding</span><span class="p">,</span>
<span class="n">has_token_type_embedding</span><span class="o">=</span><span class="n">has_token_type_embedding</span><span class="p">,</span>
<span class="n">tokens_per_block</span><span class="o">=</span><span class="n">tokens_per_block</span><span class="p">,</span>
<span class="n">max_prompt_embedding_table_size</span><span class="o">=</span><span class="n">max_prompt_embedding_table_size</span><span class="p">,</span>
<span class="n">quant_mode</span><span class="o">=</span><span class="n">quant_mode</span><span class="p">,</span>
<span class="n">gather_context_logits</span><span class="o">=</span><span class="n">gather_context_logits</span><span class="p">,</span>
<span class="n">gather_generation_logits</span><span class="o">=</span><span class="n">gather_generation_logits</span><span class="p">,</span>
<span class="n">dtype</span><span class="o">=</span><span class="n">dtype</span><span class="p">,</span>
<span class="n">lora_plugin</span><span class="o">=</span><span class="n">lora_plugin</span><span class="p">,</span>
<span class="n">lora_target_modules</span><span class="o">=</span><span class="n">lora_target_modules</span><span class="p">,</span>
<span class="n">trtllm_modules_to_hf_modules</span><span class="o">=</span><span class="n">lora_trtllm_modules_to_hf_modules</span><span class="p">,</span>
<span class="n">num_medusa_heads</span><span class="o">=</span><span class="n">num_medusa_heads</span><span class="p">,</span>
<span class="n">max_medusa_tokens</span><span class="o">=</span><span class="n">max_medusa_token_len</span><span class="p">,</span>
<span class="n">skip_cross_attn_blocks</span><span class="o">=</span><span class="n">skip_cross_attn_blocks</span><span class="p">,</span>
<span class="c1"># ReDrafter</span>
<span class="n">redrafter_num_beams</span><span class="o">=</span><span class="n">redrafter_num_beams</span><span class="p">,</span>
<span class="n">redrafter_draft_len_per_beam</span><span class="o">=</span><span class="n">redrafter_draft_len_per_beam</span><span class="p">,</span>
<span class="p">)</span>
<span class="n">other_config</span> <span class="o">=</span> <span class="p">{</span>
<span class="s1">&#39;world_size&#39;</span><span class="p">:</span> <span class="n">world_size</span><span class="p">,</span>
<span class="s1">&#39;tp_size&#39;</span><span class="p">:</span> <span class="n">tp_size</span><span class="p">,</span>
<span class="s1">&#39;pp_size&#39;</span><span class="p">:</span> <span class="n">pp_size</span><span class="p">,</span>
<span class="s1">&#39;max_batch_size&#39;</span><span class="p">:</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;max_batch_size&#39;</span><span class="p">],</span>
<span class="s1">&#39;max_input_len&#39;</span><span class="p">:</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;max_input_len&#39;</span><span class="p">],</span>
<span class="s1">&#39;max_output_len&#39;</span><span class="p">:</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;max_output_len&#39;</span><span class="p">],</span>
<span class="s1">&#39;max_beam_width&#39;</span><span class="p">:</span> <span class="n">builder_config</span><span class="p">[</span><span class="s1">&#39;max_beam_width&#39;</span><span class="p">]</span>
<span class="p">}</span>
<span class="k">return</span> <span class="n">model_config</span><span class="p">,</span> <span class="n">other_config</span>
<span class="k">def</span> <span class="nf">_engine_config_to_model_config</span><span class="p">(</span><span class="n">engine_config</span><span class="p">:</span> <span class="n">EngineConfig</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="n">pretrained_config</span> <span class="o">=</span> <span class="n">engine_config</span><span class="o">.</span><span class="n">pretrained_config</span>
<span class="n">build_config</span> <span class="o">=</span> <span class="n">engine_config</span><span class="o">.</span><span class="n">build_config</span>
<span class="n">tp_size</span> <span class="o">=</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">tp_size</span>
<span class="n">num_heads</span> <span class="o">=</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_attention_heads</span> <span class="o">//</span> <span class="n">tp_size</span>
<span class="n">num_kv_heads</span> <span class="o">=</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_key_value_heads</span>
<span class="n">num_kv_heads</span> <span class="o">=</span> <span class="p">(</span><span class="n">num_kv_heads</span> <span class="o">+</span> <span class="n">tp_size</span> <span class="o">-</span> <span class="mi">1</span><span class="p">)</span> <span class="o">//</span> <span class="n">tp_size</span>
<span class="n">hidden_size</span> <span class="o">=</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">hidden_size</span> <span class="o">//</span> <span class="n">tp_size</span>
<span class="n">head_size</span> <span class="o">=</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">head_size</span>
<span class="n">rnn_config_items</span> <span class="o">=</span> <span class="p">[</span>
<span class="s1">&#39;conv_kernel&#39;</span><span class="p">,</span> <span class="s1">&#39;layer_types&#39;</span><span class="p">,</span> <span class="s1">&#39;rnn_hidden_size&#39;</span><span class="p">,</span> <span class="s1">&#39;state_size&#39;</span><span class="p">,</span>
<span class="s1">&#39;state_dtype&#39;</span><span class="p">,</span> <span class="s1">&#39;rnn_head_size&#39;</span><span class="p">,</span> <span class="s1">&#39;rnn_conv_dim_size&#39;</span>
<span class="p">]</span>
<span class="n">rnn_configs_kwargs</span> <span class="o">=</span> <span class="p">{}</span>
<span class="k">for</span> <span class="n">item</span> <span class="ow">in</span> <span class="n">rnn_config_items</span><span class="p">:</span>
<span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span> <span class="n">item</span><span class="p">):</span>
<span class="n">rnn_configs_kwargs</span><span class="p">[</span><span class="n">item</span><span class="p">]</span> <span class="o">=</span> <span class="nb">getattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span> <span class="n">item</span><span class="p">)</span>
<span class="k">if</span> <span class="ow">not</span> <span class="nb">hasattr</span><span class="p">(</span><span class="n">build_config</span><span class="p">,</span> <span class="s1">&#39;kv_cache_type&#39;</span><span class="p">):</span>
<span class="n">logger</span><span class="o">.</span><span class="n">Warning</span><span class="p">(</span>
<span class="s1">&#39;Build config doesn</span><span class="se">\&#39;</span><span class="s1">t have kv_cache_type, you might need to rebuild your enigne.&#39;</span>
<span class="p">)</span>
<span class="c1"># TODO(oargov): this is a hack, make it prettier!</span>
<span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span> <span class="s2">&quot;num_kv_heads_per_layer&quot;</span><span class="p">):</span>
<span class="n">num_kv_heads_per_layer</span> <span class="o">=</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_kv_heads_per_layer</span>
<span class="k">elif</span> <span class="nb">hasattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span> <span class="s2">&quot;get_layer_num_kv_heads&quot;</span><span class="p">):</span>
<span class="c1"># each layer has a different number of kv heads</span>
<span class="n">attention_layers</span> <span class="o">=</span> <span class="p">[</span>
<span class="n">layer_idx</span> <span class="k">for</span> <span class="n">layer_idx</span><span class="p">,</span> <span class="n">layer_type</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span>
<span class="n">pretrained_config</span><span class="o">.</span><span class="n">layer_types</span><span class="p">)</span> <span class="k">if</span> <span class="n">layer_type</span> <span class="o">==</span> <span class="s2">&quot;attention&quot;</span>
<span class="p">]</span> <span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span> <span class="s2">&quot;layer_types&quot;</span><span class="p">)</span> <span class="k">else</span> <span class="nb">list</span><span class="p">(</span>
<span class="nb">range</span><span class="p">(</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_hidden_layers</span><span class="p">))</span>
<span class="n">num_kv_heads_per_layer</span> <span class="o">=</span> <span class="p">[</span>
<span class="n">pretrained_config</span><span class="o">.</span><span class="n">get_layer_num_kv_heads</span><span class="p">(</span><span class="n">layer_idx</span><span class="p">)</span>
<span class="k">if</span> <span class="n">layer_idx</span> <span class="ow">in</span> <span class="n">attention_layers</span> <span class="k">else</span> <span class="mi">0</span>
<span class="k">for</span> <span class="n">layer_idx</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_hidden_layers</span><span class="p">)</span>
<span class="p">]</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">num_kv_heads_per_layer</span> <span class="o">=</span> <span class="kc">None</span>
<span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span> <span class="s2">&quot;num_kv_heads_per_cross_attn_layer&quot;</span><span class="p">):</span>
<span class="n">num_kv_heads_per_cross_attn_layer</span> <span class="o">=</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_kv_heads_per_cross_attn_layer</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">num_kv_heads_per_cross_attn_layer</span> <span class="o">=</span> <span class="kc">None</span>
<span class="k">return</span> <span class="n">ModelConfig</span><span class="p">(</span>
<span class="n">max_batch_size</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">max_batch_size</span><span class="p">,</span>
<span class="n">max_beam_width</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">max_beam_width</span><span class="p">,</span>
<span class="n">vocab_size</span><span class="o">=</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">vocab_size</span><span class="p">,</span>
<span class="n">num_layers</span><span class="o">=</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_hidden_layers</span><span class="p">,</span>
<span class="n">num_heads</span><span class="o">=</span><span class="n">num_heads</span><span class="p">,</span>
<span class="n">num_kv_heads</span><span class="o">=</span><span class="n">num_kv_heads</span><span class="p">,</span>
<span class="n">hidden_size</span><span class="o">=</span><span class="n">hidden_size</span><span class="p">,</span>
<span class="n">head_size</span><span class="o">=</span><span class="n">head_size</span><span class="p">,</span>
<span class="n">gpt_attention_plugin</span><span class="o">=</span><span class="nb">bool</span><span class="p">(</span>
<span class="n">build_config</span><span class="o">.</span><span class="n">plugin_config</span><span class="o">.</span><span class="n">gpt_attention_plugin</span><span class="p">),</span>
<span class="n">mamba_conv1d_plugin</span><span class="o">=</span><span class="nb">bool</span><span class="p">(</span>
<span class="n">build_config</span><span class="o">.</span><span class="n">plugin_config</span><span class="o">.</span><span class="n">mamba_conv1d_plugin</span><span class="p">),</span>
<span class="n">remove_input_padding</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">plugin_config</span><span class="o">.</span><span class="n">remove_input_padding</span><span class="p">,</span>
<span class="n">paged_state</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">plugin_config</span><span class="o">.</span><span class="n">paged_state</span><span class="p">,</span>
<span class="n">tokens_per_block</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">plugin_config</span><span class="o">.</span><span class="n">tokens_per_block</span><span class="p">,</span>
<span class="n">quant_mode</span><span class="o">=</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">quant_mode</span><span class="p">,</span>
<span class="n">gather_context_logits</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">gather_context_logits</span><span class="p">,</span>
<span class="n">gather_generation_logits</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">gather_generation_logits</span><span class="p">,</span>
<span class="n">dtype</span><span class="o">=</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">dtype</span><span class="p">,</span>
<span class="n">max_prompt_embedding_table_size</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span>
<span class="n">max_prompt_embedding_table_size</span><span class="p">,</span>
<span class="n">lora_plugin</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">plugin_config</span><span class="o">.</span><span class="n">lora_plugin</span><span class="p">,</span>
<span class="n">lora_target_modules</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">lora_config</span><span class="o">.</span><span class="n">lora_target_modules</span><span class="p">,</span>
<span class="n">trtllm_modules_to_hf_modules</span><span class="o">=</span><span class="n">build_config</span><span class="o">.</span><span class="n">lora_config</span><span class="o">.</span>
<span class="n">trtllm_modules_to_hf_modules</span><span class="p">,</span>
<span class="n">max_medusa_tokens</span><span class="o">=</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">max_draft_len</span> <span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span>
<span class="n">pretrained_config</span><span class="p">,</span> <span class="s1">&#39;max_draft_len&#39;</span><span class="p">)</span> <span class="k">else</span> <span class="mi">0</span><span class="p">,</span>
<span class="n">num_medusa_heads</span><span class="o">=</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_medusa_heads</span> <span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span>
<span class="n">pretrained_config</span><span class="p">,</span> <span class="s1">&#39;num_medusa_heads&#39;</span><span class="p">)</span> <span class="k">else</span> <span class="mi">0</span><span class="p">,</span>
<span class="o">**</span><span class="n">rnn_configs_kwargs</span><span class="p">,</span>
<span class="n">num_kv_heads_per_layer</span><span class="o">=</span><span class="n">num_kv_heads_per_layer</span><span class="p">,</span>
<span class="n">num_kv_heads_per_cross_attn_layer</span><span class="o">=</span><span class="n">num_kv_heads_per_cross_attn_layer</span><span class="p">,</span>
<span class="n">redrafter_num_beams</span><span class="o">=</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">redrafter_num_beams</span> <span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span>
<span class="n">pretrained_config</span><span class="p">,</span> <span class="s1">&#39;redrafter_num_beams&#39;</span><span class="p">)</span> <span class="k">else</span> <span class="mi">0</span><span class="p">,</span>
<span class="n">redrafter_draft_len_per_beam</span><span class="o">=</span><span class="n">pretrained_config</span><span class="o">.</span>
<span class="n">redrafter_draft_len_per_beam</span>
<span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span> <span class="s1">&#39;redrafter_draft_len_per_beam&#39;</span><span class="p">)</span> <span class="k">else</span> <span class="mi">0</span><span class="p">,</span>
<span class="n">kv_cache_type</span><span class="o">=</span><span class="nb">getattr</span><span class="p">(</span><span class="n">build_config</span><span class="p">,</span> <span class="s1">&#39;kv_cache_type&#39;</span><span class="p">,</span>
<span class="n">KVCacheType</span><span class="o">.</span><span class="n">CONTINUOUS</span><span class="p">),</span>
<span class="n">cross_attention</span><span class="o">=</span><span class="nb">getattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span> <span class="s1">&#39;cross_attention&#39;</span><span class="p">,</span> <span class="kc">False</span><span class="p">),</span>
<span class="n">has_position_embedding</span><span class="o">=</span><span class="nb">getattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span>
<span class="s1">&#39;has_position_embedding&#39;</span><span class="p">,</span> <span class="kc">True</span><span class="p">),</span>
<span class="n">skip_cross_attn_blocks</span><span class="o">=</span><span class="nb">getattr</span><span class="p">(</span><span class="n">pretrained_config</span><span class="p">,</span>
<span class="s1">&#39;skip_cross_attn_blocks&#39;</span><span class="p">,</span> <span class="kc">False</span><span class="p">),</span>
<span class="o">**</span><span class="n">kwargs</span><span class="p">)</span>
<span class="k">class</span> <span class="nc">ModelRunnerMixin</span><span class="p">:</span>
<span class="k">def</span> <span class="nf">_check_inputs</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">batch_input_ids</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">],</span>
<span class="n">sampling_config</span><span class="p">:</span> <span class="n">SamplingConfig</span><span class="p">):</span>
<span class="n">batch_size</span> <span class="o">=</span> <span class="nb">len</span><span class="p">(</span><span class="n">batch_input_ids</span><span class="p">)</span>
<span class="k">if</span> <span class="n">batch_size</span> <span class="o">&gt;</span> <span class="bp">self</span><span class="o">.</span><span class="n">max_batch_size</span><span class="p">:</span>
<span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span>
<span class="sa">f</span><span class="s2">&quot;Input batch size (</span><span class="si">{</span><span class="n">batch_size</span><span class="si">}</span><span class="s2">) exceeds the engine or specified limit (</span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">max_batch_size</span><span class="si">}</span><span class="s2">)&quot;</span>
<span class="p">)</span>
<span class="n">input_lengths</span> <span class="o">=</span> <span class="p">[</span><span class="n">x</span><span class="o">.</span><span class="n">size</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span> <span class="k">for</span> <span class="n">x</span> <span class="ow">in</span> <span class="n">batch_input_ids</span><span class="p">]</span>
<span class="n">max_length</span> <span class="o">=</span> <span class="nb">max</span><span class="p">(</span><span class="n">input_lengths</span><span class="p">)</span>
<span class="k">if</span> <span class="n">max_length</span> <span class="o">&gt;</span> <span class="bp">self</span><span class="o">.</span><span class="n">max_input_len</span><span class="p">:</span>
<span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span>
<span class="sa">f</span><span class="s2">&quot;Maximum input length (</span><span class="si">{</span><span class="n">max_length</span><span class="si">}</span><span class="s2">) exceeds the engine or specified limit (</span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">max_input_len</span><span class="si">}</span><span class="s2">)&quot;</span>
<span class="p">)</span>
<span class="k">if</span> <span class="n">max_length</span> <span class="o">+</span> <span class="n">sampling_config</span><span class="o">.</span><span class="n">max_new_tokens</span> <span class="o">&gt;</span> <span class="bp">self</span><span class="o">.</span><span class="n">max_seq_len</span><span class="p">:</span>
<span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span>
<span class="sa">f</span><span class="s2">&quot;Maximum input length (</span><span class="si">{</span><span class="n">max_length</span><span class="si">}</span><span class="s2">) + maximum new tokens (</span><span class="si">{</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">max_new_tokens</span><span class="si">}</span><span class="s2">) exceeds the engine or specified limit (</span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">max_seq_len</span><span class="si">}</span><span class="s2">)&quot;</span>
<span class="p">)</span>
<span class="k">if</span> <span class="n">sampling_config</span><span class="o">.</span><span class="n">num_beams</span> <span class="o">&gt;</span> <span class="bp">self</span><span class="o">.</span><span class="n">max_beam_width</span><span class="p">:</span>
<span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span>
<span class="sa">f</span><span class="s2">&quot;Num beams (</span><span class="si">{</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">num_beams</span><span class="si">}</span><span class="s2">) exceeds the engine or specified limit (</span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">max_beam_width</span><span class="si">}</span><span class="s2">)&quot;</span>
<span class="p">)</span>
<span class="k">def</span> <span class="nf">_prepare_inputs</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">batch_input_ids</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">],</span>
<span class="n">pad_id</span><span class="p">:</span> <span class="nb">int</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">Tuple</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">]:</span>
<span class="c1"># Cast to int32</span>
<span class="n">batch_input_ids</span> <span class="o">=</span> <span class="p">[</span><span class="n">x</span><span class="o">.</span><span class="n">type</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">int32</span><span class="p">)</span> <span class="k">for</span> <span class="n">x</span> <span class="ow">in</span> <span class="n">batch_input_ids</span><span class="p">]</span>
<span class="n">input_lengths</span> <span class="o">=</span> <span class="p">[</span><span class="n">x</span><span class="o">.</span><span class="n">size</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span> <span class="k">for</span> <span class="n">x</span> <span class="ow">in</span> <span class="n">batch_input_ids</span><span class="p">]</span>
<span class="n">max_length</span> <span class="o">=</span> <span class="nb">max</span><span class="p">(</span><span class="n">input_lengths</span><span class="p">)</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">remove_input_padding</span><span class="p">:</span>
<span class="n">batch_input_ids</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">concat</span><span class="p">(</span><span class="n">batch_input_ids</span><span class="p">)</span>
<span class="k">else</span><span class="p">:</span>
<span class="c1"># Right padding for trt-llm</span>
<span class="n">paddings</span> <span class="o">=</span> <span class="p">[</span>
<span class="n">torch</span><span class="o">.</span><span class="n">ones</span><span class="p">(</span><span class="n">max_length</span> <span class="o">-</span> <span class="n">l</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">int32</span><span class="p">)</span> <span class="o">*</span> <span class="n">pad_id</span>
<span class="k">for</span> <span class="n">l</span> <span class="ow">in</span> <span class="n">input_lengths</span>
<span class="p">]</span>
<span class="n">batch_input_ids</span> <span class="o">=</span> <span class="p">[</span>
<span class="n">torch</span><span class="o">.</span><span class="n">cat</span><span class="p">([</span><span class="n">x</span><span class="p">,</span> <span class="n">pad</span><span class="p">])</span> <span class="k">for</span> <span class="n">x</span><span class="p">,</span> <span class="n">pad</span> <span class="ow">in</span> <span class="nb">zip</span><span class="p">(</span><span class="n">batch_input_ids</span><span class="p">,</span> <span class="n">paddings</span><span class="p">)</span>
<span class="p">]</span>
<span class="n">batch_input_ids</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">stack</span><span class="p">(</span><span class="n">batch_input_ids</span><span class="p">)</span>
<span class="n">input_lengths</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="n">input_lengths</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">int32</span><span class="p">)</span>
<span class="k">return</span> <span class="n">batch_input_ids</span><span class="p">,</span> <span class="n">input_lengths</span>
<span class="k">def</span> <span class="nf">_prepare_outputs</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">outputs</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">dict</span><span class="p">],</span>
<span class="n">input_lengths</span><span class="p">:</span> <span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">dict</span><span class="p">:</span>
<span class="k">if</span> <span class="n">outputs</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">batch_size</span> <span class="o">=</span> <span class="n">input_lengths</span><span class="o">.</span><span class="n">size</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span>
<span class="k">if</span> <span class="s1">&#39;context_logits&#39;</span> <span class="ow">in</span> <span class="n">outputs</span><span class="p">:</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">has_pp</span><span class="p">():</span>
<span class="c1"># If pp size &gt; 1, the context logits and generation logits are both in last pp</span>
<span class="c1"># Last pp rank send context logits and generation logits to rank 0</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">is_last_pp_rank</span><span class="p">():</span>
<span class="n">context_logits</span> <span class="o">=</span> <span class="n">outputs</span><span class="p">[</span><span class="s1">&#39;context_logits&#39;</span><span class="p">]</span>
<span class="n">context_logits_host</span> <span class="o">=</span> <span class="n">context_logits</span><span class="o">.</span><span class="n">cpu</span><span class="p">()</span>
<span class="n">mpi_comm</span><span class="p">()</span><span class="o">.</span><span class="n">send</span><span class="p">(</span><span class="n">context_logits_host</span><span class="p">,</span> <span class="n">dest</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
<span class="k">elif</span> <span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">is_first_pp_rank</span><span class="p">():</span>
<span class="n">context_logits_host</span> <span class="o">=</span> <span class="n">mpi_comm</span><span class="p">()</span><span class="o">.</span><span class="n">recv</span><span class="p">(</span>
<span class="n">source</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">prev_pp_rank</span><span class="p">()</span>
<span class="p">)</span> <span class="c1"># Prev pp rank of rank=0 is the last pp</span>
<span class="n">context_logits</span> <span class="o">=</span> <span class="n">context_logits_host</span><span class="o">.</span><span class="n">to</span><span class="p">(</span>
<span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda:0&#39;</span><span class="p">))</span>
<span class="n">outputs</span><span class="p">[</span><span class="s1">&#39;context_logits&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="n">context_logits</span>
<span class="n">context_logits</span> <span class="o">=</span> <span class="n">outputs</span><span class="p">[</span><span class="s1">&#39;context_logits&#39;</span><span class="p">]</span>
<span class="n">context_logits_output</span> <span class="o">=</span> <span class="p">[]</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">remove_input_padding</span><span class="p">:</span>
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="p">,</span> <span class="n">Executor</span><span class="p">)</span> <span class="ow">and</span> <span class="n">batch_size</span> <span class="o">&gt;</span> <span class="mi">1</span><span class="p">:</span>
<span class="c1"># The starting position of the context logits buffer of each micro batch is separated</span>
<span class="n">num_batches</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">pp_size</span>
<span class="n">micro_batch_size</span> <span class="o">=</span> <span class="n">math</span><span class="o">.</span><span class="n">ceil</span><span class="p">(</span><span class="n">batch_size</span> <span class="o">/</span>
<span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">pp_size</span><span class="p">)</span>
<span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="n">num_batches</span><span class="p">):</span>
<span class="n">start_idx</span> <span class="o">=</span> <span class="n">i</span> <span class="o">*</span> <span class="n">micro_batch_size</span>
<span class="n">end_idx</span> <span class="o">=</span> <span class="nb">min</span><span class="p">(</span><span class="n">start_idx</span> <span class="o">+</span> <span class="n">micro_batch_size</span><span class="p">,</span>
<span class="n">batch_size</span><span class="p">)</span>
<span class="n">micro_context_logits</span> <span class="o">=</span> <span class="n">context_logits</span><span class="p">[</span>
<span class="n">start_idx</span><span class="p">:</span><span class="n">end_idx</span><span class="p">]</span>
<span class="n">micro_input_lengths</span> <span class="o">=</span> <span class="n">input_lengths</span><span class="p">[</span>
<span class="n">start_idx</span><span class="p">:</span><span class="n">end_idx</span><span class="p">]</span>
<span class="n">micro_context_logits</span> <span class="o">=</span> <span class="n">micro_context_logits</span><span class="o">.</span><span class="n">flatten</span><span class="p">(</span>
<span class="n">end_dim</span><span class="o">=-</span><span class="mi">2</span><span class="p">)</span>
<span class="n">seg_points</span> <span class="o">=</span> <span class="p">[</span><span class="mi">0</span><span class="p">]</span> <span class="o">+</span> <span class="n">micro_input_lengths</span><span class="o">.</span><span class="n">cumsum</span><span class="p">(</span>
<span class="n">dim</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span><span class="o">.</span><span class="n">tolist</span><span class="p">()</span>
<span class="n">context_logits_output</span> <span class="o">+=</span> <span class="p">[</span>
<span class="n">micro_context_logits</span><span class="p">[</span><span class="n">s</span><span class="p">:</span><span class="n">e</span><span class="p">]</span>
<span class="k">for</span> <span class="n">s</span><span class="p">,</span> <span class="n">e</span> <span class="ow">in</span> <span class="nb">zip</span><span class="p">(</span><span class="n">seg_points</span><span class="p">[:</span><span class="o">-</span><span class="mi">1</span><span class="p">],</span> <span class="n">seg_points</span><span class="p">[</span><span class="mi">1</span><span class="p">:])</span>
<span class="p">]</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">context_logits</span> <span class="o">=</span> <span class="n">context_logits</span><span class="o">.</span><span class="n">flatten</span><span class="p">(</span><span class="n">end_dim</span><span class="o">=-</span><span class="mi">2</span><span class="p">)</span>
<span class="n">seg_points</span> <span class="o">=</span> <span class="p">[</span><span class="mi">0</span><span class="p">]</span> <span class="o">+</span> <span class="n">input_lengths</span><span class="o">.</span><span class="n">cumsum</span><span class="p">(</span><span class="n">dim</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span><span class="o">.</span><span class="n">tolist</span><span class="p">()</span>
<span class="n">context_logits_output</span> <span class="o">=</span> <span class="p">[</span>
<span class="n">context_logits</span><span class="p">[</span><span class="n">s</span><span class="p">:</span><span class="n">e</span><span class="p">]</span>
<span class="k">for</span> <span class="n">s</span><span class="p">,</span> <span class="n">e</span> <span class="ow">in</span> <span class="nb">zip</span><span class="p">(</span><span class="n">seg_points</span><span class="p">[:</span><span class="o">-</span><span class="mi">1</span><span class="p">],</span> <span class="n">seg_points</span><span class="p">[</span><span class="mi">1</span><span class="p">:])</span>
<span class="p">]</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">context_logits_output</span> <span class="o">=</span> <span class="p">[</span>
<span class="n">context_logits</span><span class="p">[</span><span class="n">bidx</span><span class="p">,</span> <span class="p">:</span><span class="n">input_lengths</span><span class="p">[</span><span class="n">bidx</span><span class="p">]]</span>
<span class="k">for</span> <span class="n">bidx</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="n">batch_size</span><span class="p">)</span>
<span class="p">]</span>
<span class="k">assert</span> <span class="nb">len</span><span class="p">(</span><span class="n">context_logits_output</span><span class="p">)</span> <span class="o">==</span> <span class="n">batch_size</span>
<span class="n">outputs</span><span class="p">[</span><span class="s1">&#39;context_logits&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="n">context_logits_output</span>
<span class="k">if</span> <span class="s1">&#39;generation_logits&#39;</span> <span class="ow">in</span> <span class="n">outputs</span><span class="p">:</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">has_pp</span><span class="p">():</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">is_last_pp_rank</span><span class="p">():</span>
<span class="n">generation_logits</span> <span class="o">=</span> <span class="n">outputs</span><span class="p">[</span><span class="s1">&#39;generation_logits&#39;</span><span class="p">]</span>
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">generation_logits</span><span class="p">,</span> <span class="nb">list</span><span class="p">):</span>
<span class="n">generation_logits_host</span> <span class="o">=</span> <span class="p">[</span>
<span class="n">logits</span><span class="o">.</span><span class="n">cpu</span><span class="p">()</span> <span class="k">for</span> <span class="n">logits</span> <span class="ow">in</span> <span class="n">generation_logits</span>
<span class="p">]</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">generation_logits_host</span> <span class="o">=</span> <span class="n">generation_logits</span><span class="o">.</span><span class="n">cpu</span><span class="p">()</span>
<span class="n">mpi_comm</span><span class="p">()</span><span class="o">.</span><span class="n">send</span><span class="p">(</span><span class="n">generation_logits_host</span><span class="p">,</span> <span class="n">dest</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span>
<span class="k">elif</span> <span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">is_first_pp_rank</span><span class="p">():</span>
<span class="n">generation_logits_host</span> <span class="o">=</span> <span class="n">mpi_comm</span><span class="p">()</span><span class="o">.</span><span class="n">recv</span><span class="p">(</span>
<span class="n">source</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">prev_pp_rank</span><span class="p">()</span>
<span class="p">)</span> <span class="c1"># Prev pp rank of rank=0 is the last pp</span>
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">generation_logits_host</span><span class="p">,</span> <span class="nb">list</span><span class="p">):</span>
<span class="n">generation_logits</span> <span class="o">=</span> <span class="p">[</span>
<span class="n">logits</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda:0&#39;</span><span class="p">))</span>
<span class="k">for</span> <span class="n">logits</span> <span class="ow">in</span> <span class="n">generation_logits_host</span>
<span class="p">]</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">generation_logits</span> <span class="o">=</span> <span class="n">generation_logits_host</span><span class="o">.</span><span class="n">to</span><span class="p">(</span>
<span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">&#39;cuda:0&#39;</span><span class="p">))</span>
<span class="n">outputs</span><span class="p">[</span><span class="s1">&#39;generation_logits&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="n">generation_logits</span>
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="p">,</span> <span class="n">GenerationSession</span><span class="p">):</span>
<span class="c1"># Convert logits format to be same as GptSession</span>
<span class="n">generation_logits</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">stack</span><span class="p">(</span>
<span class="n">outputs</span><span class="p">[</span><span class="s1">&#39;generation_logits&#39;</span><span class="p">],</span> <span class="n">dim</span><span class="o">=</span><span class="mi">1</span><span class="p">)</span>
<span class="n">batch_x_beam</span><span class="p">,</span> <span class="n">max_gen_len</span><span class="p">,</span> <span class="n">voc_size</span> <span class="o">=</span> <span class="n">generation_logits</span><span class="o">.</span><span class="n">size</span><span class="p">(</span>
<span class="p">)</span>
<span class="n">num_beams</span> <span class="o">=</span> <span class="n">batch_x_beam</span> <span class="o">//</span> <span class="n">batch_size</span>
<span class="n">generation_logits</span> <span class="o">=</span> <span class="n">generation_logits</span><span class="o">.</span><span class="n">view</span><span class="p">(</span>
<span class="n">batch_size</span><span class="p">,</span> <span class="n">num_beams</span><span class="p">,</span> <span class="n">max_gen_len</span><span class="p">,</span> <span class="n">voc_size</span><span class="p">)</span>
<span class="n">outputs</span><span class="p">[</span><span class="s1">&#39;generation_logits&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="n">generation_logits</span>
<span class="k">return</span> <span class="n">outputs</span>
<span class="k">def</span> <span class="nf">_prepare_embedding_table</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">prompt_table</span><span class="p">:</span> <span class="n">Union</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">]):</span>
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">prompt_table</span><span class="p">,</span> <span class="nb">str</span><span class="p">):</span>
<span class="n">prompt_table_data</span> <span class="o">=</span> <span class="n">numpy_to_torch</span><span class="p">(</span>
<span class="n">np</span><span class="o">.</span><span class="n">load</span><span class="p">(</span><span class="n">prompt_table</span><span class="p">))</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="n">dtype</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">dtype</span><span class="p">)</span>
<span class="k">else</span><span class="p">:</span>
<span class="k">assert</span> <span class="nb">isinstance</span><span class="p">(</span>
<span class="n">prompt_table</span><span class="p">,</span>
<span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">),</span> <span class="s2">&quot;Prompt table should be str or torch.Tensor&quot;</span>
<span class="n">prompt_table_data</span> <span class="o">=</span> <span class="n">prompt_table</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="n">dtype</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">dtype</span><span class="p">)</span>
<span class="k">return</span> <span class="n">prompt_table_data</span>
<span class="k">def</span> <span class="nf">_prepare_ptuning</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">prompt_table</span><span class="p">:</span> <span class="n">Union</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">],</span>
<span class="n">tasks</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span> <span class="n">batch_size</span><span class="p">:</span> <span class="nb">int</span><span class="p">):</span>
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">max_prompt_embedding_table_size</span> <span class="o">==</span> <span class="mi">0</span><span class="p">:</span>
<span class="k">return</span> <span class="p">{}</span>
<span class="k">if</span> <span class="n">prompt_table</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">prompt_table_data</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_prepare_embedding_table</span><span class="p">(</span><span class="n">prompt_table</span><span class="p">)</span>
<span class="k">if</span> <span class="nb">len</span><span class="p">(</span><span class="n">prompt_table_data</span><span class="o">.</span><span class="n">size</span><span class="p">())</span> <span class="o">==</span> <span class="mi">3</span><span class="p">:</span>
<span class="n">_</span><span class="p">,</span> <span class="n">task_vocab_size</span><span class="p">,</span> <span class="n">hidden_size</span> <span class="o">=</span> <span class="n">prompt_table_data</span><span class="o">.</span><span class="n">size</span><span class="p">()</span>
<span class="k">elif</span> <span class="nb">len</span><span class="p">(</span><span class="n">prompt_table_data</span><span class="o">.</span><span class="n">size</span><span class="p">())</span> <span class="o">==</span> <span class="mi">2</span><span class="p">:</span>
<span class="n">task_vocab_size</span><span class="p">,</span> <span class="n">hidden_size</span> <span class="o">=</span> <span class="n">prompt_table_data</span><span class="o">.</span><span class="n">size</span><span class="p">()</span>
<span class="n">task_vocab_size</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">([</span><span class="n">task_vocab_size</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">int32</span><span class="p">)</span>
<span class="n">prompt_table_data</span> <span class="o">=</span> <span class="n">prompt_table_data</span><span class="o">.</span><span class="n">view</span><span class="p">(</span><span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="n">hidden_size</span><span class="p">)</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">prompt_table_data</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">empty</span><span class="p">(</span>
<span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="bp">self</span><span class="o">.</span><span class="n">hidden_size</span> <span class="o">*</span> <span class="bp">self</span><span class="o">.</span><span class="n">mapping</span><span class="o">.</span><span class="n">tp_size</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">dtype</span><span class="p">)</span>
<span class="n">task_vocab_size</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">zeros</span><span class="p">([</span><span class="mi">1</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">int32</span><span class="p">)</span>
<span class="k">if</span> <span class="n">tasks</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">tasks</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">([</span><span class="nb">int</span><span class="p">(</span><span class="n">t</span><span class="p">)</span> <span class="k">for</span> <span class="n">t</span> <span class="ow">in</span> <span class="n">tasks</span><span class="o">.</span><span class="n">split</span><span class="p">(</span><span class="s1">&#39;,&#39;</span><span class="p">)],</span>
<span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">int32</span><span class="p">)</span>
<span class="k">assert</span> <span class="n">tasks</span><span class="o">.</span><span class="n">size</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span> <span class="o">==</span> <span class="n">batch_size</span><span class="p">,</span> \
<span class="sa">f</span><span class="s2">&quot;Number of supplied tasks (</span><span class="si">{</span><span class="n">tasks</span><span class="o">.</span><span class="n">size</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span><span class="si">}</span><span class="s2">) must match input batch size (</span><span class="si">{</span><span class="n">batch_size</span><span class="si">}</span><span class="s2">)&quot;</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">tasks</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">zeros</span><span class="p">([</span><span class="n">batch_size</span><span class="p">],</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">int32</span><span class="p">)</span>
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="p">,</span> <span class="n">GenerationSession</span><span class="p">):</span>
<span class="k">return</span> <span class="p">{</span>
<span class="s1">&#39;prompt_embedding_table&#39;</span><span class="p">:</span> <span class="n">prompt_table_data</span><span class="o">.</span><span class="n">cuda</span><span class="p">(),</span>
<span class="s1">&#39;tasks&#39;</span><span class="p">:</span> <span class="n">tasks</span><span class="o">.</span><span class="n">cuda</span><span class="p">(),</span>
<span class="s1">&#39;prompt_vocab_size&#39;</span><span class="p">:</span> <span class="n">task_vocab_size</span><span class="o">.</span><span class="n">cuda</span><span class="p">()</span>
<span class="p">}</span>
<span class="k">else</span><span class="p">:</span>
<span class="k">return</span> <span class="p">{</span>
<span class="s1">&#39;embedding_table&#39;</span><span class="p">:</span> <span class="n">prompt_table_data</span><span class="o">.</span><span class="n">cuda</span><span class="p">(),</span>
<span class="s1">&#39;tasks&#39;</span><span class="p">:</span> <span class="n">tasks</span><span class="o">.</span><span class="n">cuda</span><span class="p">(),</span>
<span class="s1">&#39;vocab_size&#39;</span><span class="p">:</span> <span class="n">task_vocab_size</span><span class="o">.</span><span class="n">cuda</span><span class="p">()</span>
<span class="p">}</span>
<div class="viewcode-block" id="ModelRunner">
<a class="viewcode-back" href="../../../python-api/tensorrt_llm.runtime.html#tensorrt_llm.runtime.ModelRunner">[docs]</a>
<span class="k">class</span> <span class="nc">ModelRunner</span><span class="p">(</span><span class="n">ModelRunnerMixin</span><span class="p">):</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> An interface class that wraps GenerationSession and provides generation methods.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="k">def</span> <span class="fm">__init__</span><span class="p">(</span>
<span class="bp">self</span><span class="p">,</span>
<span class="n">session</span><span class="p">:</span> <span class="n">GenerationSession</span><span class="p">,</span>
<span class="n">max_batch_size</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
<span class="n">max_input_len</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
<span class="n">max_seq_len</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
<span class="n">max_beam_width</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
<span class="n">kv_cache_type</span><span class="p">:</span> <span class="n">KVCacheType</span><span class="p">,</span>
<span class="n">lora_manager</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">LoraManager</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</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"> Create a ModelRunner instance.</span>
<span class="sd"> You are recommended to use the from_dir method to load the engine and create a ModelRunner instance.</span>
<span class="sd"> Args:</span>
<span class="sd"> session (GenerationSession):</span>
<span class="sd"> The TensorRT session created from an engine.</span>
<span class="sd"> max_batch_size (int):</span>
<span class="sd"> The maximum batch size allowed for the input.</span>
<span class="sd"> max_input_len (int):</span>
<span class="sd"> The maximum input length allowed for the input.</span>
<span class="sd"> max_seq_len (int):</span>
<span class="sd"> The maximum sequence length (input + new tokens).</span>
<span class="sd"> max_beam_width (int):</span>
<span class="sd"> The maximum beam width.</span>
<span class="sd"> lora_manager (LoraManager):</span>
<span class="sd"> The LoRA manager to handle LoRA weights.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="bp">self</span><span class="o">.</span><span class="n">session</span> <span class="o">=</span> <span class="n">session</span>
<span class="bp">self</span><span class="o">.</span><span class="n">max_batch_size</span> <span class="o">=</span> <span class="n">max_batch_size</span>
<span class="bp">self</span><span class="o">.</span><span class="n">max_input_len</span> <span class="o">=</span> <span class="n">max_input_len</span>
<span class="bp">self</span><span class="o">.</span><span class="n">max_seq_len</span> <span class="o">=</span> <span class="n">max_seq_len</span>
<span class="bp">self</span><span class="o">.</span><span class="n">max_beam_width</span> <span class="o">=</span> <span class="n">max_beam_width</span>
<span class="bp">self</span><span class="o">.</span><span class="n">lora_manager</span> <span class="o">=</span> <span class="n">lora_manager</span>
<span class="bp">self</span><span class="o">.</span><span class="n">kv_cache_type</span> <span class="o">=</span> <span class="n">kv_cache_type</span>
<span class="bp">self</span><span class="o">.</span><span class="n">enable_context_fmha_fp32_acc</span> <span class="o">=</span> <span class="kc">False</span>
<span class="bp">self</span><span class="o">.</span><span class="n">multi_block_mode</span> <span class="o">=</span> <span class="kc">True</span>
<div class="viewcode-block" id="ModelRunner.from_engine">
<a class="viewcode-back" href="../../../python-api/tensorrt_llm.runtime.html#tensorrt_llm.runtime.ModelRunner.from_engine">[docs]</a>
<span class="nd">@classmethod</span>
<span class="k">def</span> <span class="nf">from_engine</span><span class="p">(</span>
<span class="bp">cls</span><span class="p">,</span>
<span class="n">engine</span><span class="p">:</span> <span class="n">Engine</span><span class="p">,</span>
<span class="o">*</span><span class="p">,</span>
<span class="n">max_output_len</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">int</span><span class="p">],</span>
<span class="n">lora_dir</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">List</span><span class="p">[</span><span class="nb">str</span><span class="p">]],</span>
<span class="n">rank</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
<span class="n">debug_mode</span><span class="p">:</span> <span class="nb">bool</span><span class="p">,</span>
<span class="n">lora_ckpt_source</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span>
<span class="n">medusa_choices</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">List</span><span class="p">[</span><span class="nb">int</span><span class="p">]],</span>
<span class="n">stream</span><span class="p">:</span> <span class="n">torch</span><span class="o">.</span><span class="n">cuda</span><span class="o">.</span><span class="n">Stream</span><span class="p">,</span>
<span class="n">gpu_weights_percent</span><span class="p">:</span> <span class="nb">float</span><span class="p">,</span>
<span class="n">enable_context_fmha_fp32_acc</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">bool</span><span class="p">],</span>
<span class="n">multi_block_mode</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">bool</span><span class="p">],</span>
<span class="p">)</span> <span class="o">-&gt;</span> <span class="s1">&#39;ModelRunner&#39;</span><span class="p">:</span>
<span class="n">model_config</span> <span class="o">=</span> <span class="n">_engine_config_to_model_config</span><span class="p">(</span>
<span class="n">engine</span><span class="o">.</span><span class="n">config</span><span class="p">,</span> <span class="n">gpu_weights_percent</span><span class="o">=</span><span class="n">gpu_weights_percent</span><span class="p">)</span>
<span class="k">if</span> <span class="n">model_config</span><span class="o">.</span><span class="n">kv_cache_type</span> <span class="o">==</span> <span class="n">KVCacheType</span><span class="o">.</span><span class="n">DISABLED</span><span class="p">:</span>
<span class="k">assert</span> <span class="n">max_output_len</span> <span class="o">==</span> <span class="mi">1</span> <span class="ow">or</span> <span class="n">max_output_len</span> <span class="ow">is</span> <span class="kc">None</span><span class="p">,</span> <span class="s1">&#39;Disabled KV cache is intended for context phase only now.&#39;</span>
<span class="n">pretrained_config</span> <span class="o">=</span> <span class="n">engine</span><span class="o">.</span><span class="n">config</span><span class="o">.</span><span class="n">pretrained_config</span>
<span class="n">build_config</span> <span class="o">=</span> <span class="n">engine</span><span class="o">.</span><span class="n">config</span><span class="o">.</span><span class="n">build_config</span>
<span class="n">max_batch_size</span> <span class="o">=</span> <span class="n">build_config</span><span class="o">.</span><span class="n">max_batch_size</span>
<span class="n">max_input_len</span> <span class="o">=</span> <span class="n">build_config</span><span class="o">.</span><span class="n">max_input_len</span>
<span class="n">max_seq_len</span> <span class="o">=</span> <span class="n">build_config</span><span class="o">.</span><span class="n">max_seq_len</span>
<span class="n">max_beam_width</span> <span class="o">=</span> <span class="n">build_config</span><span class="o">.</span><span class="n">max_beam_width</span>
<span class="k">if</span> <span class="s1">&#39;GLM&#39;</span> <span class="ow">in</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">architecture</span> <span class="ow">and</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">chatglm_version</span> <span class="ow">in</span> <span class="p">[</span>
<span class="s1">&#39;glm&#39;</span><span class="p">,</span> <span class="s1">&#39;chatglm&#39;</span>
<span class="p">]:</span>
<span class="n">session_cls</span> <span class="o">=</span> <span class="n">ChatGLMGenerationSession</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">session_cls</span> <span class="o">=</span> <span class="n">GenerationSession</span>
<span class="n">engine_buffer</span> <span class="o">=</span> <span class="n">engine</span><span class="o">.</span><span class="n">engine</span>
<span class="n">runtime_mapping</span> <span class="o">=</span> <span class="n">pretrained_config</span><span class="o">.</span><span class="n">mapping</span>
<span class="k">if</span> <span class="n">medusa_choices</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="k">assert</span> <span class="n">session_cls</span> <span class="o">==</span> <span class="n">GenerationSession</span><span class="p">,</span> <span class="s2">&quot;Medusa is only supported by GenerationSession&quot;</span>
<span class="k">assert</span> <span class="n">model_config</span><span class="o">.</span><span class="n">max_medusa_tokens</span> <span class="o">&gt;</span> <span class="mi">0</span><span class="p">,</span> \
<span class="s2">&quot;medusa_chioce is specified but model_config.max_medusa_tokens is 0.&quot;</span>
<span class="k">if</span> <span class="n">MpiComm</span><span class="o">.</span><span class="n">size</span><span class="p">()</span> <span class="o">&gt;</span> <span class="n">runtime_mapping</span><span class="o">.</span><span class="n">gpus_per_node</span><span class="p">:</span>
<span class="k">assert</span> <span class="n">MpiComm</span><span class="o">.</span><span class="n">local_size</span><span class="p">()</span> <span class="o">==</span> <span class="n">runtime_mapping</span><span class="o">.</span><span class="n">gpus_per_node</span>
<span class="k">if</span> <span class="ow">not</span> <span class="n">DISABLE_TORCH_DEVICE_SET</span><span class="p">:</span>
<span class="n">torch</span><span class="o">.</span><span class="n">cuda</span><span class="o">.</span><span class="n">set_device</span><span class="p">(</span><span class="n">rank</span> <span class="o">%</span> <span class="n">runtime_mapping</span><span class="o">.</span><span class="n">gpus_per_node</span><span class="p">)</span>
<span class="n">session</span> <span class="o">=</span> <span class="n">session_cls</span><span class="p">(</span><span class="n">model_config</span><span class="p">,</span>
<span class="n">engine_buffer</span><span class="p">,</span>
<span class="n">runtime_mapping</span><span class="p">,</span>
<span class="n">debug_mode</span><span class="o">=</span><span class="n">debug_mode</span><span class="p">,</span>
<span class="n">stream</span><span class="o">=</span><span class="n">stream</span><span class="p">)</span>
<span class="k">if</span> <span class="n">session</span><span class="o">.</span><span class="n">runtime</span><span class="o">.</span><span class="n">engine</span><span class="o">.</span><span class="n">streamable_weights_size</span><span class="p">:</span>
<span class="n">session</span><span class="o">.</span><span class="n">runtime</span><span class="o">.</span><span class="n">_set_weight_streaming</span><span class="p">(</span><span class="n">gpu_weights_percent</span><span class="p">)</span>
<span class="k">if</span> <span class="n">session</span><span class="o">.</span><span class="n">use_lora_plugin</span><span class="p">:</span>
<span class="n">lora_manager</span> <span class="o">=</span> <span class="n">LoraManager</span><span class="p">()</span>
<span class="k">if</span> <span class="n">lora_dir</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">lora_manager</span><span class="o">.</span><span class="n">load_from_ckpt</span><span class="p">(</span><span class="n">lora_dir</span><span class="p">,</span>
<span class="n">model_config</span><span class="o">=</span><span class="n">model_config</span><span class="p">,</span>
<span class="n">runtime_mapping</span><span class="o">=</span><span class="n">runtime_mapping</span><span class="p">,</span>
<span class="n">ckpt_source</span><span class="o">=</span><span class="n">lora_ckpt_source</span><span class="p">)</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">lora_manager</span> <span class="o">=</span> <span class="kc">None</span>
<span class="n">runner</span> <span class="o">=</span> <span class="bp">cls</span><span class="p">(</span><span class="n">session</span><span class="o">=</span><span class="n">session</span><span class="p">,</span>
<span class="n">max_batch_size</span><span class="o">=</span><span class="n">max_batch_size</span><span class="p">,</span>
<span class="n">max_input_len</span><span class="o">=</span><span class="n">max_input_len</span><span class="p">,</span>
<span class="n">max_seq_len</span><span class="o">=</span><span class="n">max_seq_len</span><span class="p">,</span>
<span class="n">max_beam_width</span><span class="o">=</span><span class="n">max_beam_width</span><span class="p">,</span>
<span class="n">kv_cache_type</span><span class="o">=</span><span class="n">model_config</span><span class="o">.</span><span class="n">kv_cache_type</span><span class="p">,</span>
<span class="n">lora_manager</span><span class="o">=</span><span class="n">lora_manager</span><span class="p">)</span>
<span class="n">runner</span><span class="o">.</span><span class="n">enable_context_fmha_fp32_acc</span> <span class="o">=</span> <span class="n">enable_context_fmha_fp32_acc</span>
<span class="n">runner</span><span class="o">.</span><span class="n">multi_block_mode</span> <span class="o">=</span> <span class="n">multi_block_mode</span>
<span class="k">return</span> <span class="n">runner</span></div>
<div class="viewcode-block" id="ModelRunner.from_dir">
<a class="viewcode-back" href="../../../python-api/tensorrt_llm.runtime.html#tensorrt_llm.runtime.ModelRunner.from_dir">[docs]</a>
<span class="nd">@classmethod</span>
<span class="k">def</span> <span class="nf">from_dir</span><span class="p">(</span>
<span class="bp">cls</span><span class="p">,</span>
<span class="n">engine_dir</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span>
<span class="o">*</span><span class="p">,</span>
<span class="n">max_output_len</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">int</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">lora_dir</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">List</span><span class="p">[</span><span class="nb">str</span><span class="p">]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">rank</span><span class="p">:</span> <span class="nb">int</span> <span class="o">=</span> <span class="mi">0</span><span class="p">,</span>
<span class="n">debug_mode</span><span class="p">:</span> <span class="nb">bool</span> <span class="o">=</span> <span class="kc">False</span><span class="p">,</span>
<span class="n">lora_ckpt_source</span><span class="p">:</span> <span class="nb">str</span> <span class="o">=</span> <span class="s2">&quot;hf&quot;</span><span class="p">,</span>
<span class="n">medusa_choices</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">List</span><span class="p">[</span><span class="nb">int</span><span class="p">]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">stream</span><span class="p">:</span> <span class="n">torch</span><span class="o">.</span><span class="n">cuda</span><span class="o">.</span><span class="n">Stream</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">gpu_weights_percent</span><span class="p">:</span> <span class="nb">float</span> <span class="o">=</span> <span class="mi">1</span><span class="p">,</span>
<span class="n">enable_context_fmha_fp32_acc</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">bool</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">multi_block_mode</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">bool</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="p">)</span> <span class="o">-&gt;</span> <span class="s1">&#39;ModelRunner&#39;</span><span class="p">:</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> Create a ModelRunner instance from an engine directory.</span>
<span class="sd"> Args:</span>
<span class="sd"> engine_dir (str):</span>
<span class="sd"> The directory that contains the serialized engine files and config files.</span>
<span class="sd"> max_output_len (Optional[int]):</span>
<span class="sd"> max_output_len, this arg might be available only when loading time, generate will still to check when disable_kv_cache is enabled.</span>
<span class="sd"> lora_dir (Optional[List[str]]):</span>
<span class="sd"> The directories that contain LoRA weights.</span>
<span class="sd"> rank (int):</span>
<span class="sd"> The runtime rank id.</span>
<span class="sd"> debug_mode (bool):</span>
<span class="sd"> Whether or not to turn on the debug mode.</span>
<span class="sd"> medusa_choices (List[List[int]]):</span>
<span class="sd"> Medusa choices to use when in Medusa decoding</span>
<span class="sd"> stream (torch.cuda.Stream):</span>
<span class="sd"> Stream to use.</span>
<span class="sd"> multi_block_mode (bool):</span>
<span class="sd"> Whether to distribute the work across multiple CUDA thread-blocks on the GPU for masked MHA kernel.</span>
<span class="sd"> Returns:</span>
<span class="sd"> ModelRunner: An instance of ModelRunner.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="n">engine_version</span> <span class="o">=</span> <span class="n">get_engine_version</span><span class="p">(</span><span class="n">engine_dir</span><span class="p">)</span>
<span class="n">profiler</span><span class="o">.</span><span class="n">start</span><span class="p">(</span><span class="s1">&#39;load tensorrt_llm engine&#39;</span><span class="p">)</span>
<span class="c1"># the old engine format</span>
<span class="k">if</span> <span class="n">engine_version</span> <span class="ow">is</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">engine_dir</span> <span class="o">=</span> <span class="n">Path</span><span class="p">(</span><span class="n">engine_dir</span><span class="p">)</span>
<span class="n">config_path</span> <span class="o">=</span> <span class="n">engine_dir</span> <span class="o">/</span> <span class="s2">&quot;config.json&quot;</span>
<span class="n">model_config</span><span class="p">,</span> <span class="n">other_config</span> <span class="o">=</span> <span class="n">read_config</span><span class="p">(</span><span class="n">config_path</span><span class="p">)</span>
<span class="n">world_size</span> <span class="o">=</span> <span class="n">other_config</span><span class="o">.</span><span class="n">pop</span><span class="p">(</span><span class="s1">&#39;world_size&#39;</span><span class="p">)</span>
<span class="n">tp_size</span> <span class="o">=</span> <span class="n">other_config</span><span class="o">.</span><span class="n">pop</span><span class="p">(</span><span class="s1">&#39;tp_size&#39;</span><span class="p">)</span>
<span class="n">pp_size</span> <span class="o">=</span> <span class="n">other_config</span><span class="o">.</span><span class="n">pop</span><span class="p">(</span><span class="s1">&#39;pp_size&#39;</span><span class="p">)</span>
<span class="n">max_batch_size</span> <span class="o">=</span> <span class="n">other_config</span><span class="o">.</span><span class="n">pop</span><span class="p">(</span><span class="s1">&#39;max_batch_size&#39;</span><span class="p">)</span>
<span class="n">max_input_len</span> <span class="o">=</span> <span class="n">other_config</span><span class="o">.</span><span class="n">pop</span><span class="p">(</span><span class="s1">&#39;max_input_len&#39;</span><span class="p">)</span>
<span class="n">max_output_len</span> <span class="o">=</span> <span class="n">other_config</span><span class="o">.</span><span class="n">pop</span><span class="p">(</span><span class="s1">&#39;max_output_len&#39;</span><span class="p">)</span>
<span class="n">max_beam_width</span> <span class="o">=</span> <span class="n">other_config</span><span class="o">.</span><span class="n">pop</span><span class="p">(</span><span class="s1">&#39;max_beam_width&#39;</span><span class="p">)</span>
<span class="n">runtime_mapping</span> <span class="o">=</span> <span class="n">Mapping</span><span class="p">(</span><span class="n">world_size</span><span class="o">=</span><span class="n">world_size</span><span class="p">,</span>
<span class="n">rank</span><span class="o">=</span><span class="n">rank</span><span class="p">,</span>
<span class="n">tp_size</span><span class="o">=</span><span class="n">tp_size</span><span class="p">,</span>
<span class="n">pp_size</span><span class="o">=</span><span class="n">pp_size</span><span class="p">)</span>
<span class="n">engine_name</span> <span class="o">=</span> <span class="n">get_engine_name</span><span class="p">(</span><span class="n">model_config</span><span class="o">.</span><span class="n">model_name</span><span class="p">,</span>
<span class="n">model_config</span><span class="o">.</span><span class="n">dtype</span><span class="p">,</span> <span class="n">tp_size</span><span class="p">,</span> <span class="n">pp_size</span><span class="p">,</span>
<span class="n">rank</span><span class="p">)</span>
<span class="n">serialize_path</span> <span class="o">=</span> <span class="n">engine_dir</span> <span class="o">/</span> <span class="n">engine_name</span>
<span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="n">serialize_path</span><span class="p">,</span> <span class="s1">&#39;rb&#39;</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
<span class="n">engine_buffer</span> <span class="o">=</span> <span class="n">f</span><span class="o">.</span><span class="n">read</span><span class="p">()</span>
<span class="k">if</span> <span class="n">model_config</span><span class="o">.</span><span class="n">model_name</span> <span class="ow">in</span> <span class="p">(</span><span class="s1">&#39;chatglm_6b&#39;</span><span class="p">,</span> <span class="s1">&#39;glm_10b&#39;</span><span class="p">):</span>
<span class="n">session_cls</span> <span class="o">=</span> <span class="n">ChatGLMGenerationSession</span>
<span class="k">elif</span> <span class="n">model_config</span><span class="o">.</span><span class="n">model_name</span> <span class="o">==</span> <span class="s1">&#39;qwen&#39;</span><span class="p">:</span>
<span class="n">session_cls</span> <span class="o">=</span> <span class="n">QWenForCausalLMGenerationSession</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">session_cls</span> <span class="o">=</span> <span class="n">GenerationSession</span>
<span class="k">if</span> <span class="n">medusa_choices</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="k">assert</span> <span class="n">model_config</span><span class="o">.</span><span class="n">max_medusa_tokens</span> <span class="o">&gt;</span> <span class="mi">0</span><span class="p">,</span> \
<span class="s2">&quot;medusa_choice is specified but model_config.max_medusa_tokens is 0.&quot;</span>
<span class="k">if</span> <span class="ow">not</span> <span class="n">DISABLE_TORCH_DEVICE_SET</span><span class="p">:</span>
<span class="n">torch</span><span class="o">.</span><span class="n">cuda</span><span class="o">.</span><span class="n">set_device</span><span class="p">(</span><span class="n">rank</span> <span class="o">%</span> <span class="n">runtime_mapping</span><span class="o">.</span><span class="n">gpus_per_node</span><span class="p">)</span>
<span class="n">session</span> <span class="o">=</span> <span class="n">session_cls</span><span class="p">(</span><span class="n">model_config</span><span class="p">,</span>
<span class="n">engine_buffer</span><span class="p">,</span>
<span class="n">runtime_mapping</span><span class="p">,</span>
<span class="n">debug_mode</span><span class="o">=</span><span class="n">debug_mode</span><span class="p">,</span>
<span class="n">stream</span><span class="o">=</span><span class="n">stream</span><span class="p">)</span>
<span class="k">if</span> <span class="n">session</span><span class="o">.</span><span class="n">use_lora_plugin</span><span class="p">:</span>
<span class="n">lora_manager</span> <span class="o">=</span> <span class="n">LoraManager</span><span class="p">()</span>
<span class="k">if</span> <span class="n">lora_dir</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">lora_manager</span><span class="o">.</span><span class="n">load_from_ckpt</span><span class="p">(</span><span class="n">lora_dir</span><span class="p">,</span>
<span class="n">model_config</span><span class="o">=</span><span class="n">model_config</span><span class="p">,</span>
<span class="n">runtime_mapping</span><span class="o">=</span><span class="n">runtime_mapping</span><span class="p">,</span>
<span class="n">ckpt_source</span><span class="o">=</span><span class="n">lora_ckpt_source</span><span class="p">)</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">lora_manager</span> <span class="o">=</span> <span class="kc">None</span>
<span class="k">if</span> <span class="n">session</span><span class="o">.</span><span class="n">runtime</span><span class="o">.</span><span class="n">engine</span><span class="o">.</span><span class="n">streamable_weights_size</span><span class="p">:</span>
<span class="n">session</span><span class="o">.</span><span class="n">runtime</span><span class="o">.</span><span class="n">_set_weight_streaming</span><span class="p">(</span><span class="n">gpu_weights_percent</span><span class="p">)</span>
<span class="n">profiler</span><span class="o">.</span><span class="n">stop</span><span class="p">(</span><span class="s1">&#39;load tensorrt_llm engine&#39;</span><span class="p">)</span>
<span class="n">loading_time</span> <span class="o">=</span> <span class="n">profiler</span><span class="o">.</span><span class="n">elapsed_time_in_sec</span><span class="p">(</span>
<span class="s2">&quot;load tensorrt_llm engine&quot;</span><span class="p">)</span>
<span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="sa">f</span><span class="s1">&#39;Load engine takes: </span><span class="si">{</span><span class="n">loading_time</span><span class="si">}</span><span class="s1"> sec&#39;</span><span class="p">)</span>
<span class="n">runner</span> <span class="o">=</span> <span class="bp">cls</span><span class="p">(</span><span class="n">session</span><span class="o">=</span><span class="n">session</span><span class="p">,</span>
<span class="n">max_batch_size</span><span class="o">=</span><span class="n">max_batch_size</span><span class="p">,</span>
<span class="n">max_input_len</span><span class="o">=</span><span class="n">max_input_len</span><span class="p">,</span>
<span class="n">max_seq_len</span><span class="o">=</span><span class="n">max_input_len</span> <span class="o">+</span> <span class="n">max_output_len</span><span class="p">,</span>
<span class="n">max_beam_width</span><span class="o">=</span><span class="n">max_beam_width</span><span class="p">,</span>
<span class="n">kv_cache_type</span><span class="o">=</span><span class="n">KVCacheType</span><span class="o">.</span><span class="n">CONTINUOUS</span><span class="p">,</span>
<span class="n">lora_manager</span><span class="o">=</span><span class="n">lora_manager</span><span class="p">)</span>
<span class="n">runner</span><span class="o">.</span><span class="n">enable_context_fmha_fp32_acc</span> <span class="o">=</span> <span class="n">enable_context_fmha_fp32_acc</span>
<span class="n">runner</span><span class="o">.</span><span class="n">multi_block_mode</span> <span class="o">=</span> <span class="n">multi_block_mode</span>
<span class="k">return</span> <span class="n">runner</span>
<span class="k">else</span><span class="p">:</span>
<span class="c1"># the new engine format</span>
<span class="n">engine</span> <span class="o">=</span> <span class="n">Engine</span><span class="o">.</span><span class="n">from_dir</span><span class="p">(</span><span class="n">engine_dir</span><span class="p">,</span> <span class="n">rank</span><span class="p">)</span>
<span class="k">if</span> <span class="n">lora_dir</span> <span class="ow">is</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">config_lora_dir</span> <span class="o">=</span> <span class="n">engine</span><span class="o">.</span><span class="n">config</span><span class="o">.</span><span class="n">build_config</span><span class="o">.</span><span class="n">lora_config</span><span class="o">.</span><span class="n">lora_dir</span>
<span class="k">if</span> <span class="nb">len</span><span class="p">(</span><span class="n">config_lora_dir</span><span class="p">)</span> <span class="o">&gt;</span> <span class="mi">0</span><span class="p">:</span>
<span class="n">lora_dir</span> <span class="o">=</span> <span class="p">[</span>
<span class="sa">f</span><span class="s2">&quot;</span><span class="si">{</span><span class="n">engine_dir</span><span class="si">}</span><span class="s2">/</span><span class="si">{</span><span class="nb">dir</span><span class="si">}</span><span class="s2">&quot;</span> <span class="k">for</span> <span class="nb">dir</span> <span class="ow">in</span> <span class="n">config_lora_dir</span>
<span class="p">]</span>
<span class="n">lora_ckpt_source</span> <span class="o">=</span> <span class="n">engine</span><span class="o">.</span><span class="n">config</span><span class="o">.</span><span class="n">build_config</span><span class="o">.</span><span class="n">lora_config</span><span class="o">.</span><span class="n">lora_ckpt_source</span>
<span class="n">runner</span> <span class="o">=</span> <span class="n">ModelRunner</span><span class="o">.</span><span class="n">from_engine</span><span class="p">(</span>
<span class="n">engine</span><span class="o">=</span><span class="n">engine</span><span class="p">,</span>
<span class="n">max_output_len</span><span class="o">=</span><span class="n">max_output_len</span><span class="p">,</span>
<span class="n">lora_dir</span><span class="o">=</span><span class="n">lora_dir</span><span class="p">,</span>
<span class="n">rank</span><span class="o">=</span><span class="n">rank</span><span class="p">,</span>
<span class="n">debug_mode</span><span class="o">=</span><span class="n">debug_mode</span><span class="p">,</span>
<span class="n">lora_ckpt_source</span><span class="o">=</span><span class="n">lora_ckpt_source</span><span class="p">,</span>
<span class="n">medusa_choices</span><span class="o">=</span><span class="n">medusa_choices</span><span class="p">,</span>
<span class="n">stream</span><span class="o">=</span><span class="n">stream</span><span class="p">,</span>
<span class="n">gpu_weights_percent</span><span class="o">=</span><span class="n">gpu_weights_percent</span><span class="p">,</span>
<span class="n">enable_context_fmha_fp32_acc</span><span class="o">=</span><span class="n">enable_context_fmha_fp32_acc</span><span class="p">,</span>
<span class="n">multi_block_mode</span><span class="o">=</span><span class="n">multi_block_mode</span><span class="p">,</span>
<span class="p">)</span>
<span class="n">profiler</span><span class="o">.</span><span class="n">stop</span><span class="p">(</span><span class="s1">&#39;load tensorrt_llm engine&#39;</span><span class="p">)</span>
<span class="n">loading_time</span> <span class="o">=</span> <span class="n">profiler</span><span class="o">.</span><span class="n">elapsed_time_in_sec</span><span class="p">(</span>
<span class="s2">&quot;load tensorrt_llm engine&quot;</span><span class="p">)</span>
<span class="n">logger</span><span class="o">.</span><span class="n">info</span><span class="p">(</span><span class="sa">f</span><span class="s1">&#39;Load engine takes: </span><span class="si">{</span><span class="n">loading_time</span><span class="si">}</span><span class="s1"> sec&#39;</span><span class="p">)</span>
<span class="k">return</span> <span class="n">runner</span></div>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">dtype</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">torch</span><span class="o">.</span><span class="n">dtype</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">dtype</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">vocab_size</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">int</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">vocab_size</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">vocab_size_padded</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">int</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">vocab_size_padded</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">hidden_size</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">int</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">hidden_size</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">num_heads</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">int</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">num_heads</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">num_layers</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">int</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">num_layers</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">max_sequence_length</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">int</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">max_seq_len</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">remove_input_padding</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">bool</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">remove_input_padding</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">use_lora_plugin</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">bool</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">use_lora_plugin</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">max_prompt_embedding_table_size</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">int</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">max_prompt_embedding_table_size</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">mapping</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">Mapping</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">mapping</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">gather_context_logits</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">bool</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">gather_context_logits</span>
<span class="nd">@property</span>
<span class="k">def</span> <span class="nf">gather_generation_logits</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">bool</span><span class="p">:</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">gather_generation_logits</span>
<div class="viewcode-block" id="ModelRunner.generate">
<a class="viewcode-back" href="../../../python-api/tensorrt_llm.runtime.html#tensorrt_llm.runtime.ModelRunner.generate">[docs]</a>
<span class="k">def</span> <span class="nf">generate</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span>
<span class="n">batch_input_ids</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">],</span>
<span class="n">position_ids</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">sampling_config</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">SamplingConfig</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">prompt_table</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">Union</span><span class="p">[</span><span class="nb">str</span><span class="p">,</span> <span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">prompt_tasks</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">str</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">lora_uids</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="nb">list</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">streaming</span><span class="p">:</span> <span class="nb">bool</span> <span class="o">=</span> <span class="kc">False</span><span class="p">,</span>
<span class="n">stopping_criteria</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">StoppingCriteria</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">logits_processor</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">LogitsProcessor</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">medusa_choices</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">List</span><span class="p">[</span><span class="n">List</span><span class="p">[</span><span class="nb">int</span><span class="p">]]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">encoder_max_input_length</span><span class="p">:</span> <span class="nb">int</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">encoder_input_features</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">encoder_output_lengths</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
<span class="n">cross_attention_masks</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</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">Union</span><span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">Tensor</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"> Generates sequences of token ids.</span>
<span class="sd"> The generation-controlling parameters are set in the sampling_config; it will be set to a default one if not passed.</span>
<span class="sd"> You can override any sampling_config&#39;s attributes by passing corresponding parameters.</span>
<span class="sd"> Args:</span>
<span class="sd"> batch_input_ids (List[torch.Tensor]):</span>
<span class="sd"> A list of input id tensors. Each tensor is of shape (sequence_length, ).</span>
<span class="sd"> sampling_config (SamplingConfig):</span>
<span class="sd"> The sampling configuration to be used as base parametrization for the generation call.</span>
<span class="sd"> The passed **kwargs matching the sampling_config&#39;s attributes will override them.</span>
<span class="sd"> If the sampling_config is not provided, a default will be used.</span>
<span class="sd"> prompt_table (str or torch.Tensor):</span>
<span class="sd"> The file path of prompt table (.npy format, exported by nemo_prompt_convert.py) or the prompt table itself.</span>
<span class="sd"> prompt_tasks (str):</span>
<span class="sd"> The prompt tuning task ids for the input batch, in format of comma-separated list (e.g., 0,3,1,0).</span>
<span class="sd"> lora_uids (list):</span>
<span class="sd"> The uids of LoRA weights for the input batch. Use -1 to disable the LoRA module.</span>
<span class="sd"> streaming (bool):</span>
<span class="sd"> Whether or not to use streaming mode for generation.</span>
<span class="sd"> stopping_criteria (StoppingCriteria):</span>
<span class="sd"> Custom stopping criteria.</span>
<span class="sd"> logits_processor (LogitsProcessor):</span>
<span class="sd"> Custom logits processors.</span>
<span class="sd"> medusa_choices (List[List[int]]):</span>
<span class="sd"> Medusa decoding choices.</span>
<span class="sd"> kwargs (Dict[str, Any]:</span>
<span class="sd"> Ad hoc parametrization of sampling_config.</span>
<span class="sd"> The passed **kwargs matching the sampling_config&#39;s attributes will override them.</span>
<span class="sd"> Returns:</span>
<span class="sd"> torch.Tensor or dict:</span>
<span class="sd"> If return_dict=False, the method returns generated output_ids.</span>
<span class="sd"> If return_dict=True, the method returns a dict of output_ids,</span>
<span class="sd"> sequence_lengths (if sampling_config.output_sequence_lengths=True),</span>
<span class="sd"> context_logits and generation_logits (if self.gather_context_logits=True</span>
<span class="sd"> and self.gather_generation_logits=True, respectively).</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="c1"># Use sampling_config like HF&#39;s generation_config</span>
<span class="k">if</span> <span class="n">sampling_config</span> <span class="ow">is</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">sampling_config</span> <span class="o">=</span> <span class="n">SamplingConfig</span><span class="p">(</span><span class="n">end_id</span><span class="o">=</span><span class="kc">None</span><span class="p">,</span> <span class="n">pad_id</span><span class="o">=</span><span class="kc">None</span><span class="p">)</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">sampling_config</span> <span class="o">=</span> <span class="n">copy</span><span class="o">.</span><span class="n">deepcopy</span><span class="p">(</span><span class="n">sampling_config</span><span class="p">)</span>
<span class="n">sampling_config</span><span class="o">.</span><span class="n">update</span><span class="p">(</span><span class="o">**</span><span class="n">kwargs</span><span class="p">)</span>
<span class="c1"># To prevent numerical overflow when the temperature is set to 0.0</span>
<span class="c1"># Modify generation.SamplingConfig</span>
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">temperature</span><span class="p">,</span>
<span class="nb">float</span><span class="p">)</span> <span class="ow">and</span> <span class="n">sampling_config</span><span class="o">.</span><span class="n">temperature</span> <span class="o">==</span> <span class="mf">0.0</span><span class="p">:</span>
<span class="n">logger</span><span class="o">.</span><span class="n">warning</span><span class="p">(</span>
<span class="s2">&quot;Convert `temperature=0.0` to `temperature=1.0` and `top_k=1` to prevent overflow.&quot;</span>
<span class="p">)</span>
<span class="n">sampling_config</span><span class="o">.</span><span class="n">temperature</span> <span class="o">=</span> <span class="mf">1.0</span>
<span class="n">sampling_config</span><span class="o">.</span><span class="n">top_k</span> <span class="o">=</span> <span class="mi">1</span>
<span class="bp">self</span><span class="o">.</span><span class="n">_check_inputs</span><span class="p">(</span><span class="n">batch_input_ids</span><span class="p">,</span> <span class="n">sampling_config</span><span class="p">)</span>
<span class="k">if</span> <span class="n">kwargs</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s1">&#39;num_return_sequences&#39;</span><span class="p">,</span> <span class="kc">None</span><span class="p">)</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span>
<span class="s1">&#39;num_return_sequences will be ignored since &#39;</span>
<span class="s1">&#39;num_return_sequences &gt; 1 is not supported on python runtime. &#39;</span>
<span class="s1">&#39;Please use C++ runtime.&#39;</span><span class="p">)</span>
<span class="n">batch_size</span> <span class="o">=</span> <span class="nb">len</span><span class="p">(</span><span class="n">batch_input_ids</span><span class="p">)</span>
<span class="n">batch_input_ids</span><span class="p">,</span> <span class="n">input_lengths</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_prepare_inputs</span><span class="p">(</span>
<span class="n">batch_input_ids</span><span class="p">,</span> <span class="n">sampling_config</span><span class="o">.</span><span class="n">pad_id</span><span class="p">)</span>
<span class="k">def</span> <span class="nf">maybe_convert_to_words_list_format</span><span class="p">(</span>
<span class="n">words_list</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">Union</span><span class="p">[</span><span class="nb">list</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">,</span> <span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">]]</span>
<span class="p">)</span> <span class="o">-&gt;</span> <span class="n">Optional</span><span class="p">[</span><span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">]:</span>
<span class="k">if</span> <span class="n">words_list</span> <span class="ow">is</span> <span class="kc">None</span> <span class="ow">or</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">words_list</span><span class="p">,</span> <span class="n">np</span><span class="o">.</span><span class="n">ndarray</span><span class="p">):</span>
<span class="k">return</span> <span class="n">words_list</span>
<span class="k">elif</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">words_list</span><span class="p">,</span> <span class="n">torch</span><span class="o">.</span><span class="n">Tensor</span><span class="p">):</span>
<span class="k">return</span> <span class="n">words_list</span><span class="o">.</span><span class="n">numpy</span><span class="p">()</span>
<span class="k">elif</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">words_list</span><span class="p">,</span> <span class="nb">list</span><span class="p">):</span>
<span class="k">return</span> <span class="n">to_word_list_format</span><span class="p">(</span><span class="n">words_list</span><span class="p">)</span>
<span class="k">else</span><span class="p">:</span>
<span class="k">raise</span> <span class="ne">TypeError</span><span class="p">(</span>
<span class="sa">f</span><span class="s2">&quot;Unexpected words_list type=</span><span class="si">{</span><span class="nb">type</span><span class="p">(</span><span class="n">words_list</span><span class="p">)</span><span class="si">}</span><span class="s2">. Only list, np.ndarray, and torch.Tensor are supported.&quot;</span>
<span class="p">)</span>
<span class="k">if</span> <span class="n">cross_attention_masks</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
<span class="n">encoder_input_features</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">concat</span><span class="p">(</span><span class="n">encoder_input_features</span><span class="p">)</span>
<span class="n">encoder_output_lengths</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">concat</span><span class="p">(</span><span class="n">encoder_output_lengths</span><span class="p">)</span>
<span class="n">sampling_config</span><span class="o">.</span><span class="n">bad_words_list</span> <span class="o">=</span> <span class="n">maybe_convert_to_words_list_format</span><span class="p">(</span>
<span class="n">sampling_config</span><span class="o">.</span><span class="n">bad_words_list</span><span class="p">)</span>
<span class="n">sampling_config</span><span class="o">.</span><span class="n">stop_words_list</span> <span class="o">=</span> <span class="n">maybe_convert_to_words_list_format</span><span class="p">(</span>
<span class="n">sampling_config</span><span class="o">.</span><span class="n">stop_words_list</span><span class="p">)</span>
<span class="k">if</span> <span class="ow">not</span> <span class="bp">self</span><span class="o">.</span><span class="n">kv_cache_type</span> <span class="ow">and</span> <span class="n">sampling_config</span><span class="o">.</span><span class="n">max_new_tokens</span> <span class="o">&gt;</span> <span class="mi">1</span><span class="p">:</span>
<span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span>
<span class="s1">&#39;Disabled KV cache is intended for context phase only now.&#39;</span><span class="p">)</span>
<span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">setup</span><span class="p">(</span>
<span class="n">batch_size</span><span class="o">=</span><span class="n">batch_size</span><span class="p">,</span>
<span class="n">max_context_length</span><span class="o">=</span><span class="n">input_lengths</span><span class="o">.</span><span class="n">max</span><span class="p">()</span><span class="o">.</span><span class="n">item</span><span class="p">(),</span>
<span class="n">max_new_tokens</span><span class="o">=</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">max_new_tokens</span><span class="p">,</span>
<span class="n">beam_width</span><span class="o">=</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">num_beams</span><span class="p">,</span>
<span class="n">max_attention_window_size</span><span class="o">=</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">max_attention_window_size</span><span class="p">,</span>
<span class="n">sink_token_length</span><span class="o">=</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">sink_token_length</span><span class="p">,</span>
<span class="n">lora_manager</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">lora_manager</span><span class="p">,</span>
<span class="n">lora_uids</span><span class="o">=</span><span class="n">lora_uids</span><span class="p">,</span>
<span class="n">medusa_choices</span><span class="o">=</span><span class="n">medusa_choices</span><span class="p">,</span>
<span class="n">enable_context_fmha_fp32_acc</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">enable_context_fmha_fp32_acc</span><span class="p">,</span>
<span class="n">multi_block_mode</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">multi_block_mode</span><span class="p">,</span>
<span class="n">encoder_max_input_length</span><span class="o">=</span><span class="n">encoder_max_input_length</span><span class="p">,</span>
<span class="p">)</span>
<span class="n">batch_input_ids</span> <span class="o">=</span> <span class="n">batch_input_ids</span><span class="o">.</span><span class="n">cuda</span><span class="p">()</span>
<span class="n">input_lengths</span> <span class="o">=</span> <span class="n">input_lengths</span><span class="o">.</span><span class="n">cuda</span><span class="p">()</span>
<span class="n">other_kwargs</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_prepare_ptuning</span><span class="p">(</span><span class="n">prompt_table</span><span class="p">,</span> <span class="n">prompt_tasks</span><span class="p">,</span>
<span class="n">batch_size</span><span class="p">)</span>
<span class="n">other_kwargs</span><span class="p">[</span><span class="s1">&#39;skip_cross_attn_blocks&#39;</span><span class="p">]</span> <span class="o">=</span> <span class="n">kwargs</span><span class="o">.</span><span class="n">get</span><span class="p">(</span>
<span class="s1">&#39;skip_cross_attn_blocks&#39;</span><span class="p">,</span> <span class="kc">None</span><span class="p">)</span>
<span class="n">outputs</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">decode</span><span class="p">(</span>
<span class="n">batch_input_ids</span><span class="p">,</span>
<span class="n">input_lengths</span><span class="p">,</span>
<span class="n">sampling_config</span><span class="p">,</span>
<span class="n">stop_words_list</span><span class="o">=</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">stop_words_list</span><span class="p">,</span>
<span class="n">bad_words_list</span><span class="o">=</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">bad_words_list</span><span class="p">,</span>
<span class="n">output_sequence_lengths</span><span class="o">=</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">output_sequence_lengths</span><span class="p">,</span>
<span class="n">return_dict</span><span class="o">=</span><span class="n">sampling_config</span><span class="o">.</span><span class="n">return_dict</span><span class="p">,</span>
<span class="n">streaming</span><span class="o">=</span><span class="n">streaming</span><span class="p">,</span>
<span class="n">stopping_criteria</span><span class="o">=</span><span class="n">stopping_criteria</span><span class="p">,</span>
<span class="n">logits_processor</span><span class="o">=</span><span class="n">logits_processor</span><span class="p">,</span>
<span class="n">position_ids</span><span class="o">=</span><span class="n">position_ids</span><span class="p">,</span>
<span class="n">encoder_output</span><span class="o">=</span><span class="n">encoder_input_features</span><span class="p">,</span>
<span class="n">encoder_input_lengths</span><span class="o">=</span><span class="n">encoder_output_lengths</span><span class="p">,</span>
<span class="n">cross_attention_mask</span><span class="o">=</span><span class="n">cross_attention_masks</span><span class="p">,</span>
<span class="o">**</span><span class="n">other_kwargs</span><span class="p">)</span>
<span class="k">if</span> <span class="n">sampling_config</span><span class="o">.</span><span class="n">return_dict</span><span class="p">:</span>
<span class="k">if</span> <span class="n">streaming</span><span class="p">:</span>
<span class="n">outputs</span> <span class="o">=</span> <span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">_prepare_outputs</span><span class="p">(</span><span class="n">curr_outputs</span><span class="p">,</span> <span class="n">input_lengths</span><span class="p">)</span>
<span class="k">for</span> <span class="n">curr_outputs</span> <span class="ow">in</span> <span class="n">outputs</span><span class="p">)</span>
<span class="k">else</span><span class="p">:</span>
<span class="n">outputs</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_prepare_outputs</span><span class="p">(</span><span class="n">outputs</span><span class="p">,</span> <span class="n">input_lengths</span><span class="p">)</span>
<span class="k">return</span> <span class="n">outputs</span></div>
<div class="viewcode-block" id="ModelRunner.serialize_engine">
<a class="viewcode-back" href="../../../python-api/tensorrt_llm.runtime.html#tensorrt_llm.runtime.ModelRunner.serialize_engine">[docs]</a>
<span class="k">def</span> <span class="nf">serialize_engine</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="n">trt</span><span class="o">.</span><span class="n">IHostMemory</span><span class="p">:</span>
<span class="w"> </span><span class="sd">&quot;&quot;&quot;</span>
<span class="sd"> Serialize the engine.</span>
<span class="sd"> Returns:</span>
<span class="sd"> bytes: The serialized engine.</span>
<span class="sd"> &quot;&quot;&quot;</span>
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">session</span><span class="o">.</span><span class="n">runtime</span><span class="o">.</span><span class="n">_serialize_engine</span><span class="p">()</span></div>
</div>
</pre></div>
</article>
<footer class="prev-next-footer d-print-none">
<div class="prev-next-area">
</div>
</footer>
</div>
<div class="bd-sidebar-secondary"></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 © 2024, NVidia.
<br/>
</p>
</div>
</div>
</div>
</footer>
</body>
</html>
Reference in New Issue View Git Blame Copy Permalink