mirror of
https://github.com/NVIDIA/TensorRT-LLM.git
synced 2026-01-14 06:27:45 +08:00
762 lines
94 KiB
HTML
762 lines
94 KiB
HTML
|
|
|
|
<!DOCTYPE html>
|
|
<html class="writer-html5" lang="en" data-content_root="../../../">
|
|
<head>
|
|
<meta charset="utf-8" />
|
|
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
|
|
<title>tensorrt_llm.llmapi.llm — tensorrt_llm documentation</title>
|
|
<link rel="stylesheet" type="text/css" href="../../../_static/pygments.css?v=80d5e7a1" />
|
|
<link rel="stylesheet" type="text/css" href="../../../_static/css/theme.css?v=e59714d7" />
|
|
<link rel="stylesheet" type="text/css" href="../../../_static/copybutton.css?v=76b2166b" />
|
|
|
|
|
|
<script src="../../../_static/jquery.js?v=5d32c60e"></script>
|
|
<script src="../../../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
|
|
<script src="../../../_static/documentation_options.js?v=5929fcd5"></script>
|
|
<script src="../../../_static/doctools.js?v=888ff710"></script>
|
|
<script src="../../../_static/sphinx_highlight.js?v=dc90522c"></script>
|
|
<script src="../../../_static/clipboard.min.js?v=a7894cd8"></script>
|
|
<script src="../../../_static/copybutton.js?v=65e89d2a"></script>
|
|
<script src="../../../_static/js/theme.js"></script>
|
|
<link rel="index" title="Index" href="../../../genindex.html" />
|
|
<link rel="search" title="Search" href="../../../search.html" />
|
|
</head>
|
|
|
|
<body class="wy-body-for-nav">
|
|
<div class="wy-grid-for-nav">
|
|
<nav data-toggle="wy-nav-shift" class="wy-nav-side">
|
|
<div class="wy-side-scroll">
|
|
<div class="wy-side-nav-search" >
|
|
|
|
|
|
|
|
<a href="../../../index.html" class="icon icon-home">
|
|
tensorrt_llm
|
|
</a>
|
|
<div role="search">
|
|
<form id="rtd-search-form" class="wy-form" action="../../../search.html" method="get">
|
|
<input type="text" name="q" placeholder="Search docs" aria-label="Search docs" />
|
|
<input type="hidden" name="check_keywords" value="yes" />
|
|
<input type="hidden" name="area" value="default" />
|
|
</form>
|
|
</div>
|
|
</div><div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="Navigation menu">
|
|
<p class="caption" role="heading"><span class="caption-text">Getting Started</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../overview.html">Overview</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../quick-start-guide.html">Quick Start Guide</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../key-features.html">Key Features</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../release-notes.html">Release Notes</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">Installation</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../installation/linux.html">Installing on Linux</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../installation/build-from-source-linux.html">Building from Source Code on Linux</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../installation/windows.html">Installing on Windows</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../installation/build-from-source-windows.html">Building from Source Code on Windows</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../installation/grace-hopper.html">Installing on Grace Hopper</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">LLM API</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../llm-api/index.html">API Introduction</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../llm-api/reference.html">API Reference</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">LLM API Examples</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../llm-api-examples/index.html">LLM Examples Introduction</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../llm-api-examples/customization.html">Common Customizations</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../llm-api-examples/llm_api_examples.html">Examples</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">Model Definition API</span></p>
|
|
<ul>
|
|
<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 class="caption" role="heading"><span class="caption-text">C++ API</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../_cpp_gen/executor.html">Executor</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../_cpp_gen/runtime.html">Runtime</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">Command-Line Reference</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../commands/trtllm-build.html">trtllm-build</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../commands/trtllm-serve.html">trtllm-serve</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">Architecture</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/overview.html">TensorRT-LLM Architecture</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/core-concepts.html">Model Definition</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/core-concepts.html#compilation">Compilation</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/core-concepts.html#runtime">Runtime</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/core-concepts.html#multi-gpu-and-multi-node-support">Multi-GPU and Multi-Node Support</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/checkpoint.html">TensorRT-LLM Checkpoint</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/workflow.html">TensorRT-LLM Build Workflow</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/add-model.html">Adding a Model</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">Advanced</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/gpt-attention.html">Multi-Head, Multi-Query, and Group-Query Attention</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/gpt-runtime.html">C++ GPT Runtime</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/executor.html">Executor API</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/graph-rewriting.html">Graph Rewriting Module</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/inference-request.html">Inference Request</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/inference-request.html#responses">Responses</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/lora.html">Run gpt-2b + LoRA using GptManager / cpp runtime</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/expert-parallelism.html">Expert Parallelism in TensorRT-LLM</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/kv-cache-reuse.html">KV cache reuse</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../advanced/speculative-decoding.html">Speculative Sampling</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">Performance</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../performance/perf-overview.html">Overview</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../performance/perf-benchmarking.html">Benchmarking</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../performance/perf-best-practices.html">Best Practices</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../performance/perf-analysis.html">Performance Analysis</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">Reference</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../reference/troubleshooting.html">Troubleshooting</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../reference/support-matrix.html">Support Matrix</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../reference/precision.html">Numerical Precision</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../reference/memory.html">Memory Usage of TensorRT-LLM</a></li>
|
|
</ul>
|
|
<p class="caption" role="heading"><span class="caption-text">Blogs</span></p>
|
|
<ul>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/H100vsA100.html">H100 has 4.6x A100 Performance in TensorRT-LLM, achieving 10,000 tok/s at 100ms to first token</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/H200launch.html">H200 achieves nearly 12,000 tokens/sec on Llama2-13B with TensorRT-LLM</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/Falcon180B-H200.html">Falcon-180B on a single H200 GPU with INT4 AWQ, and 6.7x faster Llama-70B over A100</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/quantization-in-TRT-LLM.html">Speed up inference with SOTA quantization techniques in TRT-LLM</a></li>
|
|
<li class="toctree-l1"><a class="reference internal" href="../../../blogs/XQA-kernel.html">New XQA-kernel provides 2.4x more Llama-70B throughput within the same latency budget</a></li>
|
|
</ul>
|
|
|
|
</div>
|
|
</div>
|
|
</nav>
|
|
|
|
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap"><nav class="wy-nav-top" aria-label="Mobile navigation menu" >
|
|
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
|
|
<a href="../../../index.html">tensorrt_llm</a>
|
|
</nav>
|
|
|
|
<div class="wy-nav-content">
|
|
<div class="rst-content">
|
|
<div role="navigation" aria-label="Page navigation">
|
|
<ul class="wy-breadcrumbs">
|
|
<li><a href="../../../index.html" class="icon icon-home" aria-label="Home"></a></li>
|
|
<li class="breadcrumb-item"><a href="../../index.html">Module code</a></li>
|
|
<li class="breadcrumb-item active">tensorrt_llm.llmapi.llm</li>
|
|
<li class="wy-breadcrumbs-aside">
|
|
</li>
|
|
</ul>
|
|
<hr/>
|
|
</div>
|
|
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
|
|
<div itemprop="articleBody">
|
|
|
|
<h1>Source code for tensorrt_llm.llmapi.llm</h1><div class="highlight"><pre>
|
|
<span></span><span class="kn">import</span> <span class="nn">json</span>
|
|
<span class="kn">import</span> <span class="nn">os</span>
|
|
<span class="kn">import</span> <span class="nn">shutil</span>
|
|
<span class="kn">import</span> <span class="nn">tempfile</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">Any</span><span class="p">,</span> <span class="n">List</span><span class="p">,</span> <span class="n">Literal</span><span class="p">,</span> <span class="n">Optional</span><span class="p">,</span> <span class="n">Sequence</span><span class="p">,</span> <span class="n">Union</span>
|
|
|
|
<span class="kn">from</span> <span class="nn">tqdm</span> <span class="kn">import</span> <span class="n">tqdm</span>
|
|
<span class="kn">from</span> <span class="nn">transformers</span> <span class="kn">import</span> <span class="n">PreTrainedTokenizerBase</span>
|
|
|
|
<span class="kn">from</span> <span class="nn">..</span> <span class="kn">import</span> <span class="n">bindings</span> <span class="k">as</span> <span class="n">tllm</span>
|
|
<span class="kn">from</span> <span class="nn">..bindings</span> <span class="kn">import</span> <span class="n">executor</span> <span class="k">as</span> <span class="n">tllm</span>
|
|
<span class="kn">from</span> <span class="nn">..builder</span> <span class="kn">import</span> <span class="n">EngineConfig</span>
|
|
<span class="kn">from</span> <span class="nn">..executor</span> <span class="kn">import</span> <span class="p">(</span><span class="n">GenerationExecutor</span><span class="p">,</span> <span class="n">GenerationResult</span><span class="p">,</span> <span class="n">LoRARequest</span><span class="p">,</span>
|
|
<span class="n">PromptAdapterRequest</span><span class="p">)</span>
|
|
<span class="kn">from</span> <span class="nn">..inputs</span> <span class="kn">import</span> <span class="n">PromptInputs</span><span class="p">,</span> <span class="n">create_input_processor</span><span class="p">,</span> <span class="n">prompt_inputs</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">..sampling_params</span> <span class="kn">import</span> <span class="n">SamplingParams</span>
|
|
<span class="kn">from</span> <span class="nn">.llm_utils</span> <span class="kn">import</span> <span class="p">(</span><span class="n">LLMARGS_DOCSTRING</span><span class="p">,</span> <span class="n">CachedModelLoader</span><span class="p">,</span> <span class="n">LlmArgs</span><span class="p">,</span>
|
|
<span class="n">LlmBuildStats</span><span class="p">,</span> <span class="n">ModelLoader</span><span class="p">,</span> <span class="n">_ModelRuntimeContext</span><span class="p">)</span>
|
|
<span class="kn">from</span> <span class="nn">.mpi_session</span> <span class="kn">import</span> <span class="p">(</span><span class="n">MpiCommSession</span><span class="p">,</span> <span class="n">MpiPoolSession</span><span class="p">,</span> <span class="n">MpiSession</span><span class="p">,</span>
|
|
<span class="n">external_mpi_comm_available</span><span class="p">)</span>
|
|
<span class="kn">from</span> <span class="nn">.tokenizer</span> <span class="kn">import</span> <span class="n">TokenizerBase</span><span class="p">,</span> <span class="n">_xgrammar_tokenizer_info</span>
|
|
<span class="c1"># TODO[chunweiy]: move the following symbols back to utils scope, and remove the following import</span>
|
|
<span class="kn">from</span> <span class="nn">.utils</span> <span class="kn">import</span> <span class="n">append_docstring</span><span class="p">,</span> <span class="n">exception_handler</span><span class="p">,</span> <span class="n">get_device_count</span>
|
|
|
|
|
|
<div class="viewcode-block" id="RequestOutput">
|
|
<a class="viewcode-back" href="../../../llm-api/reference.html#tensorrt_llm.llmapi.RequestOutput">[docs]</a>
|
|
<span class="k">class</span> <span class="nc">RequestOutput</span><span class="p">(</span><span class="n">GenerationResult</span><span class="p">):</span>
|
|
<span class="w"> </span><span class="sd">"""The output data of a completion request to the LLM.</span>
|
|
|
|
<span class="sd"> Fields:</span>
|
|
<span class="sd"> request_id (int): The unique ID of the request.</span>
|
|
<span class="sd"> prompt (str, optional): The prompt string of the request.</span>
|
|
<span class="sd"> prompt_token_ids (List[int]): The token ids of the prompt.</span>
|
|
<span class="sd"> outputs (List[CompletionOutput]): The output sequences of the request.</span>
|
|
<span class="sd"> context_logits (torch.Tensor, optional): The logits on the prompt token ids.</span>
|
|
<span class="sd"> finished (bool): Whether the whole request is finished.</span>
|
|
<span class="sd"> """</span>
|
|
|
|
<div class="viewcode-block" id="RequestOutput.__init__">
|
|
<a class="viewcode-back" href="../../../llm-api/reference.html#tensorrt_llm.llmapi.RequestOutput.__init__">[docs]</a>
|
|
<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">generation_result</span><span class="p">:</span> <span class="n">GenerationResult</span><span class="p">,</span>
|
|
<span class="n">prompt</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">tokenizer</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">TokenizerBase</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">)</span> <span class="o">-></span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="vm">__dict__</span><span class="o">.</span><span class="n">update</span><span class="p">(</span><span class="n">generation_result</span><span class="o">.</span><span class="vm">__dict__</span><span class="p">)</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">prompt</span> <span class="o">=</span> <span class="n">prompt</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span> <span class="o">=</span> <span class="n">tokenizer</span></div>
|
|
|
|
|
|
<div class="viewcode-block" id="RequestOutput.handle_response">
|
|
<a class="viewcode-back" href="../../../llm-api/reference.html#tensorrt_llm.llmapi.RequestOutput.handle_response">[docs]</a>
|
|
<span class="k">def</span> <span class="nf">handle_response</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">response</span><span class="p">):</span>
|
|
<span class="nb">super</span><span class="p">()</span><span class="o">.</span><span class="n">handle_response</span><span class="p">(</span><span class="n">response</span><span class="p">)</span>
|
|
|
|
<span class="n">sampling_params</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_generation_request</span><span class="o">.</span><span class="n">sampling_params</span>
|
|
<span class="n">kwargs</span> <span class="o">=</span> <span class="p">{</span>
|
|
<span class="s1">'skip_special_tokens'</span><span class="p">:</span>
|
|
<span class="n">sampling_params</span><span class="o">.</span><span class="n">skip_special_tokens</span><span class="p">,</span>
|
|
<span class="s1">'spaces_between_special_tokens'</span><span class="p">:</span>
|
|
<span class="n">sampling_params</span><span class="o">.</span><span class="n">spaces_between_special_tokens</span>
|
|
<span class="p">}</span>
|
|
<span class="k">if</span> <span class="n">sampling_params</span><span class="o">.</span><span class="n">detokenize</span> <span class="ow">and</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="k">for</span> <span class="n">beam_output</span> <span class="ow">in</span> <span class="bp">self</span><span class="o">.</span><span class="n">outputs</span><span class="p">:</span>
|
|
<span class="n">beam_output</span><span class="o">.</span><span class="n">_last_text_len</span> <span class="o">=</span> <span class="nb">len</span><span class="p">(</span><span class="n">beam_output</span><span class="o">.</span><span class="n">text</span><span class="p">)</span>
|
|
<span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="p">,</span> <span class="s1">'decode_incrementally'</span><span class="p">):</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">streaming</span> <span class="ow">and</span> <span class="ow">not</span> <span class="n">sampling_params</span><span class="o">.</span><span class="n">use_beam_search</span><span class="p">:</span>
|
|
<span class="n">beam_output</span><span class="o">.</span><span class="n">text</span><span class="p">,</span> <span class="n">beam_output</span><span class="o">.</span><span class="n">_incremental_states</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="o">.</span><span class="n">decode_incrementally</span><span class="p">(</span>
|
|
<span class="n">beam_output</span><span class="o">.</span><span class="n">token_ids_diff</span><span class="p">,</span>
|
|
<span class="n">prev_text</span><span class="o">=</span><span class="n">beam_output</span><span class="o">.</span><span class="n">text</span><span class="p">,</span>
|
|
<span class="n">states</span><span class="o">=</span><span class="n">beam_output</span><span class="o">.</span><span class="n">_incremental_states</span><span class="p">,</span>
|
|
<span class="n">flush</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">finished</span><span class="p">,</span>
|
|
<span class="o">**</span><span class="n">kwargs</span><span class="p">)</span>
|
|
<span class="k">else</span><span class="p">:</span>
|
|
<span class="n">beam_output</span><span class="o">.</span><span class="n">text</span><span class="p">,</span> <span class="n">_</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="o">.</span><span class="n">decode_incrementally</span><span class="p">(</span>
|
|
<span class="n">beam_output</span><span class="o">.</span><span class="n">token_ids</span><span class="p">,</span>
|
|
<span class="n">flush</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">finished</span><span class="p">,</span>
|
|
<span class="o">**</span><span class="n">kwargs</span><span class="p">)</span>
|
|
<span class="k">else</span><span class="p">:</span>
|
|
<span class="n">beam_output</span><span class="o">.</span><span class="n">text</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="o">.</span><span class="n">decode</span><span class="p">(</span>
|
|
<span class="n">beam_output</span><span class="o">.</span><span class="n">token_ids</span><span class="p">,</span> <span class="o">**</span><span class="n">kwargs</span><span class="p">)</span></div>
|
|
|
|
|
|
<span class="k">def</span> <span class="nf">_repr_fields</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
|
<span class="k">return</span> <span class="p">[</span>
|
|
<span class="s2">"request_id"</span><span class="p">,</span> <span class="s2">"prompt"</span><span class="p">,</span> <span class="s2">"prompt_token_ids"</span><span class="p">,</span> <span class="s2">"outputs"</span><span class="p">,</span> <span class="s2">"finished"</span>
|
|
<span class="p">]</span></div>
|
|
|
|
|
|
|
|
<div class="viewcode-block" id="LLM">
|
|
<a class="viewcode-back" href="../../../llm-api/reference.html#tensorrt_llm.llmapi.LLM">[docs]</a>
|
|
<span class="nd">@append_docstring</span><span class="p">(</span><span class="n">LLMARGS_DOCSTRING</span><span class="p">)</span>
|
|
<span class="k">class</span> <span class="nc">LLM</span><span class="p">:</span>
|
|
<span class="w"> </span><span class="sd">"""LLM class is the main class for running a LLM model.</span>
|
|
|
|
<span class="sd"> Args:</span>
|
|
<span class="sd"> """</span>
|
|
|
|
<div class="viewcode-block" id="LLM.__init__">
|
|
<a class="viewcode-back" href="../../../llm-api/reference.html#tensorrt_llm.llmapi.LLM.__init__">[docs]</a>
|
|
<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">model</span><span class="p">:</span> <span class="nb">str</span><span class="p">,</span>
|
|
<span class="n">tokenizer</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">Path</span><span class="p">,</span> <span class="n">TokenizerBase</span><span class="p">,</span>
|
|
<span class="n">PreTrainedTokenizerBase</span><span class="p">]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">tokenizer_mode</span><span class="p">:</span> <span class="n">Literal</span><span class="p">[</span><span class="s1">'auto'</span><span class="p">,</span> <span class="s1">'slow'</span><span class="p">]</span> <span class="o">=</span> <span class="s1">'auto'</span><span class="p">,</span>
|
|
<span class="n">skip_tokenizer_init</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">trust_remote_code</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">tensor_parallel_size</span><span class="p">:</span> <span class="nb">int</span> <span class="o">=</span> <span class="mi">1</span><span class="p">,</span>
|
|
<span class="n">dtype</span><span class="p">:</span> <span class="nb">str</span> <span class="o">=</span> <span class="s2">"auto"</span><span class="p">,</span>
|
|
<span class="n">revision</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">tokenizer_revision</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">speculative_model</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="o">**</span><span class="n">kwargs</span><span class="p">:</span> <span class="n">Any</span><span class="p">):</span>
|
|
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_executor_cls</span> <span class="o">=</span> <span class="n">kwargs</span><span class="o">.</span><span class="n">pop</span><span class="p">(</span><span class="s2">"executor_cls"</span><span class="p">,</span> <span class="n">GenerationExecutor</span><span class="p">)</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">MpiSession</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span>
|
|
|
|
<span class="k">try</span><span class="p">:</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">args</span> <span class="o">=</span> <span class="n">LlmArgs</span><span class="o">.</span><span class="n">from_kwargs</span><span class="p">(</span>
|
|
<span class="n">model</span><span class="o">=</span><span class="n">model</span><span class="p">,</span>
|
|
<span class="n">tokenizer</span><span class="o">=</span><span class="n">tokenizer</span><span class="p">,</span>
|
|
<span class="n">tokenizer_mode</span><span class="o">=</span><span class="n">tokenizer_mode</span><span class="p">,</span>
|
|
<span class="n">skip_tokenizer_init</span><span class="o">=</span><span class="n">skip_tokenizer_init</span><span class="p">,</span>
|
|
<span class="n">trust_remote_code</span><span class="o">=</span><span class="n">trust_remote_code</span><span class="p">,</span>
|
|
<span class="n">tensor_parallel_size</span><span class="o">=</span><span class="n">tensor_parallel_size</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">revision</span><span class="o">=</span><span class="n">revision</span><span class="p">,</span>
|
|
<span class="n">tokenizer_revision</span><span class="o">=</span><span class="n">tokenizer_revision</span><span class="p">,</span>
|
|
<span class="n">speculative_model</span><span class="o">=</span><span class="n">speculative_model</span><span class="p">,</span>
|
|
<span class="o">**</span><span class="n">kwargs</span><span class="p">)</span>
|
|
<span class="k">except</span> <span class="ne">Exception</span> <span class="k">as</span> <span class="n">e</span><span class="p">:</span>
|
|
<span class="n">logger</span><span class="o">.</span><span class="n">error</span><span class="p">(</span>
|
|
<span class="sa">f</span><span class="s2">"Failed to parse the arguments for the LLM constructor: </span><span class="si">{</span><span class="n">e</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span>
|
|
<span class="k">raise</span> <span class="n">e</span>
|
|
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">is_multi_gpu</span><span class="p">:</span>
|
|
<span class="k">if</span> <span class="n">get_device_count</span><span class="p">()</span> <span class="o"><</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">world_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">"Only </span><span class="si">{</span><span class="n">get_device_count</span><span class="p">()</span><span class="si">}</span><span class="s2"> GPUs are available, but </span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">world_size</span><span class="si">}</span><span class="s2"> are required."</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">'start MpiSession with </span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">world_size</span><span class="si">}</span><span class="s1"> workers'</span>
|
|
<span class="p">)</span>
|
|
<span class="k">if</span> <span class="ow">not</span> <span class="n">external_mpi_comm_available</span><span class="p">(</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">world_size</span><span class="p">):</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span> <span class="o">=</span> <span class="n">MpiPoolSession</span><span class="p">(</span>
|
|
<span class="n">n_workers</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">world_size</span><span class="p">)</span>
|
|
<span class="k">else</span><span class="p">:</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span> <span class="o">=</span> <span class="n">MpiCommSession</span><span class="p">(</span>
|
|
<span class="n">n_workers</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">world_size</span><span class="p">)</span>
|
|
|
|
<span class="k">try</span><span class="p">:</span>
|
|
<span class="c1"># Due to the Executor can only accept a engine path, we need to save the engine to a directory</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">Path</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_executor</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">GenerationExecutor</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_workspace</span> <span class="o">=</span> <span class="n">tempfile</span><span class="o">.</span><span class="n">TemporaryDirectory</span><span class="p">(</span>
|
|
<span class="n">suffix</span><span class="o">=</span><span class="s2">"-llm-workspace"</span><span class="p">,</span> <span class="nb">dir</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">workspace</span><span class="p">)</span>
|
|
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_hf_model_dir</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">Path</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span>
|
|
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">runtime_context</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">_ModelRuntimeContext</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">llm_build_stats</span> <span class="o">=</span> <span class="n">LlmBuildStats</span><span class="p">()</span>
|
|
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_build_model</span><span class="p">()</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">input_processor</span> <span class="o">=</span> <span class="n">create_input_processor</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="p">)</span>
|
|
<span class="k">except</span> <span class="ne">Exception</span> <span class="k">as</span> <span class="n">e</span><span class="p">:</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span><span class="o">.</span><span class="n">shutdown</span><span class="p">()</span>
|
|
<span class="k">raise</span> <span class="n">e</span>
|
|
|
|
<span class="n">exception_handler</span><span class="o">.</span><span class="n">register</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="s1">'_shutdown'</span><span class="p">)</span></div>
|
|
|
|
|
|
<span class="nd">@property</span>
|
|
<span class="k">def</span> <span class="nf">workspace</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-></span> <span class="n">Path</span><span class="p">:</span>
|
|
<span class="k">return</span> <span class="n">Path</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">_workspace</span><span class="o">.</span><span class="n">name</span><span class="p">)</span>
|
|
|
|
<div class="viewcode-block" id="LLM.generate">
|
|
<a class="viewcode-back" href="../../../llm-api/reference.html#tensorrt_llm.llmapi.LLM.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">inputs</span><span class="p">:</span> <span class="n">Union</span><span class="p">[</span><span class="n">PromptInputs</span><span class="p">,</span> <span class="n">Sequence</span><span class="p">[</span><span class="n">PromptInputs</span><span class="p">]],</span>
|
|
<span class="n">sampling_params</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="n">SamplingParams</span><span class="p">,</span>
|
|
<span class="n">List</span><span class="p">[</span><span class="n">SamplingParams</span><span class="p">]]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">use_tqdm</span><span class="p">:</span> <span class="nb">bool</span> <span class="o">=</span> <span class="kc">True</span><span class="p">,</span>
|
|
<span class="n">lora_request</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="n">LoRARequest</span><span class="p">,</span>
|
|
<span class="n">Sequence</span><span class="p">[</span><span class="n">LoRARequest</span><span class="p">]]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">prompt_adapter_request</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="n">PromptAdapterRequest</span><span class="p">,</span> <span class="n">Sequence</span><span class="p">[</span><span class="n">PromptAdapterRequest</span><span class="p">]]]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">queries</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="n">PromptInputs</span><span class="p">,</span> <span class="n">Sequence</span><span class="p">[</span><span class="n">PromptInputs</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">-></span> <span class="n">Union</span><span class="p">[</span><span class="n">RequestOutput</span><span class="p">,</span> <span class="n">List</span><span class="p">[</span><span class="n">RequestOutput</span><span class="p">]]:</span>
|
|
<span class="w"> </span><span class="sd">"""Generate output for the given prompts in the synchronous mode.</span>
|
|
<span class="sd"> Synchronous generation accepts either single prompt or batched prompts.</span>
|
|
|
|
<span class="sd"> Args:</span>
|
|
<span class="sd"> inputs (PromptInputs or Sequence[PromptInputs]): The prompt text or token ids.</span>
|
|
<span class="sd"> it can be single prompt or batched prompts.</span>
|
|
<span class="sd"> sampling_params (SamplingParams, List[SamplingParams], optional): The sampling params for the</span>
|
|
<span class="sd"> generation, a default one will be used if not provided. Defaults to None.</span>
|
|
<span class="sd"> use_tqdm (bool): Whether to use tqdm to display the progress bar. Defaults to True.</span>
|
|
<span class="sd"> lora_request (LoRARequest, Sequence[LoRARequest], optional): LoRA request to use for generation,</span>
|
|
<span class="sd"> if any. Defaults to None.</span>
|
|
<span class="sd"> prompt_adapter_request (PromptAdapterRequest, Sequence[PromptAdapterRequest], optional):</span>
|
|
<span class="sd"> Prompt Adapter request to use for generation, if any. Defaults to None.</span>
|
|
<span class="sd"> queries (PromptInputs or Sequence[PromptInputs]): The query text or token ids.</span>
|
|
<span class="sd"> it can be single prompt or batched prompts. it is used for star attention to run long context tasks.</span>
|
|
<span class="sd"> Returns:</span>
|
|
<span class="sd"> Union[RequestOutput, List[RequestOutput]]: The output data of the completion request to the LLM.</span>
|
|
<span class="sd"> """</span>
|
|
<span class="n">unbatched</span> <span class="o">=</span> <span class="ow">not</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">inputs</span><span class="p">,</span> <span class="nb">list</span><span class="p">)</span>
|
|
<span class="k">if</span> <span class="ow">not</span> <span class="n">unbatched</span><span class="p">:</span>
|
|
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">inputs</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="nb">int</span><span class="p">):</span>
|
|
<span class="n">unbatched</span> <span class="o">=</span> <span class="kc">True</span>
|
|
|
|
<span class="k">if</span> <span class="n">unbatched</span><span class="p">:</span>
|
|
<span class="n">inputs</span> <span class="o">=</span> <span class="p">[</span><span class="n">inputs</span><span class="p">]</span>
|
|
<span class="k">if</span> <span class="n">queries</span><span class="p">:</span>
|
|
<span class="n">queries</span> <span class="o">=</span> <span class="p">[</span><span class="n">queries</span><span class="p">]</span>
|
|
|
|
<span class="n">inputs</span> <span class="o">=</span> <span class="p">[</span><span class="n">prompt_inputs</span><span class="p">(</span><span class="n">i</span><span class="p">)</span> <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="n">inputs</span><span class="p">]</span>
|
|
|
|
<span class="n">futures</span> <span class="o">=</span> <span class="p">[]</span>
|
|
<span class="k">for</span> <span class="n">i</span><span class="p">,</span> <span class="n">request_inputs</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span><span class="n">inputs</span><span class="p">):</span>
|
|
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">sampling_params</span><span class="p">,</span> <span class="nb">list</span><span class="p">):</span>
|
|
<span class="n">sp</span> <span class="o">=</span> <span class="n">sampling_params</span><span class="p">[</span><span class="n">i</span><span class="p">]</span>
|
|
<span class="k">else</span><span class="p">:</span>
|
|
<span class="n">sp</span> <span class="o">=</span> <span class="n">sampling_params</span>
|
|
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">lora_request</span><span class="p">,</span> <span class="nb">list</span><span class="p">):</span>
|
|
<span class="n">lora_req</span> <span class="o">=</span> <span class="n">lora_request</span><span class="p">[</span><span class="n">i</span><span class="p">]</span>
|
|
<span class="k">else</span><span class="p">:</span>
|
|
<span class="n">lora_req</span> <span class="o">=</span> <span class="n">lora_request</span>
|
|
<span class="k">if</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">prompt_adapter_request</span><span class="p">,</span> <span class="nb">list</span><span class="p">):</span>
|
|
<span class="n">pa_req</span> <span class="o">=</span> <span class="n">prompt_adapter_request</span><span class="p">[</span><span class="n">i</span><span class="p">]</span>
|
|
<span class="k">else</span><span class="p">:</span>
|
|
<span class="n">pa_req</span> <span class="o">=</span> <span class="n">prompt_adapter_request</span>
|
|
<span class="n">request_queries</span> <span class="o">=</span> <span class="kc">None</span> <span class="k">if</span> <span class="n">queries</span> <span class="ow">is</span> <span class="kc">None</span> <span class="k">else</span> <span class="n">queries</span><span class="p">[</span><span class="n">i</span><span class="p">]</span>
|
|
<span class="n">future</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">generate_async</span><span class="p">(</span><span class="n">request_inputs</span><span class="p">,</span>
|
|
<span class="n">queries</span><span class="o">=</span><span class="n">request_queries</span><span class="p">,</span>
|
|
<span class="n">sampling_params</span><span class="o">=</span><span class="n">sp</span><span class="p">,</span>
|
|
<span class="n">lora_request</span><span class="o">=</span><span class="n">lora_req</span><span class="p">,</span>
|
|
<span class="n">prompt_adapter_request</span><span class="o">=</span><span class="n">pa_req</span><span class="p">,</span>
|
|
<span class="n">streaming</span><span class="o">=</span><span class="kc">False</span><span class="p">)</span>
|
|
<span class="n">futures</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">future</span><span class="p">)</span>
|
|
|
|
<span class="k">for</span> <span class="n">future</span> <span class="ow">in</span> <span class="n">tqdm</span><span class="p">(</span><span class="n">futures</span><span class="p">,</span>
|
|
<span class="n">desc</span><span class="o">=</span><span class="s2">"Processed requests"</span><span class="p">,</span>
|
|
<span class="n">dynamic_ncols</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
|
|
<span class="n">disable</span><span class="o">=</span><span class="ow">not</span> <span class="n">use_tqdm</span><span class="p">):</span>
|
|
<span class="n">future</span><span class="o">.</span><span class="n">result</span><span class="p">()</span>
|
|
|
|
<span class="k">if</span> <span class="n">unbatched</span><span class="p">:</span>
|
|
<span class="n">futures</span> <span class="o">=</span> <span class="n">futures</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span>
|
|
|
|
<span class="k">return</span> <span class="n">futures</span></div>
|
|
|
|
|
|
<div class="viewcode-block" id="LLM.generate_async">
|
|
<a class="viewcode-back" href="../../../llm-api/reference.html#tensorrt_llm.llmapi.LLM.generate_async">[docs]</a>
|
|
<span class="k">def</span> <span class="nf">generate_async</span><span class="p">(</span>
|
|
<span class="bp">self</span><span class="p">,</span>
|
|
<span class="n">inputs</span><span class="p">:</span> <span class="n">PromptInputs</span><span class="p">,</span>
|
|
<span class="n">sampling_params</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">SamplingParams</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">lora_request</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">LoRARequest</span><span class="p">]</span> <span class="o">=</span> <span class="kc">None</span><span class="p">,</span>
|
|
<span class="n">prompt_adapter_request</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">PromptAdapterRequest</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">queries</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">PromptInputs</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">-></span> <span class="n">RequestOutput</span><span class="p">:</span>
|
|
<span class="w"> </span><span class="sd">"""Generate output for the given prompt in the asynchronous mode.</span>
|
|
<span class="sd"> Asynchronous generation accepts single prompt only.</span>
|
|
|
|
<span class="sd"> Args:</span>
|
|
<span class="sd"> inputs (PromptInputs): The prompt text or token ids; it must be single prompt.</span>
|
|
<span class="sd"> sampling_params (SamplingParams, optional): The sampling params for the generation,</span>
|
|
<span class="sd"> a default one will be used if not provided. Defaults to None.</span>
|
|
<span class="sd"> lora_request (LoRARequest, optional): LoRA request to use for generation, if any.</span>
|
|
<span class="sd"> Defaults to None.</span>
|
|
<span class="sd"> prompt_adapter_request (PromptAdapterRequest, optional): Prompt Adapter request to</span>
|
|
<span class="sd"> use for generation, if any. Defaults to None.</span>
|
|
<span class="sd"> streaming (bool): Whether to use the streaming mode for the generation. Defaults to</span>
|
|
<span class="sd"> False.</span>
|
|
<span class="sd"> queries (PromptInputs or Sequence[PromptInputs]): The query text or token ids.</span>
|
|
<span class="sd"> it can be single prompt or batched prompts. it is used for star attention to run long context tasks.</span>
|
|
|
|
<span class="sd"> Returns:</span>
|
|
<span class="sd"> RequestOutput: The output data of the completion request to the LLM.</span>
|
|
<span class="sd"> """</span>
|
|
<span class="n">sampling_params</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_prepare_sampling_params</span><span class="p">(</span><span class="n">sampling_params</span><span class="p">)</span>
|
|
|
|
<span class="n">inputs</span> <span class="o">=</span> <span class="n">prompt_inputs</span><span class="p">(</span><span class="n">inputs</span><span class="p">)</span>
|
|
|
|
<span class="n">query_token_ids</span> <span class="o">=</span> <span class="kc">None</span>
|
|
<span class="n">prompt_tuning_config</span> <span class="o">=</span> <span class="kc">None</span>
|
|
<span class="k">if</span> <span class="s2">"prompt_token_ids"</span> <span class="ow">in</span> <span class="n">inputs</span><span class="p">:</span>
|
|
<span class="n">prompt_token_ids</span> <span class="o">=</span> <span class="n">inputs</span><span class="p">[</span><span class="s1">'prompt_token_ids'</span><span class="p">]</span>
|
|
<span class="n">prompt</span> <span class="o">=</span> <span class="kc">None</span>
|
|
<span class="k">if</span> <span class="n">queries</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="n">query_token_ids</span> <span class="o">=</span> <span class="n">queries</span>
|
|
<span class="k">elif</span> <span class="s2">"prompt"</span> <span class="ow">in</span> <span class="n">inputs</span><span class="p">:</span>
|
|
<span class="n">prompt_token_ids</span><span class="p">,</span> <span class="n">extra_processed_inputs</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">input_processor</span><span class="p">(</span>
|
|
<span class="n">inputs</span><span class="p">,</span> <span class="n">sampling_params</span><span class="p">)</span>
|
|
<span class="n">prompt</span> <span class="o">=</span> <span class="n">inputs</span><span class="p">[</span><span class="s1">'prompt'</span><span class="p">]</span>
|
|
<span class="k">if</span> <span class="n">queries</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="n">query_token_ids</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">input_processor</span><span class="p">(</span><span class="n">queries</span><span class="p">,</span> <span class="n">sampling_params</span><span class="p">)</span>
|
|
<span class="k">if</span> <span class="n">extra_processed_inputs</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="n">prompt_tuning_config</span> <span class="o">=</span> <span class="n">extra_processed_inputs</span><span class="o">.</span><span class="n">get</span><span class="p">(</span>
|
|
<span class="s1">'prompt_tuning_config'</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">"The inputs must be type str or list of int, but got </span><span class="si">{</span><span class="nb">type</span><span class="p">(</span><span class="n">inputs</span><span class="p">)</span><span class="si">}</span><span class="s2">"</span>
|
|
<span class="p">)</span>
|
|
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_check_arguments</span><span class="p">(</span>
|
|
<span class="nb">len</span><span class="p">(</span><span class="n">prompt_token_ids</span><span class="p">),</span>
|
|
<span class="nb">len</span><span class="p">(</span><span class="n">query_token_ids</span><span class="p">)</span> <span class="k">if</span> <span class="n">query_token_ids</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span> <span class="k">else</span> <span class="mi">0</span><span class="p">,</span>
|
|
<span class="n">sampling_params</span><span class="p">)</span>
|
|
<span class="n">result</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_executor</span><span class="o">.</span><span class="n">generate_async</span><span class="p">(</span>
|
|
<span class="n">prompt_token_ids</span><span class="p">,</span>
|
|
<span class="n">query_token_ids</span><span class="o">=</span><span class="n">query_token_ids</span><span class="p">,</span>
|
|
<span class="n">sampling_params</span><span class="o">=</span><span class="n">sampling_params</span><span class="p">,</span>
|
|
<span class="n">lora_request</span><span class="o">=</span><span class="n">lora_request</span><span class="p">,</span>
|
|
<span class="n">prompt_adapter_request</span><span class="o">=</span><span class="n">prompt_adapter_request</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">prompt_tuning_config</span><span class="o">=</span><span class="n">prompt_tuning_config</span><span class="p">,</span>
|
|
<span class="p">)</span>
|
|
<span class="k">return</span> <span class="n">RequestOutput</span><span class="p">(</span><span class="n">result</span><span class="p">,</span> <span class="n">prompt</span><span class="p">,</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="p">)</span></div>
|
|
|
|
|
|
<span class="k">def</span> <span class="nf">_get_stats</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">timeout</span><span class="o">=</span><span class="kc">None</span><span class="p">)</span> <span class="o">-></span> <span class="nb">str</span><span class="p">:</span>
|
|
<span class="w"> </span><span class="sd">''' Get the stats from the runtime.</span>
|
|
|
|
<span class="sd"> Exceptions:</span>
|
|
<span class="sd"> NoStatsAvailable: If the stats are not available.</span>
|
|
|
|
<span class="sd"> Returns:</span>
|
|
<span class="sd"> str: The stats in JSON format.</span>
|
|
|
|
<span class="sd"> Known issue:</span>
|
|
<span class="sd"> The `_get_stats` cannot mix with `_get_stats_async` in the same LLM instance.</span>
|
|
<span class="sd"> '''</span>
|
|
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">_executor</span><span class="o">.</span><span class="n">get_stats</span><span class="p">(</span><span class="n">timeout</span><span class="o">=</span><span class="n">timeout</span><span class="p">)</span>
|
|
|
|
<span class="k">async</span> <span class="k">def</span> <span class="nf">_get_stats_async</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">timeout</span><span class="o">=</span><span class="kc">None</span><span class="p">)</span> <span class="o">-></span> <span class="nb">str</span><span class="p">:</span>
|
|
<span class="w"> </span><span class="sd">''' Get the stats from the runtime.</span>
|
|
|
|
<span class="sd"> Exceptions:</span>
|
|
<span class="sd"> NoStatsAvailable: If the stats are not available.</span>
|
|
|
|
<span class="sd"> Returns:</span>
|
|
<span class="sd"> str: The stats in JSON format.</span>
|
|
|
|
<span class="sd"> Known issue:</span>
|
|
<span class="sd"> The `_get_stats_async` cannot mix with `_get_stats` in the same LLM instance.</span>
|
|
<span class="sd"> '''</span>
|
|
<span class="k">return</span> <span class="k">await</span> <span class="bp">self</span><span class="o">.</span><span class="n">_executor</span><span class="o">.</span><span class="n">aget_stats</span><span class="p">(</span><span class="n">timeout</span><span class="o">=</span><span class="n">timeout</span><span class="p">)</span>
|
|
|
|
<span class="k">def</span> <span class="nf">_prepare_sampling_params</span><span class="p">(</span>
|
|
<span class="bp">self</span><span class="p">,</span>
|
|
<span class="n">sampling_params</span><span class="p">:</span> <span class="n">Optional</span><span class="p">[</span><span class="n">SamplingParams</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">SamplingParams</span><span class="p">:</span>
|
|
<span class="k">if</span> <span class="n">sampling_params</span> <span class="ow">is</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span> <span class="ow">is</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="s2">"tokenizer is required to initialize a default sampling_params, or you can explicitly specify a sampling_params"</span>
|
|
<span class="p">)</span>
|
|
<span class="k">return</span> <span class="n">SamplingParams</span><span class="p">(</span><span class="n">end_id</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="o">.</span><span class="n">eos_token_id</span><span class="p">,</span>
|
|
<span class="n">pad_id</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="o">.</span><span class="n">pad_token_id</span><span class="p">)</span>
|
|
<span class="k">elif</span> <span class="nb">isinstance</span><span class="p">(</span><span class="n">sampling_params</span><span class="p">,</span> <span class="n">SamplingParams</span><span class="p">):</span>
|
|
<span class="k">if</span> <span class="n">sampling_params</span><span class="o">.</span><span class="n">end_id</span> <span class="ow">is</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span> <span class="ow">is</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="s2">"tokenizer is required to reset end_id if it is None, or you can explicitly specify the end_id for sampling_params"</span>
|
|
<span class="p">)</span>
|
|
<span class="k">return</span> <span class="n">sampling_params</span><span class="o">.</span><span class="n">setup</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</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">"The sampling_params must be type SamplingParams or None, but got </span><span class="si">{</span><span class="nb">type</span><span class="p">(</span><span class="n">sampling_params</span><span class="p">)</span><span class="si">}</span><span class="s2">"</span>
|
|
<span class="p">)</span>
|
|
|
|
<span class="k">def</span> <span class="nf">_check_arguments</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">prompt_len</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span> <span class="n">query_len</span><span class="p">:</span> <span class="nb">int</span><span class="p">,</span>
|
|
<span class="n">sampling_params</span><span class="p">:</span> <span class="n">SamplingParams</span><span class="p">)</span> <span class="o">-></span> <span class="kc">None</span><span class="p">:</span>
|
|
|
|
<span class="n">build_config</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">build_config</span>
|
|
|
|
<span class="n">built_enging_cfg_file</span> <span class="o">=</span> <span class="n">Path</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">model</span><span class="p">)</span> <span class="o">/</span> <span class="s1">'config.json'</span>
|
|
<span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="n">built_enging_cfg_file</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
|
|
<span class="n">built_enging_cfg</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="n">max_seq_len</span> <span class="o">=</span> <span class="n">built_enging_cfg</span><span class="p">[</span><span class="s1">'build_config'</span><span class="p">][</span>
|
|
<span class="s1">'max_seq_len'</span><span class="p">]</span> <span class="k">if</span> <span class="s1">'build_config'</span> <span class="ow">in</span> <span class="n">built_enging_cfg</span> <span class="k">else</span> <span class="n">build_config</span><span class="o">.</span><span class="n">max_seq_len</span>
|
|
<span class="c1"># TODO: Remove this check and left the request verification to cpp runtime</span>
|
|
|
|
<span class="c1"># NOTE: [yuhangh] the meaning of max_seq_len should be for the all sequence. It's about position embedding.</span>
|
|
<span class="k">if</span> <span class="p">(</span><span class="ow">not</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">enable_chunked_prefill</span><span class="p">)</span> <span class="ow">and</span> <span class="p">(</span>
|
|
<span class="n">prompt_len</span> <span class="o">/</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">cp_size</span> <span class="o">+</span> <span class="n">query_len</span> <span class="o">+</span>
|
|
<span class="n">sampling_params</span><span class="o">.</span><span class="n">max_tokens</span> <span class="o">></span> <span class="n">max_seq_len</span><span class="p">):</span>
|
|
<span class="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span>
|
|
<span class="sa">f</span><span class="s2">"The sum of prompt length (</span><span class="si">{</span><span class="n">prompt_len</span><span class="o">/</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">cp_size</span><span class="si">}</span><span class="s2">) and query length (</span><span class="si">{</span><span class="n">query_len</span><span class="si">}</span><span class="s2">) max_tokens (</span><span class="si">{</span><span class="n">sampling_params</span><span class="o">.</span><span class="n">max_tokens</span><span class="si">}</span><span class="s2">) should not exceed "</span>
|
|
<span class="sa">f</span><span class="s2">"max_seq_len (</span><span class="si">{</span><span class="n">build_config</span><span class="o">.</span><span class="n">max_seq_len</span><span class="si">}</span><span class="s2">)"</span><span class="p">)</span>
|
|
|
|
<span class="k">if</span> <span class="n">sampling_params</span><span class="o">.</span><span class="n">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="k">raise</span> <span class="ne">ValueError</span><span class="p">(</span>
|
|
<span class="sa">f</span><span class="s2">"sampling_params's beam_width (</span><span class="si">{</span><span class="n">sampling_params</span><span class="o">.</span><span class="n">beam_width</span><span class="si">}</span><span class="s2">) should not exceed max_beam_width (</span><span class="si">{</span><span class="n">build_config</span><span class="o">.</span><span class="n">max_beam_width</span><span class="si">}</span><span class="s2">)"</span>
|
|
<span class="p">)</span>
|
|
|
|
<span class="k">def</span> <span class="nf">_build_model</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
|
<span class="n">model_loader</span> <span class="o">=</span> <span class="n">CachedModelLoader</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="p">,</span>
|
|
<span class="n">mpi_session</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span><span class="p">,</span>
|
|
<span class="n">workspace</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">workspace</span><span class="p">,</span>
|
|
<span class="n">llm_build_stats</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">llm_build_stats</span><span class="p">)</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span><span class="p">,</span> <span class="bp">self</span><span class="o">.</span><span class="n">_hf_model_dir</span> <span class="o">=</span> <span class="n">model_loader</span><span class="p">()</span>
|
|
<span class="c1"># update the model_dir to a local dir for the runtime, such as tokenizer loading.</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">model</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span>
|
|
|
|
<span class="c1"># Tokenizer loading should be after calling model_loader(), since model_loader() may download the model from HF hub.</span>
|
|
<span class="c1"># It should also be before bindings ExecutorConfig, which may depend on tokenizer info.</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_tokenizer</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_try_load_tokenizer</span><span class="p">()</span>
|
|
|
|
<span class="n">max_batch_size</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">max_batch_size</span> <span class="ow">or</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</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_num_tokens</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">max_num_tokens</span> <span class="ow">or</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">build_config</span><span class="o">.</span><span class="n">max_num_tokens</span>
|
|
<span class="n">executor_config</span> <span class="o">=</span> <span class="n">tllm</span><span class="o">.</span><span class="n">ExecutorConfig</span><span class="p">(</span>
|
|
<span class="n">max_beam_width</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</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">scheduler_config</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">scheduler_config</span><span class="p">,</span>
|
|
<span class="n">batching_type</span><span class="o">=</span><span class="n">tllm</span><span class="o">.</span><span class="n">BatchingType</span><span class="o">.</span><span class="n">INFLIGHT</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_num_tokens</span><span class="o">=</span><span class="n">max_num_tokens</span><span class="p">)</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">kv_cache_config</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">kv_cache_config</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">kv_cache_config</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">peft_cache_config</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">peft_cache_config</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">peft_cache_config</span>
|
|
<span class="k">elif</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</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">engine_config</span> <span class="o">=</span> <span class="n">EngineConfig</span><span class="o">.</span><span class="n">from_json_file</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span> <span class="o">/</span>
|
|
<span class="s2">"config.json"</span><span class="p">)</span>
|
|
<span class="n">lora_config</span> <span class="o">=</span> <span class="n">engine_config</span><span class="o">.</span><span class="n">build_config</span><span class="o">.</span><span class="n">lora_config</span>
|
|
<span class="n">max_lora_rank</span> <span class="o">=</span> <span class="n">lora_config</span><span class="o">.</span><span class="n">max_lora_rank</span>
|
|
<span class="n">num_lora_modules</span> <span class="o">=</span> <span class="n">engine_config</span><span class="o">.</span><span class="n">pretrained_config</span><span class="o">.</span><span class="n">num_hidden_layers</span> <span class="o">*</span> \
|
|
<span class="nb">len</span><span class="p">(</span><span class="n">lora_config</span><span class="o">.</span><span class="n">lora_target_modules</span> <span class="o">+</span> <span class="n">lora_config</span><span class="o">.</span><span class="n">missing_qkv_modules</span><span class="p">)</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">peft_cache_config</span> <span class="o">=</span> <span class="n">tllm</span><span class="o">.</span><span class="n">PeftCacheConfig</span><span class="p">(</span>
|
|
<span class="n">num_device_module_layer</span><span class="o">=</span><span class="n">max_lora_rank</span> <span class="o">*</span> <span class="n">num_lora_modules</span> <span class="o">*</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">max_loras</span><span class="p">,</span>
|
|
<span class="n">num_host_module_layer</span><span class="o">=</span><span class="n">max_lora_rank</span> <span class="o">*</span> <span class="n">num_lora_modules</span> <span class="o">*</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">max_cpu_loras</span><span class="p">,</span>
|
|
<span class="p">)</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">decoding_config</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">decoding_config</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">decoding_config</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">guided_decoding_backend</span> <span class="o">==</span> <span class="s1">'xgrammar'</span><span class="p">:</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">guided_decoding_config</span> <span class="o">=</span> <span class="n">tllm</span><span class="o">.</span><span class="n">GuidedDecodingConfig</span><span class="p">(</span>
|
|
<span class="n">backend</span><span class="o">=</span><span class="n">tllm</span><span class="o">.</span><span class="n">GuidedDecodingConfig</span><span class="o">.</span><span class="n">GuidedDecodingBackend</span><span class="o">.</span>
|
|
<span class="n">XGRAMMAR</span><span class="p">,</span>
|
|
<span class="o">**</span><span class="n">_xgrammar_tokenizer_info</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">tokenizer</span><span class="p">))</span>
|
|
<span class="k">elif</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">guided_decoding_backend</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="sa">f</span><span class="s2">"Unrecognized guided decoding backend </span><span class="si">{</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">guided_decoding_backend</span><span class="si">}</span><span class="s2">"</span>
|
|
<span class="p">)</span>
|
|
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">normalize_log_probs</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">normalize_log_probs</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">enable_chunked_context</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">enable_chunked_prefill</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">max_beam_width</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</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="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">extended_runtime_perf_knob_config</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">extended_runtime_perf_knob_config</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">extended_runtime_perf_knob_config</span>
|
|
|
|
<span class="n">trt_engine_dir</span> <span class="o">=</span> <span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span> <span class="k">else</span> <span class="kc">None</span><span class="p">)</span>
|
|
<span class="c1"># PIVOT_TO_PYTHON_START</span>
|
|
<span class="n">hf_model_dir</span> <span class="o">=</span> <span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">_hf_model_dir</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">_hf_model_dir</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span> <span class="k">else</span> <span class="kc">None</span><span class="p">)</span>
|
|
<span class="kn">from</span> <span class="nn">tensorrt_llm.pyexecutor.config</span> <span class="kn">import</span> <span class="n">update_executor_config</span>
|
|
<span class="n">update_executor_config</span><span class="p">(</span>
|
|
<span class="n">executor_config</span><span class="p">,</span>
|
|
<span class="n">backend</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">backend</span><span class="p">,</span>
|
|
<span class="n">pytorch_backend_config</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">pytorch_backend_config</span><span class="p">,</span>
|
|
<span class="n">max_seq_len</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">build_config</span><span class="o">.</span><span class="n">max_seq_len</span><span class="p">,</span>
|
|
<span class="n">hf_model_dir</span><span class="o">=</span><span class="n">hf_model_dir</span><span class="p">,</span>
|
|
<span class="n">trt_engine_dir</span><span class="o">=</span><span class="n">trt_engine_dir</span><span class="p">)</span>
|
|
<span class="n">executor_config</span><span class="o">.</span><span class="n">llm_parallel_config</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span>
|
|
<span class="c1"># PIVOT_TO_PYTHON_END</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_executor</span> <span class="o">=</span> <span class="bp">self</span><span class="o">.</span><span class="n">_executor_cls</span><span class="o">.</span><span class="n">create</span><span class="p">(</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span><span class="p">,</span>
|
|
<span class="n">executor_config</span><span class="o">=</span><span class="n">executor_config</span><span class="p">,</span>
|
|
<span class="n">logits_post_processor_map</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">logits_post_processor_map</span><span class="p">,</span>
|
|
<span class="n">model_world_size</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">world_size</span><span class="p">,</span>
|
|
<span class="n">mpi_session</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span><span class="p">,</span>
|
|
<span class="n">reuse_mpi_comm</span><span class="o">=</span><span class="n">external_mpi_comm_available</span><span class="p">(</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">parallel_config</span><span class="o">.</span><span class="n">world_size</span><span class="p">),</span>
|
|
<span class="p">)</span>
|
|
|
|
<span class="k">def</span> <span class="nf">_try_load_tokenizer</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-></span> <span class="n">Optional</span><span class="p">[</span><span class="n">TokenizerBase</span><span class="p">]:</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">skip_tokenizer_init</span><span class="p">:</span>
|
|
<span class="k">return</span> <span class="kc">None</span>
|
|
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">tokenizer</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="nb">isinstance</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">tokenizer</span><span class="p">,</span> <span class="n">TokenizerBase</span><span class="p">)</span>
|
|
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">tokenizer</span>
|
|
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">runtime_context</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">runtime_context</span><span class="o">.</span><span class="n">tokenizer</span>
|
|
|
|
<span class="k">return</span> <span class="n">ModelLoader</span><span class="o">.</span><span class="n">load_hf_tokenizer</span><span class="p">(</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">model</span><span class="p">,</span>
|
|
<span class="n">trust_remote_code</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">trust_remote_code</span><span class="p">,</span>
|
|
<span class="n">use_fast</span><span class="o">=</span><span class="bp">self</span><span class="o">.</span><span class="n">args</span><span class="o">.</span><span class="n">tokenizer_mode</span> <span class="o">!=</span> <span class="s1">'slow'</span><span class="p">)</span>
|
|
|
|
<span class="nd">@property</span>
|
|
<span class="k">def</span> <span class="nf">tokenizer</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-></span> <span class="n">Optional</span><span class="p">[</span><span class="n">TokenizerBase</span><span class="p">]:</span>
|
|
<span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">_tokenizer</span>
|
|
|
|
<div class="viewcode-block" id="LLM.save">
|
|
<a class="viewcode-back" href="../../../llm-api/reference.html#tensorrt_llm.llmapi.LLM.save">[docs]</a>
|
|
<span class="k">def</span> <span class="nf">save</span><span class="p">(</span><span class="bp">self</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="w"> </span><span class="sd">"""Save the built engine to the given path.</span>
|
|
|
|
<span class="sd"> Args:</span>
|
|
<span class="sd"> engine_dir (str): The path to save the engine.</span>
|
|
|
|
<span class="sd"> Returns:</span>
|
|
<span class="sd"> None</span>
|
|
<span class="sd"> """</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="s2">"Save model to </span><span class="si">{</span><span class="n">engine_dir</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span> <span class="ow">is</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span><span class="s2">"The engine is not built yet."</span><span class="p">)</span>
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span><span class="o">.</span><span class="n">absolute</span><span class="p">()</span> <span class="o">!=</span> <span class="n">os</span><span class="o">.</span><span class="n">path</span><span class="o">.</span><span class="n">abspath</span><span class="p">(</span><span class="n">engine_dir</span><span class="p">):</span>
|
|
<span class="n">shutil</span><span class="o">.</span><span class="n">copytree</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">_engine_dir</span><span class="p">,</span> <span class="n">engine_dir</span><span class="p">,</span> <span class="n">dirs_exist_ok</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span></div>
|
|
|
|
|
|
<span class="k">def</span> <span class="nf">_shutdown</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
|
<span class="k">if</span> <span class="nb">hasattr</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="s2">"_executor"</span><span class="p">)</span> <span class="ow">and</span> <span class="bp">self</span><span class="o">.</span><span class="n">_executor</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_executor</span><span class="o">.</span><span class="n">shutdown</span><span class="p">()</span>
|
|
|
|
<span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span> <span class="ow">is</span> <span class="ow">not</span> <span class="kc">None</span><span class="p">:</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span><span class="o">.</span><span class="n">shutdown</span><span class="p">()</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">mpi_session</span> <span class="o">=</span> <span class="kc">None</span>
|
|
|
|
<span class="k">def</span> <span class="fm">__enter__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
|
<span class="k">return</span> <span class="bp">self</span>
|
|
|
|
<span class="k">def</span> <span class="fm">__exit__</span><span class="p">(</span><span class="bp">self</span><span class="p">,</span> <span class="n">exc_type</span><span class="p">,</span> <span class="n">exc_value</span><span class="p">,</span> <span class="n">traceback</span><span class="p">)</span> <span class="o">-></span> <span class="nb">bool</span><span class="p">:</span>
|
|
<span class="k">del</span> <span class="n">exc_value</span><span class="p">,</span> <span class="n">traceback</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_shutdown</span><span class="p">()</span>
|
|
<span class="k">return</span> <span class="kc">False</span> <span class="c1"># propagate exceptions</span>
|
|
|
|
<span class="k">def</span> <span class="nf">__getstate__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
|
<span class="k">raise</span> <span class="ne">RuntimeError</span><span class="p">(</span><span class="s2">"LLM object can not be pickled."</span><span class="p">)</span>
|
|
|
|
<span class="k">def</span> <span class="fm">__del__</span><span class="p">(</span><span class="bp">self</span><span class="p">):</span>
|
|
<span class="bp">self</span><span class="o">.</span><span class="n">_shutdown</span><span class="p">()</span></div>
|
|
|
|
</pre></div>
|
|
|
|
</div>
|
|
</div>
|
|
<footer>
|
|
|
|
<hr/>
|
|
|
|
<div role="contentinfo">
|
|
<jinja2.runtime.BlockReference object at 0x7fed97af7950>
|
|
|
|
<div class="footer">
|
|
<p>
|
|
Copyright © 2024 NVIDIA Corporation
|
|
</p>
|
|
<p>
|
|
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/privacy-policy/" target="_blank" rel="noopener"
|
|
data-cms-ai="0">Privacy Policy</a> |
|
|
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/privacy-center/" target="_blank" rel="noopener"
|
|
data-cms-ai="0">Manage My Privacy</a> |
|
|
<a class="Link" href="https://www.nvidia.com/en-us/preferences/start/" target="_blank" rel="noopener"
|
|
data-cms-ai="0">Do Not Sell or Share My Data</a> |
|
|
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/terms-of-service/" target="_blank"
|
|
rel="noopener" data-cms-ai="0">Terms of Service</a> |
|
|
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/accessibility/" target="_blank" rel="noopener"
|
|
data-cms-ai="0">Accessibility</a> |
|
|
<a class="Link" href="https://www.nvidia.com/en-us/about-nvidia/company-policies/" target="_blank"
|
|
rel="noopener" data-cms-ai="0">Corporate Policies</a> |
|
|
<a class="Link" href="https://www.nvidia.com/en-us/product-security/" target="_blank" rel="noopener"
|
|
data-cms-ai="0">Product Security</a> |
|
|
<a class="Link" href="https://www.nvidia.com/en-us/contact/" target="_blank" rel="noopener"
|
|
data-cms-ai="0">Contact</a>
|
|
</p>
|
|
</div>
|
|
|
|
|
|
</div>
|
|
|
|
|
|
|
|
</footer>
|
|
</div>
|
|
</div>
|
|
</section>
|
|
</div>
|
|
<script>
|
|
jQuery(function () {
|
|
SphinxRtdTheme.Navigation.enable(true);
|
|
});
|
|
</script>
|
|
|
|
</body>
|
|
</html> |