chore: import upstream snapshot with attribution
This commit is contained in:
wehub-resource-sync
@@ -0,0 +1,443 @@
|
||||
<!DOCTYPE html>
|
||||
<html lang="zh">
|
||||
<head>
|
||||
<meta http-equiv="content-type" content="text/html;charset=utf-8"/>
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1.0"/>
|
||||
<meta name="description" content="使用 Fairscale 管道并行模块微调 GPT-NEOX 偏差"/>
|
||||
|
||||
<meta name="twitter:card" content="summary"/>
|
||||
<meta name="twitter:image:src" content="https://avatars1.githubusercontent.com/u/64068543?s=400&v=4"/>
|
||||
<meta name="twitter:title" content="Fine Tune GPT-NEOX"/>
|
||||
<meta name="twitter:description" content="使用 Fairscale 管道并行模块微调 GPT-NEOX 偏差"/>
|
||||
<meta name="twitter:site" content="@labmlai"/>
|
||||
<meta name="twitter:creator" content="@labmlai"/>
|
||||
|
||||
<meta property="og:url" content="https://nn.labml.ai/neox/samples/finetune.html"/>
|
||||
<meta property="og:title" content="Fine Tune GPT-NEOX"/>
|
||||
<meta property="og:image" content="https://avatars1.githubusercontent.com/u/64068543?s=400&v=4"/>
|
||||
<meta property="og:site_name" content="Fine Tune GPT-NEOX"/>
|
||||
<meta property="og:type" content="object"/>
|
||||
<meta property="og:title" content="Fine Tune GPT-NEOX"/>
|
||||
<meta property="og:description" content="使用 Fairscale 管道并行模块微调 GPT-NEOX 偏差"/>
|
||||
|
||||
<title>Fine Tune GPT-NEOX</title>
|
||||
<link rel="shortcut icon" href="/icon.png"/>
|
||||
<link rel="stylesheet" href="../../pylit.css?v=1">
|
||||
<link rel="canonical" href="https://nn.labml.ai/neox/samples/finetune.html"/>
|
||||
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.13.18/dist/katex.min.css" integrity="sha384-zTROYFVGOfTw7JV7KUu8udsvW2fx4lWOsCEDqhBreBwlHI4ioVRtmIvEThzJHGET" crossorigin="anonymous">
|
||||
|
||||
<!-- Global site tag (gtag.js) - Google Analytics -->
|
||||
<script async src="https://www.googletagmanager.com/gtag/js?id=G-4V3HC8HBLH"></script>
|
||||
<script>
|
||||
window.dataLayer = window.dataLayer || [];
|
||||
|
||||
function gtag() {
|
||||
dataLayer.push(arguments);
|
||||
}
|
||||
|
||||
gtag('js', new Date());
|
||||
|
||||
gtag('config', 'G-4V3HC8HBLH');
|
||||
</script>
|
||||
</head>
|
||||
<body>
|
||||
<div id='container'>
|
||||
<div id="background"></div>
|
||||
<div class='section'>
|
||||
<div class='docs'>
|
||||
<p>
|
||||
<a class="parent" href="/">home</a>
|
||||
<a class="parent" href="../index.html">neox</a>
|
||||
<a class="parent" href="index.html">samples</a>
|
||||
</p>
|
||||
<p>
|
||||
<a href="https://github.com/labmlai/annotated_deep_learning_paper_implementations" target="_blank">
|
||||
<img alt="Github"
|
||||
src="https://img.shields.io/github/stars/labmlai/annotated_deep_learning_paper_implementations?style=social"
|
||||
style="max-width:100%;"/></a>
|
||||
<a href="https://twitter.com/labmlai" rel="nofollow" target="_blank">
|
||||
<img alt="Twitter"
|
||||
src="https://img.shields.io/twitter/follow/labmlai?style=social"
|
||||
style="max-width:100%;"/></a>
|
||||
</p>
|
||||
<p>
|
||||
<a href="https://github.com/labmlai/annotated_deep_learning_paper_implementations/tree/master/labml_nn/neox/samples/finetune.py" target="_blank">
|
||||
View code on Github</a>
|
||||
</p>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-0'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-0'>#</a>
|
||||
</div>
|
||||
<h1>Fine Tune GPT-NEOX</h1>
|
||||
<p>这说明了如何利用管道并行度对 GPT-NEOX 进行微调。</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">13</span><span></span><span class="kn">import</span> <span class="nn">fairscale</span>
|
||||
<span class="lineno">14</span><span class="kn">import</span> <span class="nn">torch</span>
|
||||
<span class="lineno">15</span><span class="kn">import</span> <span class="nn">torch.nn</span> <span class="k">as</span> <span class="nn">nn</span>
|
||||
<span class="lineno">16</span><span class="kn">import</span> <span class="nn">torch.utils.data</span>
|
||||
<span class="lineno">17</span><span class="kn">import</span> <span class="nn">torch.utils.data</span>
|
||||
<span class="lineno">18</span><span class="kn">import</span> <span class="nn">typing</span>
|
||||
<span class="lineno">19</span><span class="kn">from</span> <span class="nn">torch.utils.data</span> <span class="kn">import</span> <span class="n">DataLoader</span><span class="p">,</span> <span class="n">RandomSampler</span>
|
||||
<span class="lineno">20</span>
|
||||
<span class="lineno">21</span><span class="kn">from</span> <span class="nn">labml</span> <span class="kn">import</span> <span class="n">experiment</span><span class="p">,</span> <span class="n">monit</span><span class="p">,</span> <span class="n">tracker</span><span class="p">,</span> <span class="n">lab</span>
|
||||
<span class="lineno">22</span><span class="kn">from</span> <span class="nn">labml.configs</span> <span class="kn">import</span> <span class="n">option</span>
|
||||
<span class="lineno">23</span><span class="kn">from</span> <span class="nn">labml.logger</span> <span class="kn">import</span> <span class="n">inspect</span>
|
||||
<span class="lineno">24</span><span class="kn">from</span> <span class="nn">labml_nn.neox.utils.text_dataset</span> <span class="kn">import</span> <span class="n">get_training_data</span>
|
||||
<span class="lineno">25</span><span class="kn">from</span> <span class="nn">labml_nn.neox.utils.finetune</span> <span class="kn">import</span> <span class="n">FineTuneBiases</span>
|
||||
<span class="lineno">26</span><span class="kn">from</span> <span class="nn">labml_nn.neox.model</span> <span class="kn">import</span> <span class="n">LayerGenerator</span><span class="p">,</span> <span class="n">NeoXModule</span>
|
||||
<span class="lineno">27</span><span class="kn">from</span> <span class="nn">labml_nn.neox.utils</span> <span class="kn">import</span> <span class="n">balance_layers_simple</span>
|
||||
<span class="lineno">28</span><span class="kn">from</span> <span class="nn">labml_nn.neox.utils.trainer</span> <span class="kn">import</span> <span class="n">PipelineParallelTrainerConf</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-1'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-1'>#</a>
|
||||
</div>
|
||||
<h3>加载 GPT-NEOX 图层</h3>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">31</span><span class="nd">@option</span><span class="p">(</span><span class="n">PipelineParallelTrainerConf</span><span class="o">.</span><span class="n">layers</span><span class="p">,</span> <span class="s1">'PipelineBiases'</span><span class="p">)</span>
|
||||
<span class="lineno">32</span><span class="k">def</span> <span class="nf">neox_layers</span><span class="p">(</span><span class="n">c</span><span class="p">:</span> <span class="n">PipelineParallelTrainerConf</span><span class="p">):</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-2'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-2'>#</a>
|
||||
</div>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">36</span> <span class="k">return</span> <span class="nb">list</span><span class="p">(</span><span class="n">LayerGenerator</span><span class="p">(</span><span class="n">is_clone_layers</span><span class="o">=</span><span class="n">c</span><span class="o">.</span><span class="n">is_clone_layers</span><span class="p">,</span>
|
||||
<span class="lineno">37</span> <span class="n">filter_layers</span><span class="o">=</span><span class="n">c</span><span class="o">.</span><span class="n">filter_layers</span><span class="p">,</span>
|
||||
<span class="lineno">38</span> <span class="n">dtype</span><span class="o">=</span><span class="n">c</span><span class="o">.</span><span class="n">dtype</span><span class="p">,</span>
|
||||
<span class="lineno">39</span> <span class="p">)</span><span class="o">.</span><span class="n">load</span><span class="p">())</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-3'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-3'>#</a>
|
||||
</div>
|
||||
<h3>为偏见创建精细的调谐器</h3>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">42</span><span class="nd">@option</span><span class="p">(</span><span class="n">PipelineParallelTrainerConf</span><span class="o">.</span><span class="n">fine_tuner</span><span class="p">,</span> <span class="s1">'PipelineBiases'</span><span class="p">)</span>
|
||||
<span class="lineno">43</span><span class="k">def</span> <span class="nf">fine_tune_biases</span><span class="p">(</span><span class="n">c</span><span class="p">:</span> <span class="n">PipelineParallelTrainerConf</span><span class="p">):</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-4'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-4'>#</a>
|
||||
</div>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">48</span> <span class="n">fine_tuner</span> <span class="o">=</span> <span class="n">FineTuneBiases</span><span class="p">(</span><span class="n">typing</span><span class="o">.</span><span class="n">cast</span><span class="p">(</span><span class="n">typing</span><span class="o">.</span><span class="n">List</span><span class="p">[</span><span class="n">NeoXModule</span><span class="p">],</span> <span class="n">c</span><span class="o">.</span><span class="n">layers</span><span class="p">))</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-5'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-5'>#</a>
|
||||
</div>
|
||||
<p>将偏见标记为可训练</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">50</span> <span class="n">fine_tuner</span><span class="o">.</span><span class="n">set_trainable_params</span><span class="p">()</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-6'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-6'>#</a>
|
||||
</div>
|
||||
<p></p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">53</span> <span class="k">return</span> <span class="n">fine_tuner</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-7'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-7'>#</a>
|
||||
</div>
|
||||
<h3>创建管道并行模型</h3>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">56</span><span class="nd">@option</span><span class="p">(</span><span class="n">PipelineParallelTrainerConf</span><span class="o">.</span><span class="n">model</span><span class="p">,</span> <span class="s1">'PipelineBiases'</span><span class="p">)</span>
|
||||
<span class="lineno">57</span><span class="k">def</span> <span class="nf">pipe_model</span><span class="p">(</span><span class="n">c</span><span class="p">:</span> <span class="n">PipelineParallelTrainerConf</span><span class="p">):</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-8'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-8'>#</a>
|
||||
</div>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">62</span> <span class="k">if</span> <span class="n">c</span><span class="o">.</span><span class="n">is_checkpointing</span><span class="p">:</span>
|
||||
<span class="lineno">63</span> <span class="k">raise</span> <span class="ne">NotImplementedError</span><span class="p">()</span>
|
||||
<span class="lineno">64</span> <span class="k">else</span><span class="p">:</span>
|
||||
<span class="lineno">65</span> <span class="n">layers</span> <span class="o">=</span> <span class="n">c</span><span class="o">.</span><span class="n">layers</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-9'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-9'>#</a>
|
||||
</div>
|
||||
<p>确保微调器已初始化</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">68</span> <span class="n">_</span> <span class="o">=</span> <span class="n">c</span><span class="o">.</span><span class="n">fine_tuner</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-10'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-10'>#</a>
|
||||
</div>
|
||||
<p>创建管道模块</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">71</span> <span class="k">with</span> <span class="n">monit</span><span class="o">.</span><span class="n">section</span><span class="p">(</span><span class="s1">'Pipe'</span><span class="p">):</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-11'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-11'>#</a>
|
||||
</div>
|
||||
<p>获取跨 GPU 的层分布</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">73</span> <span class="n">balance</span> <span class="o">=</span> <span class="n">balance_layers_simple</span><span class="p">(</span><span class="nb">len</span><span class="p">(</span><span class="n">layers</span><span class="p">),</span> <span class="n">c</span><span class="o">.</span><span class="n">n_gpus</span><span class="p">)</span>
|
||||
<span class="lineno">74</span> <span class="n">inspect</span><span class="p">(</span><span class="n">balance</span><span class="o">=</span><span class="n">balance</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-12'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-12'>#</a>
|
||||
</div>
|
||||
<p>每个 GPU 的设备</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">76</span> <span class="n">devices</span> <span class="o">=</span> <span class="p">[</span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="sa">f</span><span class="s1">'cuda:</span><span class="si">{</span><span class="n">i</span><span class="si">}</span><span class="s1">'</span><span class="p">)</span> <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="n">c</span><span class="o">.</span><span class="n">n_gpus</span><span class="p">)]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-13'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-13'>#</a>
|
||||
</div>
|
||||
<p>创建公平规模管道模块</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">78</span> <span class="n">pipe_model</span> <span class="o">=</span> <span class="n">fairscale</span><span class="o">.</span><span class="n">nn</span><span class="o">.</span><span class="n">Pipe</span><span class="p">(</span><span class="n">nn</span><span class="o">.</span><span class="n">Sequential</span><span class="p">(</span><span class="o">*</span><span class="n">layers</span><span class="p">),</span>
|
||||
<span class="lineno">79</span> <span class="n">balance</span><span class="o">=</span><span class="n">balance</span><span class="p">,</span>
|
||||
<span class="lineno">80</span> <span class="n">devices</span><span class="o">=</span><span class="n">devices</span><span class="p">,</span>
|
||||
<span class="lineno">81</span> <span class="n">chunks</span><span class="o">=</span><span class="n">c</span><span class="o">.</span><span class="n">chunks</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-14'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-14'>#</a>
|
||||
</div>
|
||||
<p></p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">84</span> <span class="k">return</span> <span class="n">pipe_model</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-15'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-15'>#</a>
|
||||
</div>
|
||||
<h4>小莎士比亚数据集</h4>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">87</span><span class="nd">@option</span><span class="p">(</span><span class="n">PipelineParallelTrainerConf</span><span class="o">.</span><span class="n">train_loader</span><span class="p">)</span>
|
||||
<span class="lineno">88</span><span class="k">def</span> <span class="nf">tiny_shakespeare</span><span class="p">(</span><span class="n">c</span><span class="p">:</span> <span class="n">PipelineParallelTrainerConf</span><span class="p">):</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-16'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-16'>#</a>
|
||||
</div>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">92</span> <span class="n">dataset</span> <span class="o">=</span> <span class="n">get_training_data</span><span class="p">(</span><span class="n">c</span><span class="o">.</span><span class="n">max_seq_len</span><span class="p">)</span>
|
||||
<span class="lineno">93</span>
|
||||
<span class="lineno">94</span> <span class="k">return</span> <span class="n">DataLoader</span><span class="p">(</span><span class="n">dataset</span><span class="p">,</span>
|
||||
<span class="lineno">95</span> <span class="n">batch_size</span><span class="o">=</span><span class="n">c</span><span class="o">.</span><span class="n">batch_size</span><span class="p">,</span>
|
||||
<span class="lineno">96</span> <span class="n">sampler</span><span class="o">=</span><span class="n">RandomSampler</span><span class="p">(</span><span class="n">dataset</span><span class="p">,</span> <span class="n">replacement</span><span class="o">=</span><span class="kc">True</span><span class="p">))</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-17'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-17'>#</a>
|
||||
</div>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">99</span><span class="k">def</span> <span class="nf">main</span><span class="p">():</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-18'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-18'>#</a>
|
||||
</div>
|
||||
<p>创建实验</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">101</span> <span class="n">experiment</span><span class="o">.</span><span class="n">create</span><span class="p">(</span><span class="n">name</span><span class="o">=</span><span class="s1">'pipe_neox_biases'</span><span class="p">,</span>
|
||||
<span class="lineno">102</span> <span class="n">writers</span><span class="o">=</span><span class="p">{</span><span class="s1">'screen'</span><span class="p">,</span> <span class="s1">'web_api'</span><span class="p">})</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-19'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-19'>#</a>
|
||||
</div>
|
||||
<p>初始化配置</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">105</span> <span class="n">conf</span> <span class="o">=</span> <span class="n">PipelineParallelTrainerConf</span><span class="p">()</span>
|
||||
<span class="lineno">106</span> <span class="n">experiment</span><span class="o">.</span><span class="n">configs</span><span class="p">(</span><span class="n">conf</span><span class="p">,</span> <span class="p">{</span>
|
||||
<span class="lineno">107</span> <span class="s1">'learning_rate'</span><span class="p">:</span> <span class="mf">3e-4</span><span class="p">,</span>
|
||||
<span class="lineno">108</span> <span class="s1">'is_checkpointing'</span><span class="p">:</span> <span class="kc">False</span><span class="p">,</span>
|
||||
<span class="lineno">109</span> <span class="s1">'max_seq_len'</span><span class="p">:</span> <span class="mi">128</span><span class="p">,</span>
|
||||
<span class="lineno">110</span> <span class="s1">'batch_size'</span><span class="p">:</span> <span class="mi">64</span><span class="p">,</span>
|
||||
<span class="lineno">111</span> <span class="s1">'chunks'</span><span class="p">:</span> <span class="mi">8</span><span class="p">,</span>
|
||||
<span class="lineno">112</span> <span class="p">})</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-20'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-20'>#</a>
|
||||
</div>
|
||||
<p>开始实验</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">115</span> <span class="k">with</span> <span class="n">experiment</span><span class="o">.</span><span class="n">start</span><span class="p">():</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-21'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-21'>#</a>
|
||||
</div>
|
||||
<p>初始化模型。在循环之前执行此操作以获得更清晰的日志。</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">117</span> <span class="n">_</span> <span class="o">=</span> <span class="n">conf</span><span class="o">.</span><span class="n">model</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-22'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-22'>#</a>
|
||||
</div>
|
||||
<p>火车</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">120</span> <span class="k">for</span> <span class="n">epoch</span> <span class="ow">in</span> <span class="n">monit</span><span class="o">.</span><span class="n">loop</span><span class="p">(</span><span class="n">conf</span><span class="o">.</span><span class="n">epochs</span><span class="p">):</span>
|
||||
<span class="lineno">121</span> <span class="n">conf</span><span class="o">.</span><span class="n">train_epoch</span><span class="p">()</span>
|
||||
<span class="lineno">122</span> <span class="n">tracker</span><span class="o">.</span><span class="n">new_line</span><span class="p">()</span>
|
||||
<span class="lineno">123</span> <span class="n">torch</span><span class="o">.</span><span class="n">save</span><span class="p">(</span><span class="n">conf</span><span class="o">.</span><span class="n">fine_tuner</span><span class="o">.</span><span class="n">state_dict</span><span class="p">(),</span> <span class="nb">str</span><span class="p">(</span><span class="n">lab</span><span class="o">.</span><span class="n">get_data_path</span><span class="p">()</span> <span class="o">/</span> <span class="s1">'fine_tune.pt'</span><span class="p">))</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-23'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-23'>#</a>
|
||||
</div>
|
||||
<p></p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">127</span><span class="k">if</span> <span class="vm">__name__</span> <span class="o">==</span> <span class="s1">'__main__'</span><span class="p">:</span>
|
||||
<span class="lineno">128</span> <span class="n">main</span><span class="p">()</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='footer'>
|
||||
<a href="https://labml.ai">labml.ai</a>
|
||||
</div>
|
||||
</div>
|
||||
<script src=../../interactive.js?v=1"></script>
|
||||
<script>
|
||||
function handleImages() {
|
||||
var images = document.querySelectorAll('p>img')
|
||||
|
||||
for (var i = 0; i < images.length; ++i) {
|
||||
handleImage(images[i])
|
||||
}
|
||||
}
|
||||
|
||||
function handleImage(img) {
|
||||
img.parentElement.style.textAlign = 'center'
|
||||
|
||||
var modal = document.createElement('div')
|
||||
modal.id = 'modal'
|
||||
|
||||
var modalContent = document.createElement('div')
|
||||
modal.appendChild(modalContent)
|
||||
|
||||
var modalImage = document.createElement('img')
|
||||
modalContent.appendChild(modalImage)
|
||||
|
||||
var span = document.createElement('span')
|
||||
span.classList.add('close')
|
||||
span.textContent = 'x'
|
||||
modal.appendChild(span)
|
||||
|
||||
img.onclick = function () {
|
||||
console.log('clicked')
|
||||
document.body.appendChild(modal)
|
||||
modalImage.src = img.src
|
||||
}
|
||||
|
||||
span.onclick = function () {
|
||||
document.body.removeChild(modal)
|
||||
}
|
||||
}
|
||||
|
||||
handleImages()
|
||||
</script>
|
||||
</body>
|
||||
</html>
|
||||
@@ -0,0 +1,405 @@
|
||||
<!DOCTYPE html>
|
||||
<html lang="zh">
|
||||
<head>
|
||||
<meta http-equiv="content-type" content="text/html;charset=utf-8"/>
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1.0"/>
|
||||
<meta name="description" content="使用 GPT-NEOX 生成文本"/>
|
||||
|
||||
<meta name="twitter:card" content="summary"/>
|
||||
<meta name="twitter:image:src" content="https://avatars1.githubusercontent.com/u/64068543?s=400&v=4"/>
|
||||
<meta name="twitter:title" content="使用 GPT-NEOX 生成文本"/>
|
||||
<meta name="twitter:description" content="使用 GPT-NEOX 生成文本"/>
|
||||
<meta name="twitter:site" content="@labmlai"/>
|
||||
<meta name="twitter:creator" content="@labmlai"/>
|
||||
|
||||
<meta property="og:url" content="https://nn.labml.ai/neox/samples/generate.html"/>
|
||||
<meta property="og:title" content="使用 GPT-NEOX 生成文本"/>
|
||||
<meta property="og:image" content="https://avatars1.githubusercontent.com/u/64068543?s=400&v=4"/>
|
||||
<meta property="og:site_name" content="使用 GPT-NEOX 生成文本"/>
|
||||
<meta property="og:type" content="object"/>
|
||||
<meta property="og:title" content="使用 GPT-NEOX 生成文本"/>
|
||||
<meta property="og:description" content="使用 GPT-NEOX 生成文本"/>
|
||||
|
||||
<title>使用 GPT-NEOX 生成文本</title>
|
||||
<link rel="shortcut icon" href="/icon.png"/>
|
||||
<link rel="stylesheet" href="../../pylit.css?v=1">
|
||||
<link rel="canonical" href="https://nn.labml.ai/neox/samples/generate.html"/>
|
||||
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.13.18/dist/katex.min.css" integrity="sha384-zTROYFVGOfTw7JV7KUu8udsvW2fx4lWOsCEDqhBreBwlHI4ioVRtmIvEThzJHGET" crossorigin="anonymous">
|
||||
|
||||
<!-- Global site tag (gtag.js) - Google Analytics -->
|
||||
<script async src="https://www.googletagmanager.com/gtag/js?id=G-4V3HC8HBLH"></script>
|
||||
<script>
|
||||
window.dataLayer = window.dataLayer || [];
|
||||
|
||||
function gtag() {
|
||||
dataLayer.push(arguments);
|
||||
}
|
||||
|
||||
gtag('js', new Date());
|
||||
|
||||
gtag('config', 'G-4V3HC8HBLH');
|
||||
</script>
|
||||
</head>
|
||||
<body>
|
||||
<div id='container'>
|
||||
<div id="background"></div>
|
||||
<div class='section'>
|
||||
<div class='docs'>
|
||||
<p>
|
||||
<a class="parent" href="/">home</a>
|
||||
<a class="parent" href="../index.html">neox</a>
|
||||
<a class="parent" href="index.html">samples</a>
|
||||
</p>
|
||||
<p>
|
||||
<a href="https://github.com/labmlai/annotated_deep_learning_paper_implementations" target="_blank">
|
||||
<img alt="Github"
|
||||
src="https://img.shields.io/github/stars/labmlai/annotated_deep_learning_paper_implementations?style=social"
|
||||
style="max-width:100%;"/></a>
|
||||
<a href="https://twitter.com/labmlai" rel="nofollow" target="_blank">
|
||||
<img alt="Twitter"
|
||||
src="https://img.shields.io/twitter/follow/labmlai?style=social"
|
||||
style="max-width:100%;"/></a>
|
||||
</p>
|
||||
<p>
|
||||
<a href="https://github.com/labmlai/annotated_deep_learning_paper_implementations/tree/master/labml_nn/neox/samples/generate.py" target="_blank">
|
||||
View code on Github</a>
|
||||
</p>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-0'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-0'>#</a>
|
||||
</div>
|
||||
<h1>使用 GPT-NEOX 生成文本</h1>
|
||||
<p>这说明了如何使用单个 GPU 从 GPT-NEOX 生成文本。</p>
|
||||
<p>这需要一个内存超过45GB的GPU。</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">15</span><span></span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-1'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-1'>#</a>
|
||||
</div>
|
||||
<p>进口</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">16</span><span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">List</span>
|
||||
<span class="lineno">17</span>
|
||||
<span class="lineno">18</span><span class="kn">import</span> <span class="nn">torch</span>
|
||||
<span class="lineno">19</span><span class="kn">from</span> <span class="nn">torch</span> <span class="kn">import</span> <span class="n">nn</span>
|
||||
<span class="lineno">20</span>
|
||||
<span class="lineno">21</span><span class="kn">from</span> <span class="nn">labml</span> <span class="kn">import</span> <span class="n">monit</span>
|
||||
<span class="lineno">22</span><span class="kn">from</span> <span class="nn">labml_nn.neox.model</span> <span class="kn">import</span> <span class="n">LayerGenerator</span>
|
||||
<span class="lineno">23</span><span class="kn">from</span> <span class="nn">labml_nn.neox.utils</span> <span class="kn">import</span> <span class="n">get_tokens</span><span class="p">,</span> <span class="n">print_tokens</span>
|
||||
<span class="lineno">24</span><span class="kn">from</span> <span class="nn">labml_nn.neox.utils.cache</span> <span class="kn">import</span> <span class="n">get_cache</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-2'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-2'>#</a>
|
||||
</div>
|
||||
<p>要加载的图层列表。这用于测试。您可以将层的子集分配给变压器层，<code class="highlight"><span></span><span class="p">{</span><span class="mi">0</span><span class="p">,</span> <span class="mi">1</span><span class="p">}</span></code>
|
||||
使其仅将第一个层加载到变压器层。</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">29</span><span class="n">LAYERS</span> <span class="o">=</span> <span class="kc">None</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-3'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-3'>#</a>
|
||||
</div>
|
||||
<p>提示完成</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">32</span><span class="n">PROMPT</span> <span class="o">=</span> <span class="s1">'Einstein was born in the German Empire, but moved to Switzerland in 1895, forsaking his German'</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-4'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-4'>#</a>
|
||||
</div>
|
||||
<h3>预测下一个代币</h3>
|
||||
<ul><li><code class="highlight"><span></span><span class="n">model</span></code>
|
||||
是模特吗</li>
|
||||
<li><code class="highlight"><span></span><span class="n">ids</span></code>
|
||||
是输入令牌 ID</li>
|
||||
<li><code class="highlight"><span></span><span class="n">device</span></code>
|
||||
是该型号的设备</li></ul>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">35</span><span class="k">def</span> <span class="nf">infer</span><span class="p">(</span><span class="n">model</span><span class="p">:</span> <span class="n">nn</span><span class="o">.</span><span class="n">Module</span><span class="p">,</span> <span class="n">ids</span><span class="p">:</span> <span class="n">List</span><span class="p">[</span><span class="nb">int</span><span class="p">],</span> <span class="n">device</span><span class="p">:</span> <span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">):</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-5'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-5'>#</a>
|
||||
</div>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">44</span> <span class="k">with</span> <span class="n">torch</span><span class="o">.</span><span class="n">no_grad</span><span class="p">():</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-6'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-6'>#</a>
|
||||
</div>
|
||||
<p>获取代币</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">46</span> <span class="n">x</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">tensor</span><span class="p">(</span><span class="n">ids</span><span class="p">)[</span><span class="kc">None</span><span class="p">,</span> <span class="p">:]</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="n">device</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-7'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-7'>#</a>
|
||||
</div>
|
||||
<p>评估模型</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">48</span> <span class="n">x</span> <span class="o">=</span> <span class="n">model</span><span class="p">(</span><span class="n">x</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-8'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-8'>#</a>
|
||||
</div>
|
||||
<p>返回预测的代币</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">51</span> <span class="k">return</span> <span class="n">x</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span><span class="o">.</span><span class="n">max</span><span class="p">(</span><span class="n">dim</span><span class="o">=-</span><span class="mi">1</span><span class="p">)[</span><span class="mi">1</span><span class="p">]</span><span class="o">.</span><span class="n">tolist</span><span class="p">()</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-9'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-9'>#</a>
|
||||
</div>
|
||||
<h2>生成文本</h2>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">54</span><span class="k">def</span> <span class="nf">generate</span><span class="p">():</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-10'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-10'>#</a>
|
||||
</div>
|
||||
<p>设置<a href="../utils/cache.html">缓存</a>以缓存中间键/值对以加快生成速度</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">60</span> <span class="n">cache</span> <span class="o">=</span> <span class="n">get_cache</span><span class="p">()</span>
|
||||
<span class="lineno">61</span> <span class="n">cache</span><span class="o">.</span><span class="n">set</span><span class="p">(</span><span class="s1">'use_cache'</span><span class="p">,</span> <span class="kc">True</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-11'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-11'>#</a>
|
||||
</div>
|
||||
<p>设备</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">64</span> <span class="n">device</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">'cuda:0'</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-12'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-12'>#</a>
|
||||
</div>
|
||||
<p>加载图层</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">67</span> <span class="n">layers</span> <span class="o">=</span> <span class="nb">list</span><span class="p">(</span><span class="n">LayerGenerator</span><span class="p">(</span><span class="n">is_clone_layers</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
|
||||
<span class="lineno">68</span> <span class="n">filter_layers</span><span class="o">=</span><span class="n">LAYERS</span><span class="p">,</span>
|
||||
<span class="lineno">69</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">float16</span><span class="p">,</span>
|
||||
<span class="lineno">70</span> <span class="n">device</span><span class="o">=</span><span class="n">device</span><span class="p">,</span>
|
||||
<span class="lineno">71</span> <span class="p">)</span><span class="o">.</span><span class="n">load</span><span class="p">())</span>
|
||||
<span class="lineno">72</span>
|
||||
<span class="lineno">73</span> <span class="n">model</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Sequential</span><span class="p">(</span><span class="o">*</span><span class="n">layers</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-13'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-13'>#</a>
|
||||
</div>
|
||||
<p>获取代币 ID</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">76</span> <span class="n">ids</span> <span class="o">=</span> <span class="n">get_tokens</span><span class="p">(</span><span class="n">PROMPT</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-14'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-14'>#</a>
|
||||
</div>
|
||||
<p>运行模型</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">79</span> <span class="n">cache</span><span class="o">.</span><span class="n">set</span><span class="p">(</span><span class="s1">'state_ids'</span><span class="p">,</span> <span class="p">(</span><span class="kc">None</span><span class="p">,</span> <span class="mi">1</span><span class="p">))</span>
|
||||
<span class="lineno">80</span> <span class="k">with</span> <span class="n">monit</span><span class="o">.</span><span class="n">section</span><span class="p">(</span><span class="s1">'Infer'</span><span class="p">):</span>
|
||||
<span class="lineno">81</span> <span class="n">next_token</span> <span class="o">=</span> <span class="n">infer</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">ids</span><span class="p">,</span> <span class="n">device</span><span class="p">)[</span><span class="o">-</span><span class="mi">1</span><span class="p">]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-15'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-15'>#</a>
|
||||
</div>
|
||||
<p>追加预测的令牌</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">84</span> <span class="n">ids</span> <span class="o">+=</span> <span class="p">[</span><span class="n">next_token</span><span class="p">]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-16'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-16'>#</a>
|
||||
</div>
|
||||
<p>预测 100 个代币</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">87</span> <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">100</span><span class="p">):</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-17'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-17'>#</a>
|
||||
</div>
|
||||
<p>设置状态以使用缓存的激活</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">89</span> <span class="n">cache</span><span class="o">.</span><span class="n">set</span><span class="p">(</span><span class="s1">'state_ids'</span><span class="p">,</span> <span class="p">(</span><span class="n">i</span><span class="p">,</span> <span class="n">i</span> <span class="o">+</span> <span class="mi">1</span><span class="p">))</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-18'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-18'>#</a>
|
||||
</div>
|
||||
<p>获取下一个令牌。请注意，我们只将最后一个令牌提供给模型，因为我们缓存了先前令牌的键/值对。</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">92</span> <span class="k">with</span> <span class="n">monit</span><span class="o">.</span><span class="n">section</span><span class="p">(</span><span class="s1">'Infer'</span><span class="p">):</span>
|
||||
<span class="lineno">93</span> <span class="n">next_token</span> <span class="o">=</span> <span class="n">infer</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="p">[</span><span class="n">next_token</span><span class="p">],</span> <span class="n">device</span><span class="p">)[</span><span class="o">-</span><span class="mi">1</span><span class="p">]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-19'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-19'>#</a>
|
||||
</div>
|
||||
<p>追加预测的令牌</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">95</span> <span class="n">ids</span> <span class="o">+=</span> <span class="p">[</span><span class="n">next_token</span><span class="p">]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-20'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-20'>#</a>
|
||||
</div>
|
||||
<p>打印</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">97</span> <span class="n">print_tokens</span><span class="p">(</span><span class="n">ids</span><span class="p">,</span> <span class="p">[</span><span class="n">ids</span><span class="p">])</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-21'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-21'>#</a>
|
||||
</div>
|
||||
<p></p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">101</span><span class="k">if</span> <span class="vm">__name__</span> <span class="o">==</span> <span class="s1">'__main__'</span><span class="p">:</span>
|
||||
<span class="lineno">102</span> <span class="n">generate</span><span class="p">()</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='footer'>
|
||||
<a href="https://labml.ai">labml.ai</a>
|
||||
</div>
|
||||
</div>
|
||||
<script src=../../interactive.js?v=1"></script>
|
||||
<script>
|
||||
function handleImages() {
|
||||
var images = document.querySelectorAll('p>img')
|
||||
|
||||
for (var i = 0; i < images.length; ++i) {
|
||||
handleImage(images[i])
|
||||
}
|
||||
}
|
||||
|
||||
function handleImage(img) {
|
||||
img.parentElement.style.textAlign = 'center'
|
||||
|
||||
var modal = document.createElement('div')
|
||||
modal.id = 'modal'
|
||||
|
||||
var modalContent = document.createElement('div')
|
||||
modal.appendChild(modalContent)
|
||||
|
||||
var modalImage = document.createElement('img')
|
||||
modalContent.appendChild(modalImage)
|
||||
|
||||
var span = document.createElement('span')
|
||||
span.classList.add('close')
|
||||
span.textContent = 'x'
|
||||
modal.appendChild(span)
|
||||
|
||||
img.onclick = function () {
|
||||
console.log('clicked')
|
||||
document.body.appendChild(modal)
|
||||
modalImage.src = img.src
|
||||
}
|
||||
|
||||
span.onclick = function () {
|
||||
document.body.removeChild(modal)
|
||||
}
|
||||
}
|
||||
|
||||
handleImages()
|
||||
</script>
|
||||
</body>
|
||||
</html>
|
||||
@@ -0,0 +1,128 @@
|
||||
<!DOCTYPE html>
|
||||
<html lang="zh">
|
||||
<head>
|
||||
<meta http-equiv="content-type" content="text/html;charset=utf-8"/>
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1.0"/>
|
||||
<meta name="description" content="用于推理和微调的样本"/>
|
||||
|
||||
<meta name="twitter:card" content="summary"/>
|
||||
<meta name="twitter:image:src" content="https://avatars1.githubusercontent.com/u/64068543?s=400&v=4"/>
|
||||
<meta name="twitter:title" content="样品"/>
|
||||
<meta name="twitter:description" content="用于推理和微调的样本"/>
|
||||
<meta name="twitter:site" content="@labmlai"/>
|
||||
<meta name="twitter:creator" content="@labmlai"/>
|
||||
|
||||
<meta property="og:url" content="https://nn.labml.ai/neox/samples/index.html"/>
|
||||
<meta property="og:title" content="样品"/>
|
||||
<meta property="og:image" content="https://avatars1.githubusercontent.com/u/64068543?s=400&v=4"/>
|
||||
<meta property="og:site_name" content="样品"/>
|
||||
<meta property="og:type" content="object"/>
|
||||
<meta property="og:title" content="样品"/>
|
||||
<meta property="og:description" content="用于推理和微调的样本"/>
|
||||
|
||||
<title>样品</title>
|
||||
<link rel="shortcut icon" href="/icon.png"/>
|
||||
<link rel="stylesheet" href="../../pylit.css?v=1">
|
||||
<link rel="canonical" href="https://nn.labml.ai/neox/samples/index.html"/>
|
||||
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.13.18/dist/katex.min.css" integrity="sha384-zTROYFVGOfTw7JV7KUu8udsvW2fx4lWOsCEDqhBreBwlHI4ioVRtmIvEThzJHGET" crossorigin="anonymous">
|
||||
|
||||
<!-- Global site tag (gtag.js) - Google Analytics -->
|
||||
<script async src="https://www.googletagmanager.com/gtag/js?id=G-4V3HC8HBLH"></script>
|
||||
<script>
|
||||
window.dataLayer = window.dataLayer || [];
|
||||
|
||||
function gtag() {
|
||||
dataLayer.push(arguments);
|
||||
}
|
||||
|
||||
gtag('js', new Date());
|
||||
|
||||
gtag('config', 'G-4V3HC8HBLH');
|
||||
</script>
|
||||
</head>
|
||||
<body>
|
||||
<div id='container'>
|
||||
<div id="background"></div>
|
||||
<div class='section'>
|
||||
<div class='docs'>
|
||||
<p>
|
||||
<a class="parent" href="/">home</a>
|
||||
<a class="parent" href="../index.html">neox</a>
|
||||
<a class="parent" href="index.html">samples</a>
|
||||
</p>
|
||||
<p>
|
||||
<a href="https://github.com/labmlai/annotated_deep_learning_paper_implementations" target="_blank">
|
||||
<img alt="Github"
|
||||
src="https://img.shields.io/github/stars/labmlai/annotated_deep_learning_paper_implementations?style=social"
|
||||
style="max-width:100%;"/></a>
|
||||
<a href="https://twitter.com/labmlai" rel="nofollow" target="_blank">
|
||||
<img alt="Twitter"
|
||||
src="https://img.shields.io/twitter/follow/labmlai?style=social"
|
||||
style="max-width:100%;"/></a>
|
||||
</p>
|
||||
<p>
|
||||
<a href="https://github.com/labmlai/annotated_deep_learning_paper_implementations/tree/master/labml_nn/neox/samples/__init__.py" target="_blank">
|
||||
View code on Github</a>
|
||||
</p>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-0'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-0'>#</a>
|
||||
</div>
|
||||
<h1>样品</h1>
|
||||
<ul><li><a href="generate.html">生成文本</a></li>
|
||||
</ul><li><a href="finetune.html">通过管道平行训练微调偏差</a></li>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='footer'>
|
||||
<a href="https://labml.ai">labml.ai</a>
|
||||
</div>
|
||||
</div>
|
||||
<script src=../../interactive.js?v=1"></script>
|
||||
<script>
|
||||
function handleImages() {
|
||||
var images = document.querySelectorAll('p>img')
|
||||
|
||||
for (var i = 0; i < images.length; ++i) {
|
||||
handleImage(images[i])
|
||||
}
|
||||
}
|
||||
|
||||
function handleImage(img) {
|
||||
img.parentElement.style.textAlign = 'center'
|
||||
|
||||
var modal = document.createElement('div')
|
||||
modal.id = 'modal'
|
||||
|
||||
var modalContent = document.createElement('div')
|
||||
modal.appendChild(modalContent)
|
||||
|
||||
var modalImage = document.createElement('img')
|
||||
modalContent.appendChild(modalImage)
|
||||
|
||||
var span = document.createElement('span')
|
||||
span.classList.add('close')
|
||||
span.textContent = 'x'
|
||||
modal.appendChild(span)
|
||||
|
||||
img.onclick = function () {
|
||||
console.log('clicked')
|
||||
document.body.appendChild(modal)
|
||||
modalImage.src = img.src
|
||||
}
|
||||
|
||||
span.onclick = function () {
|
||||
document.body.removeChild(modal)
|
||||
}
|
||||
}
|
||||
|
||||
handleImages()
|
||||
</script>
|
||||
</body>
|
||||
</html>
|
||||
@@ -0,0 +1,347 @@
|
||||
<!DOCTYPE html>
|
||||
<html lang="zh">
|
||||
<head>
|
||||
<meta http-equiv="content-type" content="text/html;charset=utf-8"/>
|
||||
<meta name="viewport" content="width=device-width, initial-scale=1.0"/>
|
||||
<meta name="description" content="使用 llm.int8 () 量化使用 GPT-NEOX 生成文本"/>
|
||||
|
||||
<meta name="twitter:card" content="summary"/>
|
||||
<meta name="twitter:image:src" content="https://avatars1.githubusercontent.com/u/64068543?s=400&v=4"/>
|
||||
<meta name="twitter:title" content="使用 llm.int8 () 量化使用 GPT-NEOX 生成文本"/>
|
||||
<meta name="twitter:description" content="使用 llm.int8 () 量化使用 GPT-NEOX 生成文本"/>
|
||||
<meta name="twitter:site" content="@labmlai"/>
|
||||
<meta name="twitter:creator" content="@labmlai"/>
|
||||
|
||||
<meta property="og:url" content="https://nn.labml.ai/neox/samples/llm_int8.html"/>
|
||||
<meta property="og:title" content="使用 llm.int8 () 量化使用 GPT-NEOX 生成文本"/>
|
||||
<meta property="og:image" content="https://avatars1.githubusercontent.com/u/64068543?s=400&v=4"/>
|
||||
<meta property="og:site_name" content="使用 llm.int8 () 量化使用 GPT-NEOX 生成文本"/>
|
||||
<meta property="og:type" content="object"/>
|
||||
<meta property="og:title" content="使用 llm.int8 () 量化使用 GPT-NEOX 生成文本"/>
|
||||
<meta property="og:description" content="使用 llm.int8 () 量化使用 GPT-NEOX 生成文本"/>
|
||||
|
||||
<title>使用 llm.int8 () 量化使用 GPT-NEOX 生成文本</title>
|
||||
<link rel="shortcut icon" href="/icon.png"/>
|
||||
<link rel="stylesheet" href="../../pylit.css?v=1">
|
||||
<link rel="canonical" href="https://nn.labml.ai/neox/samples/llm_int8.html"/>
|
||||
<link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/katex@0.13.18/dist/katex.min.css" integrity="sha384-zTROYFVGOfTw7JV7KUu8udsvW2fx4lWOsCEDqhBreBwlHI4ioVRtmIvEThzJHGET" crossorigin="anonymous">
|
||||
|
||||
<!-- Global site tag (gtag.js) - Google Analytics -->
|
||||
<script async src="https://www.googletagmanager.com/gtag/js?id=G-4V3HC8HBLH"></script>
|
||||
<script>
|
||||
window.dataLayer = window.dataLayer || [];
|
||||
|
||||
function gtag() {
|
||||
dataLayer.push(arguments);
|
||||
}
|
||||
|
||||
gtag('js', new Date());
|
||||
|
||||
gtag('config', 'G-4V3HC8HBLH');
|
||||
</script>
|
||||
</head>
|
||||
<body>
|
||||
<div id='container'>
|
||||
<div id="background"></div>
|
||||
<div class='section'>
|
||||
<div class='docs'>
|
||||
<p>
|
||||
<a class="parent" href="/">home</a>
|
||||
<a class="parent" href="../index.html">neox</a>
|
||||
<a class="parent" href="index.html">samples</a>
|
||||
</p>
|
||||
<p>
|
||||
<a href="https://github.com/labmlai/annotated_deep_learning_paper_implementations" target="_blank">
|
||||
<img alt="Github"
|
||||
src="https://img.shields.io/github/stars/labmlai/annotated_deep_learning_paper_implementations?style=social"
|
||||
style="max-width:100%;"/></a>
|
||||
<a href="https://twitter.com/labmlai" rel="nofollow" target="_blank">
|
||||
<img alt="Twitter"
|
||||
src="https://img.shields.io/twitter/follow/labmlai?style=social"
|
||||
style="max-width:100%;"/></a>
|
||||
</p>
|
||||
<p>
|
||||
<a href="https://github.com/labmlai/annotated_deep_learning_paper_implementations/tree/master/labml_nn/neox/samples/llm_int8.py" target="_blank">
|
||||
View code on Github</a>
|
||||
</p>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-0'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-0'>#</a>
|
||||
</div>
|
||||
<h1>使用 llm.int8 () 量化使用 GPT-NEOX 生成文本</h1>
|
||||
<p>这说明了如何使用 <a href="../utils/llm_int8.html">llm.int8 () 量化</a>从 GPT-NEOX 生成文本。</p>
|
||||
<p>这需要一个具有 24GB 内存的 GPU。</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">15</span><span></span><span class="kn">import</span> <span class="nn">torch</span>
|
||||
<span class="lineno">16</span><span class="kn">from</span> <span class="nn">torch</span> <span class="kn">import</span> <span class="n">nn</span>
|
||||
<span class="lineno">17</span>
|
||||
<span class="lineno">18</span><span class="kn">from</span> <span class="nn">labml</span> <span class="kn">import</span> <span class="n">monit</span>
|
||||
<span class="lineno">19</span><span class="kn">from</span> <span class="nn">labml_nn.neox.model</span> <span class="kn">import</span> <span class="n">LayerGenerator</span>
|
||||
<span class="lineno">20</span><span class="kn">from</span> <span class="nn">labml_nn.neox.samples.generate</span> <span class="kn">import</span> <span class="n">PROMPT</span><span class="p">,</span> <span class="n">infer</span>
|
||||
<span class="lineno">21</span><span class="kn">from</span> <span class="nn">labml_nn.neox.utils</span> <span class="kn">import</span> <span class="n">get_tokens</span><span class="p">,</span> <span class="n">print_tokens</span>
|
||||
<span class="lineno">22</span><span class="kn">from</span> <span class="nn">labml_nn.neox.utils.cache</span> <span class="kn">import</span> <span class="n">get_cache</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-1'>
|
||||
<div class='docs doc-strings'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-1'>#</a>
|
||||
</div>
|
||||
<h2>生成文本</h2>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">25</span><span class="k">def</span> <span class="nf">generate</span><span class="p">():</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-2'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-2'>#</a>
|
||||
</div>
|
||||
<p>设置<a href="../utils/cache.html">缓存</a>以缓存中间键/值对以加快生成速度</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">31</span> <span class="n">cache</span> <span class="o">=</span> <span class="n">get_cache</span><span class="p">()</span>
|
||||
<span class="lineno">32</span> <span class="n">cache</span><span class="o">.</span><span class="n">set</span><span class="p">(</span><span class="s1">'use_cache'</span><span class="p">,</span> <span class="kc">True</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-3'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-3'>#</a>
|
||||
</div>
|
||||
<p>设备</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">35</span> <span class="n">device</span> <span class="o">=</span> <span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">'cuda:0'</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-4'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-4'>#</a>
|
||||
</div>
|
||||
将 <p>float16 中的层加载到 CPU 中。我们稍后将图层转换为int8，因为在将图层加载到GPU后即时执行此操作会导致CUDA内存碎片（大约3GB的内存可能会由于碎片而丢失）。</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">40</span> <span class="n">layer_generator</span> <span class="o">=</span> <span class="n">LayerGenerator</span><span class="p">(</span><span class="n">is_clone_layers</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
|
||||
<span class="lineno">41</span> <span class="n">dtype</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">float16</span><span class="p">,</span>
|
||||
<span class="lineno">42</span> <span class="n">device</span><span class="o">=</span><span class="n">torch</span><span class="o">.</span><span class="n">device</span><span class="p">(</span><span class="s1">'cpu'</span><span class="p">),</span>
|
||||
<span class="lineno">43</span> <span class="n">is_llm_int8</span><span class="o">=</span><span class="kc">False</span><span class="p">,</span>
|
||||
<span class="lineno">44</span> <span class="p">)</span>
|
||||
<span class="lineno">45</span> <span class="n">layers</span> <span class="o">=</span> <span class="nb">list</span><span class="p">(</span><span class="n">layer_generator</span><span class="o">.</span><span class="n">load</span><span class="p">())</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-5'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-5'>#</a>
|
||||
</div>
|
||||
<p>这减少了 CUDA 内存碎片</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">48</span> <span class="k">for</span> <span class="n">layer</span> <span class="ow">in</span> <span class="n">monit</span><span class="o">.</span><span class="n">iterate</span><span class="p">(</span><span class="s1">'Convert to int8'</span><span class="p">,</span> <span class="n">layers</span><span class="p">,</span> <span class="n">is_children_silent</span><span class="o">=</span><span class="kc">True</span><span class="p">):</span>
|
||||
<span class="lineno">49</span> <span class="n">layer_generator</span><span class="o">.</span><span class="n">post_load_prepare</span><span class="p">(</span><span class="n">layer</span><span class="p">,</span>
|
||||
<span class="lineno">50</span> <span class="n">device</span><span class="o">=</span><span class="n">device</span><span class="p">,</span>
|
||||
<span class="lineno">51</span> <span class="n">is_llm_int8</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span>
|
||||
<span class="lineno">52</span> <span class="n">llm_int8_threshold</span><span class="o">=</span><span class="mf">6.0</span><span class="p">,</span>
|
||||
<span class="lineno">53</span> <span class="p">)</span>
|
||||
<span class="lineno">54</span> <span class="n">layer</span><span class="o">.</span><span class="n">to</span><span class="p">(</span><span class="n">device</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-6'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-6'>#</a>
|
||||
</div>
|
||||
<p>创建<code class="highlight"><span></span><span class="n">nn</span><span class="o">.</span><span class="n">Sequential</span></code>
|
||||
模型</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">57</span> <span class="n">model</span> <span class="o">=</span> <span class="n">nn</span><span class="o">.</span><span class="n">Sequential</span><span class="p">(</span><span class="o">*</span><span class="n">layers</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-7'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-7'>#</a>
|
||||
</div>
|
||||
<p>清除缓存和打印内存摘要以进行调试</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">60</span> <span class="n">torch</span><span class="o">.</span><span class="n">cuda</span><span class="o">.</span><span class="n">empty_cache</span><span class="p">()</span>
|
||||
<span class="lineno">61</span> <span class="nb">print</span><span class="p">(</span><span class="n">torch</span><span class="o">.</span><span class="n">cuda</span><span class="o">.</span><span class="n">memory_summary</span><span class="p">())</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-8'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-8'>#</a>
|
||||
</div>
|
||||
<p>获取代币 ID</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">64</span> <span class="n">ids</span> <span class="o">=</span> <span class="n">get_tokens</span><span class="p">(</span><span class="n">PROMPT</span><span class="p">)</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-9'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-9'>#</a>
|
||||
</div>
|
||||
<p>运行模型。我们使用中定义的<a href="generate.html"><code class="highlight"><span></span><span class="n">infer</span></code>
|
||||
</a>函数 <a href="generate.html"><code class="highlight"><span></span><span class="n">generate</span><span class="o">.</span><span class="n">py</span></code>
|
||||
</a></p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">68</span> <span class="n">cache</span><span class="o">.</span><span class="n">set</span><span class="p">(</span><span class="s1">'state_ids'</span><span class="p">,</span> <span class="p">(</span><span class="kc">None</span><span class="p">,</span> <span class="mi">1</span><span class="p">))</span>
|
||||
<span class="lineno">69</span> <span class="k">with</span> <span class="n">monit</span><span class="o">.</span><span class="n">section</span><span class="p">(</span><span class="s1">'Infer'</span><span class="p">):</span>
|
||||
<span class="lineno">70</span> <span class="n">next_token</span> <span class="o">=</span> <span class="n">infer</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="n">ids</span><span class="p">,</span> <span class="n">device</span><span class="p">)[</span><span class="o">-</span><span class="mi">1</span><span class="p">]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-10'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-10'>#</a>
|
||||
</div>
|
||||
<p>追加预测的令牌</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">73</span> <span class="n">ids</span> <span class="o">+=</span> <span class="p">[</span><span class="n">next_token</span><span class="p">]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-11'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-11'>#</a>
|
||||
</div>
|
||||
<p>预测 100 个代币</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">76</span> <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">100</span><span class="p">):</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-12'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-12'>#</a>
|
||||
</div>
|
||||
<p>设置状态以使用缓存的激活</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">78</span> <span class="n">cache</span><span class="o">.</span><span class="n">set</span><span class="p">(</span><span class="s1">'state_ids'</span><span class="p">,</span> <span class="p">(</span><span class="n">i</span><span class="p">,</span> <span class="n">i</span> <span class="o">+</span> <span class="mi">1</span><span class="p">))</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-13'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-13'>#</a>
|
||||
</div>
|
||||
<p>获取下一个令牌。请注意，我们只将最后一个令牌提供给模型，因为我们缓存了先前令牌的键/值对。</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">81</span> <span class="k">with</span> <span class="n">monit</span><span class="o">.</span><span class="n">section</span><span class="p">(</span><span class="s1">'Infer'</span><span class="p">):</span>
|
||||
<span class="lineno">82</span> <span class="n">next_token</span> <span class="o">=</span> <span class="n">infer</span><span class="p">(</span><span class="n">model</span><span class="p">,</span> <span class="p">[</span><span class="n">next_token</span><span class="p">],</span> <span class="n">device</span><span class="p">)[</span><span class="o">-</span><span class="mi">1</span><span class="p">]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-14'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-14'>#</a>
|
||||
</div>
|
||||
<p>追加预测的令牌</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">84</span> <span class="n">ids</span> <span class="o">+=</span> <span class="p">[</span><span class="n">next_token</span><span class="p">]</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-15'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-15'>#</a>
|
||||
</div>
|
||||
<p>打印</p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">86</span> <span class="n">print_tokens</span><span class="p">(</span><span class="n">ids</span><span class="p">,</span> <span class="p">[</span><span class="n">ids</span><span class="p">])</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='section' id='section-16'>
|
||||
<div class='docs'>
|
||||
<div class='section-link'>
|
||||
<a href='#section-16'>#</a>
|
||||
</div>
|
||||
<p></p>
|
||||
|
||||
</div>
|
||||
<div class='code'>
|
||||
<div class="highlight"><pre><span class="lineno">90</span><span class="k">if</span> <span class="vm">__name__</span> <span class="o">==</span> <span class="s1">'__main__'</span><span class="p">:</span>
|
||||
<span class="lineno">91</span> <span class="n">generate</span><span class="p">()</span></pre></div>
|
||||
</div>
|
||||
</div>
|
||||
<div class='footer'>
|
||||
<a href="https://labml.ai">labml.ai</a>
|
||||
</div>
|
||||
</div>
|
||||
<script src=../../interactive.js?v=1"></script>
|
||||
<script>
|
||||
function handleImages() {
|
||||
var images = document.querySelectorAll('p>img')
|
||||
|
||||
for (var i = 0; i < images.length; ++i) {
|
||||
handleImage(images[i])
|
||||
}
|
||||
}
|
||||
|
||||
function handleImage(img) {
|
||||
img.parentElement.style.textAlign = 'center'
|
||||
|
||||
var modal = document.createElement('div')
|
||||
modal.id = 'modal'
|
||||
|
||||
var modalContent = document.createElement('div')
|
||||
modal.appendChild(modalContent)
|
||||
|
||||
var modalImage = document.createElement('img')
|
||||
modalContent.appendChild(modalImage)
|
||||
|
||||
var span = document.createElement('span')
|
||||
span.classList.add('close')
|
||||
span.textContent = 'x'
|
||||
modal.appendChild(span)
|
||||
|
||||
img.onclick = function () {
|
||||
console.log('clicked')
|
||||
document.body.appendChild(modal)
|
||||
modalImage.src = img.src
|
||||
}
|
||||
|
||||
span.onclick = function () {
|
||||
document.body.removeChild(modal)
|
||||
}
|
||||
}
|
||||
|
||||
handleImages()
|
||||
</script>
|
||||
</body>
|
||||
</html>
|
||||
Reference in New Issue
Block a user