Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"Custom Model","local":"custom-model","sections":[{"title":"Creating a Custom Model","local":"creating-a-custom-model","sections":[],"depth":2},{"title":"Running the Evaluation","local":"running-the-evaluation","sections":[{"title":"Using the Command Line","local":"using-the-command-line","sections":[],"depth":3},{"title":"Using the Python API","local":"using-the-python-api","sections":[],"depth":3}],"depth":2},{"title":"Required Methods","local":"required-methods","sections":[],"depth":2},{"title":"Enabling Caching (Recommended)","local":"enabling-caching-recommended","sections":[],"depth":2},{"title":"Best Practices","local":"best-practices","sections":[],"depth":2}],"depth":1}"> | |
| <link href="/docs/lighteval/pr_476/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/entry/start.9e6d7572.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/chunks/scheduler.7da89386.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/chunks/singletons.21444ba4.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/chunks/paths.5bc62363.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/entry/app.75f264ac.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/chunks/index.20910acc.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/nodes/0.38ea3089.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/chunks/each.e59479a4.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/nodes/7.aa7465dd.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/chunks/Tip.53e22153.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/chunks/CodeBlock.143bd81e.js"> | |
| <link rel="modulepreload" href="/docs/lighteval/pr_476/en/_app/immutable/chunks/getInferenceSnippets.cb95a28d.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{"title":"Custom Model","local":"custom-model","sections":[{"title":"Creating a Custom Model","local":"creating-a-custom-model","sections":[],"depth":2},{"title":"Running the Evaluation","local":"running-the-evaluation","sections":[{"title":"Using the Command Line","local":"using-the-command-line","sections":[],"depth":3},{"title":"Using the Python API","local":"using-the-python-api","sections":[],"depth":3}],"depth":2},{"title":"Required Methods","local":"required-methods","sections":[],"depth":2},{"title":"Enabling Caching (Recommended)","local":"enabling-caching-recommended","sections":[],"depth":2},{"title":"Best Practices","local":"best-practices","sections":[],"depth":2}],"depth":1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <h1 class="relative group"><a id="custom-model" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#custom-model"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Custom Model</span></h1> <p data-svelte-h="svelte-f5zt17">Lighteval allows you to evaluate custom model implementations by creating a custom model class that inherits from <code>LightevalModel</code>. | |
| This is useful when you want to evaluate models that aren’t directly supported by the standard backends and providers (transformers, vllm, etc), or | |
| if you want to add your own pre/post processing.</p> <h2 class="relative group"><a id="creating-a-custom-model" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#creating-a-custom-model"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Creating a Custom Model</span></h2> <ol data-svelte-h="svelte-5nepfb"><li>Create a Python file containing your custom model implementation. The model must inherit from <code>LightevalModel</code> and implement all required methods.</li></ol> <p data-svelte-h="svelte-1frc3d">Here’s a basic example:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START --><span class="hljs-keyword">from</span> typing <span class="hljs-keyword">import</span> <span class="hljs-type">List</span> | |
| <span class="hljs-keyword">from</span> lighteval.models.abstract_model <span class="hljs-keyword">import</span> LightevalModel | |
| <span class="hljs-keyword">from</span> lighteval.models.model_output <span class="hljs-keyword">import</span> ModelResponse | |
| <span class="hljs-keyword">from</span> lighteval.tasks.requests <span class="hljs-keyword">import</span> Doc | |
| <span class="hljs-keyword">from</span> lighteval.utils.cache_management <span class="hljs-keyword">import</span> SampleCache, cached | |
| <span class="hljs-keyword">class</span> <span class="hljs-title class_">MyCustomModel</span>(<span class="hljs-title class_ inherited__">LightevalModel</span>): | |
| <span class="hljs-keyword">def</span> <span class="hljs-title function_">__init__</span>(<span class="hljs-params">self, config</span>): | |
| <span class="hljs-built_in">super</span>().__init__(config) | |
| <span class="hljs-comment"># Initialize your model here...</span> | |
| <span class="hljs-comment"># Enable caching (recommended)</span> | |
| self._cache = SampleCache(config) | |
| <span class="hljs-meta"> @cached(<span class="hljs-params"><span class="hljs-string">"predictions"</span></span>) </span><span class="hljs-comment"># Enable caching for better performance</span> | |
| <span class="hljs-keyword">def</span> <span class="hljs-title function_">greedy_until</span>(<span class="hljs-params">self, docs: <span class="hljs-type">List</span>[Doc]</span>) -> <span class="hljs-type">List</span>[ModelResponse]: | |
| <span class="hljs-comment"># Implement generation logic</span> | |
| <span class="hljs-keyword">pass</span> | |
| <span class="hljs-meta"> @cached(<span class="hljs-params"><span class="hljs-string">"predictions"</span></span>) </span><span class="hljs-comment"># Enable caching for better performance</span> | |
| <span class="hljs-keyword">def</span> <span class="hljs-title function_">loglikelihood</span>(<span class="hljs-params">self, docs: <span class="hljs-type">List</span>[Doc]</span>) -> <span class="hljs-type">List</span>[ModelResponse]: | |
| <span class="hljs-comment"># Implement loglikelihood computation</span> | |
| <span class="hljs-keyword">pass</span> | |
| <span class="hljs-meta"> @cached(<span class="hljs-params"><span class="hljs-string">"predictions"</span></span>) </span><span class="hljs-comment"># Enable caching for better performance</span> | |
| <span class="hljs-keyword">def</span> <span class="hljs-title function_">loglikelihood_rolling</span>(<span class="hljs-params">self, docs: <span class="hljs-type">List</span>[Doc]</span>) -> <span class="hljs-type">List</span>[ModelResponse]: | |
| <span class="hljs-comment"># Implement rolling loglikelihood computation</span> | |
| <span class="hljs-keyword">pass</span><!-- HTML_TAG_END --></pre></div> <ol start="2" data-svelte-h="svelte-1uhsqbo"><li>The custom model file should contain exactly one class that inherits from <code>LightevalModel</code>. This class will be automatically detected and instantiated when loading the model.</li></ol> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-1s5ktfo">You can find a complete example of a custom model implementation in <code>examples/custom_models/google_translate_model.py</code>.</p></div> <h2 class="relative group"><a id="running-the-evaluation" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#running-the-evaluation"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Running the Evaluation</span></h2> <p data-svelte-h="svelte-1icw64v">You can evaluate your custom model using either the command line interface or the Python API.</p> <h3 class="relative group"><a id="using-the-command-line" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#using-the-command-line"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Using the Command Line</span></h3> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START -->lighteval custom \ | |
| <span class="hljs-string">"google-translate"</span> \ | |
| <span class="hljs-string">"examples/custom_models/google_translate_model.py"</span> \ | |
| <span class="hljs-string">"lighteval|wmt20:fr-de|0|0"</span> \ | |
| --max-samples 10<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-9ha93v">The command takes three required arguments:</p> <ul data-svelte-h="svelte-1dz6vvj"><li>The model name (used for tracking in results/logs)</li> <li>The path to your model implementation file</li> <li>The tasks to evaluate on (same format as other backends)</li></ul> <h3 class="relative group"><a id="using-the-python-api" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#using-the-python-api"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Using the Python API</span></h3> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START --><span class="hljs-keyword">from</span> lighteval.logging.evaluation_tracker <span class="hljs-keyword">import</span> EvaluationTracker | |
| <span class="hljs-keyword">from</span> lighteval.models.custom.custom_model <span class="hljs-keyword">import</span> CustomModelConfig | |
| <span class="hljs-keyword">from</span> lighteval.pipeline <span class="hljs-keyword">import</span> Pipeline, PipelineParameters | |
| <span class="hljs-comment"># Set up evaluation tracking</span> | |
| evaluation_tracker = EvaluationTracker( | |
| output_dir=<span class="hljs-string">"results"</span>, | |
| save_details=<span class="hljs-literal">True</span> | |
| ) | |
| <span class="hljs-comment"># Configure the pipeline</span> | |
| pipeline_params = PipelineParameters( | |
| launcher_type=ParallelismManager.CUSTOM, | |
| ) | |
| <span class="hljs-comment"># Configure your custom model</span> | |
| model_config = CustomModelConfig( | |
| model=<span class="hljs-string">"my-custom-model"</span>, | |
| model_definition_file_path=<span class="hljs-string">"path/to/my_model.py"</span> | |
| ) | |
| <span class="hljs-comment"># Create and run the pipeline</span> | |
| pipeline = Pipeline( | |
| tasks=<span class="hljs-string">"leaderboard|truthfulqa:mc|0|0"</span>, | |
| pipeline_parameters=pipeline_params, | |
| evaluation_tracker=evaluation_tracker, | |
| model_config=model_config | |
| ) | |
| pipeline.evaluate() | |
| pipeline.save_and_push_results()<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="required-methods" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#required-methods"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Required Methods</span></h2> <p data-svelte-h="svelte-abp4ax">Your custom model must implement these core methods:</p> <ul data-svelte-h="svelte-yt56c1"><li><code>greedy_until</code>: For generating text until a stop sequence or max tokens is reached - this is used for generative evaluations</li> <li><code>loglikelihood</code>: For computing log probabilities of specific continuations - this is used for multiple choice logprob evaluations</li> <li><code>loglikelihood_rolling</code>: For computing rolling log probabilities of sequences - this is used for perplexity metrics</li></ul> <p data-svelte-h="svelte-845vib">See the <code>LightevalModel</code> base class documentation for detailed method signatures and requirements.</p> <h2 class="relative group"><a id="enabling-caching-recommended" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#enabling-caching-recommended"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Enabling Caching (Recommended)</span></h2> <p data-svelte-h="svelte-1eshig5">Lighteval includes a caching system that can significantly speed up evaluations by storing and reusing model predictions. | |
| To enable caching in your custom model:</p> <ol><li><p data-svelte-h="svelte-sl2427"><strong>Import caching components</strong>:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START --><span class="hljs-keyword">from</span> lighteval.utils.cache_management <span class="hljs-keyword">import</span> SampleCache, cached<!-- HTML_TAG_END --></pre></div></li> <li><p data-svelte-h="svelte-1nvuilw"><strong>Initialize cache in constructor</strong>:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START --><span class="hljs-keyword">def</span> <span class="hljs-title function_">__init__</span>(<span class="hljs-params">self, config</span>): | |
| <span class="hljs-comment"># Your initialization code...</span> | |
| self._cache = SampleCache(config)<!-- HTML_TAG_END --></pre></div></li> <li><p data-svelte-h="svelte-m52y4l"><strong>Add cache decorators</strong> to your prediction methods:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START --><span class="hljs-meta">@cached(<span class="hljs-params"><span class="hljs-string">"predictions"</span></span>)</span> | |
| <span class="hljs-keyword">def</span> <span class="hljs-title function_">greedy_until</span>(<span class="hljs-params">self, docs: <span class="hljs-type">List</span>[Doc]</span>) -> <span class="hljs-type">List</span>[ModelResponse]: | |
| <span class="hljs-comment"># Your implementation...</span><!-- HTML_TAG_END --></pre></div></li></ol> <p data-svelte-h="svelte-t0b8b6">For detailed information about the caching system, see the <a href="./caching.mdx">Caching Documentation</a>.</p> <h2 class="relative group"><a id="best-practices" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#best-practices"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>Best Practices</span></h2> <ol data-svelte-h="svelte-1auebwz"><li><p><strong>Error Handling</strong>: Implement robust error handling in your model methods to gracefully handle edge cases.</p></li> <li><p><strong>Batching</strong>: Consider implementing efficient batching in your model methods to improve performance.</p></li> <li><p><strong>Documentation</strong>: Add clear docstrings to your model class and methods explaining any specific requirements or limitations.</p></li> <li><p><strong>Caching</strong>: Enable caching to speed up repeated evaluations and development iterations.</p></li></ol> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/lighteval/blob/main/docs/source/evaluating-a-custom-model.mdx" target="_blank"><span data-svelte-h="svelte-1kd6by1"><</span> <span data-svelte-h="svelte-x0xyl0">></span> <span data-svelte-h="svelte-1dajgef"><span class="underline ml-1.5">Update</span> on GitHub</span></a> <p></p> | |
| <script> | |
| { | |
| __sveltekit_xuqd6b = { | |
| assets: "/docs/lighteval/pr_476/en", | |
| base: "/docs/lighteval/pr_476/en", | |
| env: {} | |
| }; | |
| const element = document.currentScript.parentElement; | |
| const data = [null,null]; | |
| Promise.all([ | |
| import("/docs/lighteval/pr_476/en/_app/immutable/entry/start.9e6d7572.js"), | |
| import("/docs/lighteval/pr_476/en/_app/immutable/entry/app.75f264ac.js") | |
| ]).then(([kit, app]) => { | |
| kit.start(app, element, { | |
| node_ids: [0, 7], | |
| data, | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 29.9 kB
- Xet hash:
- 875bf743c6d949d590f18ca64cf276add49eb4d6e644446827ca65e0b81d993f
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.