Built site for gh-pages
This commit is contained in:
@@ -820,8 +820,35 @@ gtag('config', 'G-9KYCVJBNMQ', { 'anonymize_ip': true});
|
||||
<h3 class="anchored" data-anchor-id="axolotl.core.builders.rl.HFRLTrainerBuilder">HFRLTrainerBuilder</h3>
|
||||
<div class="code-copy-outer-scaffold"><div class="sourceCode" id="cb1"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb1-1"><a href="#cb1-1" aria-hidden="true" tabindex="-1"></a>core.builders.rl.HFRLTrainerBuilder(cfg, model, tokenizer, processor<span class="op">=</span><span class="va">None</span>)</span></code></pre></div><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></div>
|
||||
<p>Trainer factory class for TRL-based RLHF trainers (e.g. DPO)</p>
|
||||
<section id="methods" class="level4">
|
||||
<h4 class="anchored" data-anchor-id="methods">Methods</h4>
|
||||
<table class="caption-top table">
|
||||
<thead>
|
||||
<tr class="header">
|
||||
<th>Name</th>
|
||||
<th>Description</th>
|
||||
</tr>
|
||||
</thead>
|
||||
<tbody>
|
||||
<tr class="odd">
|
||||
<td><a href="#axolotl.core.builders.rl.HFRLTrainerBuilder.build_collator">build_collator</a></td>
|
||||
<td>Build a data collator for preference-tuning trainers.</td>
|
||||
</tr>
|
||||
</tbody>
|
||||
</table>
|
||||
<section id="axolotl.core.builders.rl.HFRLTrainerBuilder.build_collator" class="level5">
|
||||
<h5 class="anchored" data-anchor-id="axolotl.core.builders.rl.HFRLTrainerBuilder.build_collator">build_collator</h5>
|
||||
<div class="code-copy-outer-scaffold"><div class="sourceCode" id="cb2"><pre class="sourceCode python code-with-copy"><code class="sourceCode python"><span id="cb2-1"><a href="#cb2-1" aria-hidden="true" tabindex="-1"></a>core.builders.rl.HFRLTrainerBuilder.build_collator(<span class="op">**</span>kwargs)</span></code></pre></div><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></div>
|
||||
<p>Build a data collator for preference-tuning trainers.</p>
|
||||
<p>Returns None for RL types that provide their own collator (e.g. GRPO,
|
||||
KTO), letting the trainer construct its default. For DPO/IPO/ORPO/SIMPO
|
||||
returns an <code>AxolotlDPODataCollatorWithPadding</code> when
|
||||
<code>pad_to_multiple_of</code> is set, otherwise None (so the trainer
|
||||
falls back to the TRL default).</p>
|
||||
|
||||
|
||||
</section>
|
||||
</section>
|
||||
</section>
|
||||
</section>
|
||||
</section>
|
||||
|
||||
Reference in New Issue
Block a user