Built site for gh-pages

.nojekyll

@@ -1 +1 @@
-6229eb51
+21d9bf7f

docs/api/cli.args.html

@@ -556,7 +556,8 @@ pre > code.sourceCode > span > a:first-child::before { text-decoration: underlin
 <span id="cb5-7"><a href="#cb5-7" aria-hidden="true" tabindex="-1"></a>    dtype<span class="op">=</span><span class="va">None</span>,</span>
 <span id="cb5-8"><a href="#cb5-8" aria-hidden="true" tabindex="-1"></a>    max_model_len<span class="op">=</span><span class="va">None</span>,</span>
 <span id="cb5-9"><a href="#cb5-9" aria-hidden="true" tabindex="-1"></a>    enable_prefix_caching<span class="op">=</span><span class="va">None</span>,</span>
-<span id="cb5-10"><a href="#cb5-10" aria-hidden="true" tabindex="-1"></a>)</span></code><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></pre></div>
+<span id="cb5-10"><a href="#cb5-10" aria-hidden="true" tabindex="-1"></a>    serve_module<span class="op">=</span><span class="va">None</span>,</span>
+<span id="cb5-11"><a href="#cb5-11" aria-hidden="true" tabindex="-1"></a>)</span></code><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></pre></div>
 <p>Dataclass with CLI arguments for <code>axolotl vllm-serve</code> command.</p>
 
 

docs/api/index.html

@@ -926,8 +926,12 @@ ul.task-list li input[type="checkbox"] {
 <td>data handling specific to SFT</td>
 </tr>
 <tr class="even">
-<td><a href="../../docs/api/utils.gradient_checkpointing.unsloth.html#axolotl.utils.gradient_checkpointing.unsloth">utils.gradient_checkpointing.unsloth</a></td>
-<td>Unsloth checkpointing</td>
+<td><a href="../../docs/api/utils.gradient_checkpointing.offload_cpu.html#axolotl.utils.gradient_checkpointing.offload_cpu">utils.gradient_checkpointing.offload_cpu</a></td>
+<td>CPU offloaded checkpointing</td>
+</tr>
+<tr class="odd">
+<td><a href="../../docs/api/utils.gradient_checkpointing.offload_disk.html#axolotl.utils.gradient_checkpointing.offload_disk">utils.gradient_checkpointing.offload_disk</a></td>
+<td>DISCO - DIsk-based Storage and Checkpointing with Optimized prefetching</td>
 </tr>
 </tbody>
 </table>

docs/api/utils.gradient_checkpointing.offload_cpu.html

@@ -7,7 +7,7 @@
 <meta name="viewport" content="width=device-width, initial-scale=1.0, user-scalable=yes">
 
 
-<title>utils.gradient_checkpointing.unsloth – Axolotl</title>
+<title>utils.gradient_checkpointing.offload_cpu – Axolotl</title>
 <style>
 code{white-space: pre-wrap;}
 span.smallcaps{font-variant: small-caps;}
@@ -446,11 +446,11 @@ pre > code.sourceCode > span > a:first-child::before { text-decoration: underlin
     <h2 id="toc-title">On this page</h2>
    
   <ul>
-  <li><a href="#axolotl.utils.gradient_checkpointing.unsloth" id="toc-axolotl.utils.gradient_checkpointing.unsloth" class="nav-link active" data-scroll-target="#axolotl.utils.gradient_checkpointing.unsloth">utils.gradient_checkpointing.unsloth</a>
+  <li><a href="#axolotl.utils.gradient_checkpointing.offload_cpu" id="toc-axolotl.utils.gradient_checkpointing.offload_cpu" class="nav-link active" data-scroll-target="#axolotl.utils.gradient_checkpointing.offload_cpu">utils.gradient_checkpointing.offload_cpu</a>
   <ul class="collapse">
   <li><a href="#classes" id="toc-classes" class="nav-link" data-scroll-target="#classes">Classes</a>
   <ul class="collapse">
-  <li><a href="#axolotl.utils.gradient_checkpointing.unsloth.Unsloth_Offloaded_Gradient_Checkpointer" id="toc-axolotl.utils.gradient_checkpointing.unsloth.Unsloth_Offloaded_Gradient_Checkpointer" class="nav-link" data-scroll-target="#axolotl.utils.gradient_checkpointing.unsloth.Unsloth_Offloaded_Gradient_Checkpointer">Unsloth_Offloaded_Gradient_Checkpointer</a></li>
+  <li><a href="#axolotl.utils.gradient_checkpointing.offload_cpu.CPU_Offloaded_Gradient_Checkpointer" id="toc-axolotl.utils.gradient_checkpointing.offload_cpu.CPU_Offloaded_Gradient_Checkpointer" class="nav-link" data-scroll-target="#axolotl.utils.gradient_checkpointing.offload_cpu.CPU_Offloaded_Gradient_Checkpointer">CPU_Offloaded_Gradient_Checkpointer</a></li>
   </ul></li>
   </ul></li>
   </ul>
@@ -462,10 +462,10 @@ pre > code.sourceCode > span > a:first-child::before { text-decoration: underlin
 
 
 
-<section id="axolotl.utils.gradient_checkpointing.unsloth" class="level1">
-<h1>utils.gradient_checkpointing.unsloth</h1>
-<p><code>utils.gradient_checkpointing.unsloth</code></p>
-<p>Unsloth checkpointing</p>
+<section id="axolotl.utils.gradient_checkpointing.offload_cpu" class="level1">
+<h1>utils.gradient_checkpointing.offload_cpu</h1>
+<p><code>utils.gradient_checkpointing.offload_cpu</code></p>
+<p>CPU offloaded checkpointing</p>
 <section id="classes" class="level2">
 <h2 class="anchored" data-anchor-id="classes">Classes</h2>
 <table class="caption-top table">
@@ -477,14 +477,14 @@ pre > code.sourceCode > span > a:first-child::before { text-decoration: underlin
 </thead>
 <tbody>
 <tr class="odd">
-<td><a href="#axolotl.utils.gradient_checkpointing.unsloth.Unsloth_Offloaded_Gradient_Checkpointer">Unsloth_Offloaded_Gradient_Checkpointer</a></td>
+<td><a href="#axolotl.utils.gradient_checkpointing.offload_cpu.CPU_Offloaded_Gradient_Checkpointer">CPU_Offloaded_Gradient_Checkpointer</a></td>
 <td>Saves VRAM by smartly offloading to RAM.</td>
 </tr>
 </tbody>
 </table>
-<section id="axolotl.utils.gradient_checkpointing.unsloth.Unsloth_Offloaded_Gradient_Checkpointer" class="level3">
-<h3 class="anchored" data-anchor-id="axolotl.utils.gradient_checkpointing.unsloth.Unsloth_Offloaded_Gradient_Checkpointer">Unsloth_Offloaded_Gradient_Checkpointer</h3>
-<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>utils.gradient_checkpointing.unsloth.Unsloth_Offloaded_Gradient_Checkpointer()</span></code><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></pre></div>
+<section id="axolotl.utils.gradient_checkpointing.offload_cpu.CPU_Offloaded_Gradient_Checkpointer" class="level3">
+<h3 class="anchored" data-anchor-id="axolotl.utils.gradient_checkpointing.offload_cpu.CPU_Offloaded_Gradient_Checkpointer">CPU_Offloaded_Gradient_Checkpointer</h3>
+<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>utils.gradient_checkpointing.offload_cpu.CPU_Offloaded_Gradient_Checkpointer()</span></code><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></pre></div>
 <p>Saves VRAM by smartly offloading to RAM.
 Tiny hit to performance, since we mask the movement via non blocking calls.</p>
 

docs/config.html

@@ -1007,7 +1007,7 @@ pre > code.sourceCode > span > a:first-child::before { text-decoration: underlin
 <span id="cb1-533"><a href="#cb1-533" aria-hidden="true" tabindex="-1"></a><span class="co"># Note that training loss may have an oscillating pattern with this enabled.</span></span>
 <span id="cb1-534"><a href="#cb1-534" aria-hidden="true" tabindex="-1"></a><span class="fu">group_by_length</span><span class="kw">:</span><span class="at"> </span><span class="ch">false</span></span>
 <span id="cb1-535"><a href="#cb1-535" aria-hidden="true" tabindex="-1"></a></span>
-<span id="cb1-536"><a href="#cb1-536" aria-hidden="true" tabindex="-1"></a><span class="co"># Whether to use gradient checkpointing. Available options are: true, false, "offload".</span></span>
+<span id="cb1-536"><a href="#cb1-536" aria-hidden="true" tabindex="-1"></a><span class="co"># Whether to use gradient checkpointing. Available options are: true, false, "offload", "offload_disk".</span></span>
 <span id="cb1-537"><a href="#cb1-537" aria-hidden="true" tabindex="-1"></a><span class="co"># https://huggingface.co/docs/transformers/v4.18.0/en/performance#gradient-checkpointing</span></span>
 <span id="cb1-538"><a href="#cb1-538" aria-hidden="true" tabindex="-1"></a><span class="fu">gradient_checkpointing</span><span class="kw">:</span><span class="at"> </span><span class="ch">false</span></span>
 <span id="cb1-539"><a href="#cb1-539" aria-hidden="true" tabindex="-1"></a><span class="co"># additional kwargs to pass to the trainer for gradient checkpointing</span></span>