Update docs from 09dde09

_modules/olive/evaluator/olive_evaluator.html

@@ -153,7 +153,7 @@ <h1>Source code for olive.evaluator.olive_evaluator</h1><div class="highlight"><
 <span class="kn">from</span> <span class="nn">olive.common.ort_inference</span> <span class="kn">import</span> <span class="n">OrtInferenceSession</span><span class="p">,</span> <span class="n">prepare_io_bindings</span>
 <span class="kn">from</span> <span class="nn">olive.common.pydantic_v1</span> <span class="kn">import</span> <span class="n">Field</span><span class="p">,</span> <span class="n">root_validator</span><span class="p">,</span> <span class="n">validator</span>
 <span class="kn">from</span> <span class="nn">olive.common.user_module_loader</span> <span class="kn">import</span> <span class="n">UserModuleLoader</span>
-<span class="kn">from</span> <span class="nn">olive.common.utils</span> <span class="kn">import</span> <span class="n">tensor_data_to_device</span>
+<span class="kn">from</span> <span class="nn">olive.common.utils</span> <span class="kn">import</span> <span class="n">load_weights</span><span class="p">,</span> <span class="n">tensor_data_to_device</span>
 <span class="kn">from</span> <span class="nn">olive.constants</span> <span class="kn">import</span> <span class="n">Framework</span>
 <span class="kn">from</span> <span class="nn">olive.data.config</span> <span class="kn">import</span> <span class="n">DataConfig</span>
 <span class="kn">from</span> <span class="nn">olive.data.container.dummy_data_container</span> <span class="kn">import</span> <span class="n">TRANSFORMER_DUMMY_DATA_CONTAINER</span>
@@ -530,7 +530,7 @@ <h1>Source code for olive.evaluator.olive_evaluator</h1><div class="highlight"><
         <span class="c1"># load constant inputs if any</span>
         <span class="n">constant_inputs</span> <span class="o">=</span> <span class="kc">None</span>
         <span class="k">if</span> <span class="n">model</span><span class="o">.</span><span class="n">constant_inputs_path</span><span class="p">:</span>
-            <span class="n">constant_inputs</span> <span class="o">=</span> <span class="n">OnnxEvaluator</span><span class="o">.</span><span class="n">format_input</span><span class="p">(</span><span class="nb">dict</span><span class="p">(</span><span class="n">np</span><span class="o">.</span><span class="n">load</span><span class="p">(</span><span class="n">model</span><span class="o">.</span><span class="n">constant_inputs_path</span><span class="p">)),</span> <span class="n">io_config</span><span class="p">)</span>
+            <span class="n">constant_inputs</span> <span class="o">=</span> <span class="n">OnnxEvaluator</span><span class="o">.</span><span class="n">format_input</span><span class="p">(</span><span class="n">load_weights</span><span class="p">(</span><span class="n">model</span><span class="o">.</span><span class="n">constant_inputs_path</span><span class="p">),</span> <span class="n">io_config</span><span class="p">)</span>
 
         <span class="c1"># create session wrapper</span>
         <span class="n">session_wrapper</span> <span class="o">=</span> <span class="n">OrtInferenceSession</span><span class="p">(</span>

_modules/olive/hardware/accelerator.html

@@ -151,10 +151,7 @@ <h1>Source code for olive.hardware.accelerator</h1><div class="highlight"><pre>
     <span class="n">GPU</span> <span class="o">=</span> <span class="s2">&quot;gpu&quot;</span>
     <span class="n">NPU</span> <span class="o">=</span> <span class="s2">&quot;npu&quot;</span>
     <span class="n">VPU</span> <span class="o">=</span> <span class="s2">&quot;vpu&quot;</span>
-    <span class="n">INTEL_MYRIAD</span> <span class="o">=</span> <span class="s2">&quot;intel_myriad&quot;</span>
-
-    <span class="k">def</span> <span class="fm">__str__</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">str</span><span class="p">:</span>
-        <span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">value</span></div>
+    <span class="n">INTEL_MYRIAD</span> <span class="o">=</span> <span class="s2">&quot;intel_myriad&quot;</span></div>
 
 
 <span class="nd">@dataclass</span><span class="p">(</span><span class="n">frozen</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">eq</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>

_modules/olive/model/handler/onnx.html

@@ -139,11 +139,11 @@ <h1>Source code for olive.model.handler.onnx</h1><div class="highlight"><pre>
 <span class="kn">from</span> <span class="nn">pathlib</span> <span class="kn">import</span> <span class="n">Path</span>
 <span class="kn">from</span> <span class="nn">typing</span> <span class="kn">import</span> <span class="n">Any</span><span class="p">,</span> <span class="n">ClassVar</span><span class="p">,</span> <span class="n">Dict</span><span class="p">,</span> <span class="n">List</span><span class="p">,</span> <span class="n">Optional</span><span class="p">,</span> <span class="n">Tuple</span><span class="p">,</span> <span class="n">Union</span>
 
-<span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span>
 <span class="kn">import</span> <span class="nn">onnx</span>
 <span class="kn">from</span> <span class="nn">onnx</span> <span class="kn">import</span> <span class="n">GraphProto</span><span class="p">,</span> <span class="n">ModelProto</span>
 
 <span class="kn">from</span> <span class="nn">olive.common.ort_inference</span> <span class="kn">import</span> <span class="n">OrtSessionFallbackError</span><span class="p">,</span> <span class="n">get_ort_inference_session</span>
+<span class="kn">from</span> <span class="nn">olive.common.utils</span> <span class="kn">import</span> <span class="n">load_weights</span>
 <span class="kn">from</span> <span class="nn">olive.constants</span> <span class="kn">import</span> <span class="n">Framework</span><span class="p">,</span> <span class="n">ModelFileFormat</span>
 <span class="kn">from</span> <span class="nn">olive.exception</span> <span class="kn">import</span> <span class="n">OliveEvaluationError</span>
 <span class="kn">from</span> <span class="nn">olive.hardware.accelerator</span> <span class="kn">import</span> <span class="n">AcceleratorLookup</span><span class="p">,</span> <span class="n">Device</span>
@@ -243,7 +243,9 @@ <h1>Source code for olive.model.handler.onnx</h1><div class="highlight"><pre>
         <span class="c1"># device id for ranked model</span>
         <span class="n">device_id</span> <span class="o">=</span> <span class="n">rank</span> <span class="k">if</span> <span class="n">device</span> <span class="o">==</span> <span class="n">Device</span><span class="o">.</span><span class="n">GPU</span> <span class="k">else</span> <span class="kc">None</span>
         <span class="c1"># load external initializers if available</span>
-        <span class="n">external_initializers</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">load</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">external_initializers_path</span><span class="p">)</span> <span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">external_initializers_path</span> <span class="k">else</span> <span class="kc">None</span>
+        <span class="n">external_initializers</span> <span class="o">=</span> <span class="p">(</span>
+            <span class="n">load_weights</span><span class="p">(</span><span class="bp">self</span><span class="o">.</span><span class="n">external_initializers_path</span><span class="p">)</span> <span class="k">if</span> <span class="bp">self</span><span class="o">.</span><span class="n">external_initializers_path</span> <span class="k">else</span> <span class="kc">None</span>
+        <span class="p">)</span>
 
         <span class="k">try</span><span class="p">:</span>
             <span class="k">return</span> <span class="n">get_ort_inference_session</span><span class="p">(</span>

_modules/olive/resource_path.html

@@ -167,10 +167,7 @@ <h1>Source code for olive.resource_path</h1><div class="highlight"><pre>
     <span class="n">AzureMLModel</span> <span class="o">=</span> <span class="s2">&quot;azureml_model&quot;</span>
     <span class="n">AzureMLRegistryModel</span> <span class="o">=</span> <span class="s2">&quot;azureml_registry_model&quot;</span>
     <span class="n">AzureMLDatastore</span> <span class="o">=</span> <span class="s2">&quot;azureml_datastore&quot;</span>
-    <span class="n">AzureMLJobOutput</span> <span class="o">=</span> <span class="s2">&quot;azureml_job_output&quot;</span>
-
-    <span class="k">def</span> <span class="fm">__str__</span><span class="p">(</span><span class="bp">self</span><span class="p">)</span> <span class="o">-&gt;</span> <span class="nb">str</span><span class="p">:</span>
-        <span class="k">return</span> <span class="bp">self</span><span class="o">.</span><span class="n">value</span></div>
+    <span class="n">AzureMLJobOutput</span> <span class="o">=</span> <span class="s2">&quot;azureml_job_output&quot;</span></div>
 
 
 <span class="n">LOCAL_RESOURCE_TYPES</span> <span class="o">=</span> <span class="p">(</span><span class="n">ResourceType</span><span class="o">.</span><span class="n">LocalFile</span><span class="p">,</span> <span class="n">ResourceType</span><span class="o">.</span><span class="n">LocalFolder</span><span class="p">)</span>

api/passes.html

@@ -3350,6 +3350,15 @@
 <p><strong>searchable_values:</strong> None</p>
 </dd></dl>
 
+<dl class="std option">
+<dt class="sig sig-object std" id="cmdoption-arg-save_format">
+<span class="sig-name descname"><span class="pre">save_format</span></span><span class="sig-prename descclassname"></span><a class="headerlink" href="#cmdoption-arg-save_format" title="Permalink to this definition">¶</a></dt>
+<dd><p>Format to save the weights in.</p>
+<p><strong>type:</strong> olive.common.utils.WeightsFileFormat</p>
+<p><strong>default_value:</strong> numpy</p>
+<p><strong>searchable_values:</strong> None</p>
+</dd></dl>
+
 <dl class="std option">
 <dt class="sig sig-object std" id="cmdoption-arg-184">
 <span class="sig-name descname"><span class="pre">save_as_external_data</span></span><span class="sig-prename descclassname"></span><a class="headerlink" href="#cmdoption-arg-184" title="Permalink to this definition">¶</a></dt>

features/cli.html

@@ -335,11 +335,11 @@ <h4>remote options<a class="headerlink" href="#remote-options" title="Permalink
 </section>
 <section id="export-adapters">
 <h3>export-adapters<a class="headerlink" href="#export-adapters" title="Permalink to this heading">¶</a></h3>
-<p>Export lora adapter weights to a .npz file that will be consumed by ONNX models generated by Olive ExtractedAdapters pass.</p>
-<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">olive</span> <span class="n">export</span><span class="o">-</span><span class="n">adapters</span> <span class="p">[</span><span class="o">-</span><span class="n">h</span><span class="p">]</span> <span class="p">[</span><span class="o">--</span><span class="n">adapter_path</span> <span class="n">ADAPTER_PATH</span><span class="p">]</span>
-                      <span class="p">[</span><span class="o">--</span><span class="n">output_path</span> <span class="n">OUTPUT_PATH</span><span class="p">]</span> <span class="p">[</span><span class="o">--</span><span class="n">dtype</span> <span class="p">{</span><span class="n">float32</span><span class="p">,</span><span class="n">float16</span><span class="p">}]</span>
-                      <span class="p">[</span><span class="o">--</span><span class="n">pack_weights</span><span class="p">]</span> <span class="p">[</span><span class="o">--</span><span class="n">quantize_int4</span><span class="p">]</span>
-                      <span class="p">[</span><span class="o">--</span><span class="n">int4_block_size</span> <span class="p">{</span><span class="mi">16</span><span class="p">,</span><span class="mi">32</span><span class="p">,</span><span class="mi">64</span><span class="p">,</span><span class="mi">128</span><span class="p">,</span><span class="mi">256</span><span class="p">}]</span>
+<p>Export lora adapter weights to a file that will be consumed by ONNX models generated by Olive ExtractedAdapters pass.</p>
+<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">olive</span> <span class="n">export</span><span class="o">-</span><span class="n">adapters</span> <span class="p">[</span><span class="o">-</span><span class="n">h</span><span class="p">]</span> <span class="o">--</span><span class="n">adapter_path</span> <span class="n">ADAPTER_PATH</span>
+                      <span class="p">[</span><span class="o">--</span><span class="n">save_format</span> <span class="p">{</span><span class="n">pt</span><span class="p">,</span><span class="n">numpy</span><span class="p">,</span><span class="n">safetensors</span><span class="p">}]</span> <span class="o">--</span><span class="n">output_path</span>
+                      <span class="n">OUTPUT_PATH</span> <span class="p">[</span><span class="o">--</span><span class="n">dtype</span> <span class="p">{</span><span class="n">float32</span><span class="p">,</span><span class="n">float16</span><span class="p">}]</span> <span class="p">[</span><span class="o">--</span><span class="n">pack_weights</span><span class="p">]</span>
+                      <span class="p">[</span><span class="o">--</span><span class="n">quantize_int4</span><span class="p">]</span> <span class="p">[</span><span class="o">--</span><span class="n">int4_block_size</span> <span class="p">{</span><span class="mi">16</span><span class="p">,</span><span class="mi">32</span><span class="p">,</span><span class="mi">64</span><span class="p">,</span><span class="mi">128</span><span class="p">,</span><span class="mi">256</span><span class="p">}]</span>
                       <span class="p">[</span><span class="o">--</span><span class="n">int4_quantization_mode</span> <span class="p">{</span><span class="n">symmetric</span><span class="p">,</span><span class="n">asymmetric</span><span class="p">}]</span>
 </pre></div>
 </div>
@@ -349,8 +349,13 @@ <h4>Named Arguments<a class="headerlink" href="#named-arguments_repeat2" title="
 <dt><kbd>--adapter_path</kbd></dt>
 <dd><p>Path to the adapters weights saved after peft fine-tuning. Can be a local folder or huggingface id.</p>
 </dd>
+<dt><kbd>--save_format</kbd></dt>
+<dd><p>Possible choices: pt, numpy, safetensors</p>
+<p>Format to save the weights in. Default is numpy.</p>
+<p>Default: “numpy”</p>
+</dd>
 <dt><kbd>--output_path</kbd></dt>
-<dd><p>Path to save the exported weights. Will be saved as a .npz file.</p>
+<dd><p>Path to save the exported weights. Will be saved in the <cite>save_format</cite> format.</p>
 </dd>
 <dt><kbd>--dtype</kbd></dt>
 <dd><p>Possible choices: float32, float16</p>

genindex.html

@@ -798,6 +798,8 @@ <h2 id="C">C</h2>
         <li><a href="api/passes.html#cmdoption-arg-round_interval">round_interval</a>
 </li>
         <li><a href="api/passes.html#cmdoption-arg-1">save_as_external_data</a>, <a href="api/passes.html#cmdoption-arg-100">[1]</a>, <a href="api/passes.html#cmdoption-arg-107">[2]</a>, <a href="api/passes.html#cmdoption-arg-11">[3]</a>, <a href="api/passes.html#cmdoption-arg-127">[4]</a>, <a href="api/passes.html#cmdoption-arg-148">[5]</a>, <a href="api/passes.html#cmdoption-arg-167">[6]</a>, <a href="api/passes.html#cmdoption-arg-173">[7]</a>, <a href="api/passes.html#cmdoption-arg-179">[8]</a>, <a href="api/passes.html#cmdoption-arg-184">[9]</a>, <a href="api/passes.html#cmdoption-arg-20">[10]</a>, <a href="api/passes.html#cmdoption-arg-223">[11]</a>, <a href="api/passes.html#cmdoption-arg-25">[12]</a>, <a href="api/passes.html#cmdoption-arg-258">[13]</a>, <a href="api/passes.html#cmdoption-arg-31">[14]</a>, <a href="api/passes.html#cmdoption-arg-36">[15]</a>, <a href="api/passes.html#cmdoption-arg-41">[16]</a>, <a href="api/passes.html#cmdoption-arg-46">[17]</a>, <a href="api/passes.html#cmdoption-arg-6">[18]</a>, <a href="api/passes.html#cmdoption-arg-68">[19]</a>, <a href="api/passes.html#cmdoption-arg-95">[20]</a>, <a href="api/passes.html#cmdoption-arg-save_as_external_data">[21]</a>
+</li>
+        <li><a href="api/passes.html#cmdoption-arg-save_format">save_format</a>
 </li>
         <li><a href="api/passes.html#cmdoption-arg-save_metadata_for_token_generation">save_metadata_for_token_generation</a>
 </li>
@@ -2441,6 +2443,13 @@ <h2 id="S">S</h2>
 
       <ul>
         <li><a href="api/passes.html#cmdoption-arg-1">command line option</a>, <a href="api/passes.html#cmdoption-arg-100">[1]</a>, <a href="api/passes.html#cmdoption-arg-107">[2]</a>, <a href="api/passes.html#cmdoption-arg-11">[3]</a>, <a href="api/passes.html#cmdoption-arg-127">[4]</a>, <a href="api/passes.html#cmdoption-arg-148">[5]</a>, <a href="api/passes.html#cmdoption-arg-167">[6]</a>, <a href="api/passes.html#cmdoption-arg-173">[7]</a>, <a href="api/passes.html#cmdoption-arg-179">[8]</a>, <a href="api/passes.html#cmdoption-arg-184">[9]</a>, <a href="api/passes.html#cmdoption-arg-20">[10]</a>, <a href="api/passes.html#cmdoption-arg-223">[11]</a>, <a href="api/passes.html#cmdoption-arg-25">[12]</a>, <a href="api/passes.html#cmdoption-arg-258">[13]</a>, <a href="api/passes.html#cmdoption-arg-31">[14]</a>, <a href="api/passes.html#cmdoption-arg-36">[15]</a>, <a href="api/passes.html#cmdoption-arg-41">[16]</a>, <a href="api/passes.html#cmdoption-arg-46">[17]</a>, <a href="api/passes.html#cmdoption-arg-6">[18]</a>, <a href="api/passes.html#cmdoption-arg-68">[19]</a>, <a href="api/passes.html#cmdoption-arg-95">[20]</a>, <a href="api/passes.html#cmdoption-arg-save_as_external_data">[21]</a>
+</li>
+      </ul></li>
+      <li>
+    save_format
+
+      <ul>
+        <li><a href="api/passes.html#cmdoption-arg-save_format">command line option</a>
 </li>
       </ul></li>
       <li>