Merge branch 'main' of github.com:microsoft/Trace into main

docs/_config.yml

@@ -34,6 +34,8 @@ html:
   extra_footer: "<a href='mailto:[email protected]'>Contact Us</a> | <a href='http://go.microsoft.com/fwlink/?LinkId=521839'>Privacy &amp; Cookies</a> | <a href='https://go.microsoft.com/fwlink/?linkid=2259814'>Consumer Health Privacy</a> | <a href='https://go.microsoft.com/fwlink/?LinkID=206977'>Terms Of Use</a> | <a href='https://www.microsoft.com/trademarks'>Trademarks</a>"
 
 sphinx:
+  html_context:
+    default_mode: light
   extra_extensions:
   - autodoc2
   config:

docs/_static/custom.css

@@ -24,4 +24,12 @@
 .admonition, div.admonition {
     background-color: #eef9fd;
     border-color: #4cb3d4;
+}
+
+.admonition p {
+    color: #474747;
+}
+
+.text_html p {
+    color: #474747;
 }

docs/tutorials/basic_tutorial.ipynb

@@ -296,7 +296,7 @@
     "\n",
     "In addition to magic methods, we can use `bundle` to write custom methods that are traceable. When decorating a method with `bundle`, it needs a description of the method. It has a format of `[method_name] description`. `bundle` will automatically add all nodes whose `data` attribute is used within the function as the parents of the output `MessageNode`.\n",
     "\n",
-    "Given a function `fun`, the decorated function `bundle(description)(fun)` by default will unpack all the inputs (i.e. it unpacks all the data inside nodes), send them to `fun`, and then creates a `MessageNode` to wrap the output of `fun` which has parents containing all the nodes used in this operation. "
+    "Given a function `fun`, the decorated function by default will unpack all the inputs (i.e. it unpacks all the data inside nodes), send them to `fun`, and then creates a `MessageNode` to wrap the output of `fun` which has parents containing all the nodes used in this operation. "
    ]
   },
   {
@@ -900,7 +900,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "newtrace",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
    "name": "python3"
   },
@@ -914,7 +914,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.9.19"
+   "version": "3.9.5"
   }
  },
  "nbformat": 4,

docs/tutorials/quick_start.ipynb

@@ -277,7 +277,7 @@
     "In order for the optimization code to run, create a file with name `OAI_CONFIG_LIST` in the same folder as this notebook. This file should look the same as [OAI_CONFIG_LIST](https://github.com/microsoft/autogen/blob/main/OAI_CONFIG_LIST_sample).\n",
     "\n",
     "The code below looks like any PyTorch code. We can break it down to a few steps:\n",
-    "1. We first create an optimizer from `opto.optimizers.FunctionOptimizer`.\n",
+    "1. We first import an optimizer from `opto.optimizers.OptoPrime`.\n",
     "2. The bundled function `strange_sort_list` has a convenience method `.parameters()` for us to grab all the trainable parameters in it. In this case, the trainable parameter is just the function code itself.\n",
     "3. We need to perform `backward` pass on a `MessageNode` object, which is `correctness`. We also need to pass in the `feedback` into the optimizer as well.\n",
     "4. We call `optimizer.step()` to perform the actual optimization"
@@ -541,6 +541,14 @@
     "print(new_test_output)"
    ]
   },
+  {
+   "cell_type": "markdown",
+   "id": "afd8d1b8-7d88-40c3-b1eb-6abd85edfad7",
+   "metadata": {},
+   "source": [
+    "**Now the answer is correct!** (alternating numbers between min, max, min, max!)"
+   ]
+  },
   {
    "cell_type": "markdown",
    "id": "a58756c0-c740-4d8e-850b-4c14a5f80486",
@@ -570,14 +578,6 @@
     "\n",
     "The functionalities provided by Trace seem simple and straightforward right now. But using these as basic building blocks, we can construct LLM-based workflow that is complex and mind-blowing. Continue to the next section to find out!"
    ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "691b2644-324b-4cd8-a12a-69aa8d5eb094",
-   "metadata": {},
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {