Baal in Production Notebook | Classification | NLP | Hugging Face

docs/tutorials/index.md

@@ -6,7 +6,8 @@ latter on how we integrate with other common frameworks such as Label Studio, Hu
 ## :material-file-tree: How to
 
 * [Run an active learning experiments](notebooks/active_learning_process.ipynb)
-* [Active learning in production](notebooks/baal_prod_cls.ipynb)
+* [Active learning in production (Image Classification)](notebooks/production/baal_prod_cls.ipynb)
+* [Active learning in production (Text Classification)](notebooks/production/baal_prod_cls_nlp_hf.ipynb)
 * [Deep Ensembles](../notebooks/deep_ensemble.ipynb)
 
 ## :material-file-tree: Compatibility

mkdocs.yml

@@ -91,8 +91,10 @@ nav:
       - Compatibility:
         - HuggingFace: notebooks/compatibility/nlp_classification.ipynb
         - Scikit-learn: notebooks/compatibility/sklearn_tutorial.ipynb
+      - Production use cases:
+        - Computer vision: notebooks/production/baal_prod_cls.ipynb
+        - Text classification: notebooks/production/baal_prod_cls_nlp_hf.ipynb
       - Active learning for research: notebooks/active_learning_process.ipynb
-      - Active learning for production: notebooks/baal_prod_cls.ipynb
       - Deep Ensembles for active learning: notebooks/deep_ensemble.ipynb
   - Research:
       - research/index.md

notebooks/baal_prod_cls.ipynb → notebooks/production/baal_prod_cls.ipynb

@@ -3,13 +3,10 @@
   {
    "cell_type": "markdown",
    "metadata": {
-    "collapsed": true,
-    "pycharm": {
-     "name": "#%% md\n"
-    }
+    "collapsed": true
    },
    "source": [
-    "# Use Baal in production (Classification)\n",
+    "# Use Baal in production (Image classification)\n",
     "\n",
     "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/baal-org/baal/blob/master/notebooks/baal_prod_cls.ipynb)\n",
     "\n",
@@ -35,8 +32,7 @@
    "execution_count": 1,
    "metadata": {
     "pycharm": {
-     "is_executing": false,
-     "name": "#%%\n"
+     "is_executing": false
     }
    },
    "outputs": [
@@ -60,11 +56,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
+   "metadata": {},
    "source": [
     "Introducing `baal.active.FileDataset` and `baal.active.ActiveLearningDataset`\n",
     "\n",
@@ -84,8 +76,7 @@
    "execution_count": 2,
    "metadata": {
     "pycharm": {
-     "is_executing": false,
-     "name": "#%%\n"
+     "is_executing": false
     }
    },
    "outputs": [],
@@ -114,11 +105,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
+   "metadata": {},
    "source": [
     "\n",
     "We now have two unlabeled datasets : train and validation. We encapsulate the training dataset in a \n",
@@ -139,8 +126,7 @@
    "execution_count": 3,
    "metadata": {
     "pycharm": {
-     "is_executing": false,
-     "name": "#%%\n"
+     "is_executing": false
     }
    },
    "outputs": [],
@@ -167,11 +153,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
+   "metadata": {},
    "source": [
     "### Heuristics\n",
     "\n",
@@ -185,8 +167,7 @@
    "execution_count": 4,
    "metadata": {
     "pycharm": {
-     "is_executing": false,
-     "name": "#%%\n"
+     "is_executing": false
     }
    },
    "outputs": [],
@@ -197,11 +178,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
+   "metadata": {},
    "source": [
     "### Oracle\n",
     "When the AL process requires a new item to labeled, we need to provide an Oracle. In your case, the Oracle will\n",
@@ -213,8 +190,7 @@
    "execution_count": 5,
    "metadata": {
     "pycharm": {
-     "is_executing": false,
-     "name": "#%%\n"
+     "is_executing": false
     }
    },
    "outputs": [],
@@ -227,11 +203,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
+   "metadata": {},
    "source": [
     "### Labeling process\n",
     "The labeling will go like this:\n",
@@ -248,8 +220,7 @@
    "execution_count": 6,
    "metadata": {
     "pycharm": {
-     "is_executing": false,
-     "name": "#%%\n"
+     "is_executing": false
     }
    },
    "outputs": [
@@ -282,8 +253,7 @@
    "execution_count": 7,
    "metadata": {
     "pycharm": {
-     "is_executing": false,
-     "name": "#%%\n"
+     "is_executing": false
     }
    },
    "outputs": [
@@ -326,11 +296,7 @@
   {
    "cell_type": "code",
    "execution_count": 8,
-   "metadata": {
-    "pycharm": {
-     "name": "#%%\n"
-    }
-   },
+   "metadata": {},
    "outputs": [
     {
      "name": "stdout",
@@ -355,11 +321,7 @@
   {
    "cell_type": "code",
    "execution_count": 9,
-   "metadata": {
-    "pycharm": {
-     "name": "#%%\n"
-    }
-   },
+   "metadata": {},
    "outputs": [
     {
      "name": "stdout",
@@ -383,7 +345,6 @@
    "execution_count": null,
    "metadata": {
     "pycharm": {
-     "name": "#%%\n",
      "is_executing": true
     }
    },
@@ -415,11 +376,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
+   "metadata": {},
    "source": [
     "And we're done!\n",
     "Be sure to save the dataset and the model.\n"
@@ -428,11 +385,7 @@
   {
    "cell_type": "code",
    "execution_count": 11,
-   "metadata": {
-    "pycharm": {
-     "name": "#%%\n"
-    }
-   },
+   "metadata": {},
    "outputs": [],
    "source": [
     "torch.save({\n",
@@ -444,11 +397,7 @@
   },
   {
    "cell_type": "markdown",
-   "metadata": {
-    "pycharm": {
-     "name": "#%% md\n"
-    }
-   },
+   "metadata": {},
    "source": [
     "## Support\n",
     "Submit an issue or reach us to our Slack!"
@@ -476,4 +425,4 @@
  },
  "nbformat": 4,
  "nbformat_minor": 1
-}
+}