Created a basic demo

.gitignore

@@ -0,0 +1,163 @@
+# My stuff
+raw_data/
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/

all_genres.csv

@@ -0,0 +1,21 @@
+genre
+History
+Adventure
+Fantasy
+Action
+Crime
+Family
+Documentary
+Foreign
+Romance
+Drama
+Science Fiction
+Music
+Western
+Comedy
+Thriller
+War
+Mystery
+Horror
+TV Movie
+Animation

app.py

@@ -0,0 +1,136 @@
+from transformers import DistilBertTokenizer, DistilBertModel, \
+                         BertTokenizer, BertModel, \
+                         RobertaTokenizer, RobertaModel, \
+                         AutoTokenizer, AutoModelForMaskedLM
+import gradio as gr
+import pandas as pd
+import numpy as np
+from typing import Tuple
+from sklearn.cluster import KMeans
+
+# global variables
+# global variables
+encoder_options = [
+    'distilbert-base-uncased',
+    'bert-base-uncased',
+    'bert-base-cased',
+    'roberta-base',
+    'xlm-roberta-base',
+    ]
+
+tokenizer = None
+model = None
+
+genres = pd.read_csv("./all_genres.csv")
+genres = set(genres["genre"].to_list())
+
+
+def update_models(current_encoder: str) -> None:
+    global model, tokenizer
+    if current_encoder == 'distilbert-base-uncased':
+        tokenizer = DistilBertTokenizer.from_pretrained(
+            'distilbert-base-uncased'
+            )
+        model = DistilBertModel.from_pretrained('distilbert-base-uncased')
+    elif current_encoder == 'bert-base-uncased':
+        tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')
+        model = BertModel.from_pretrained('bert-base-uncased')
+    elif current_encoder == 'bert-base-cased':
+        tokenizer = BertTokenizer.from_pretrained('bert-base-cased')
+        model = BertModel.from_pretrained('bert-base-cased')
+    elif current_encoder == 'roberta-base':
+        tokenizer = RobertaTokenizer.from_pretrained('roberta-base')
+        model = RobertaModel.from_pretrained('roberta-base')
+    elif current_encoder == 'xlm-roberta-base':
+        tokenizer = AutoTokenizer.from_pretrained('xlm-roberta-base')
+        model = AutoModelForMaskedLM.from_pretrained('xlm-roberta-base')
+
+
+def embed_string() -> np.ndarray:
+    output = []
+    for text in genres:
+        encoded_input = tokenizer(text, return_tensors='pt')
+        # forward pass
+        new_output = model(**encoded_input)
+        to_append = new_output.last_hidden_state
+        to_append = to_append[:, -1, :]
+        to_append = to_append.flatten().detach().cpu().numpy()
+        output.append(to_append)
+    np_output = np.zeros((len(output), output[0].shape[0]))
+    for i, vector in enumerate(output):
+        np_output[i, :] = vector
+    return np_output
+
+
+def gen_clusters(
+        input_strs: np.ndarray,
+        num_clusters: int
+        ) -> Tuple[KMeans, np.ndarray, float]:
+    clustering_algo = KMeans(n_clusters=num_clusters)
+    predicted_labels = clustering_algo.fit_predict(input_strs)
+
+    cluster_error = 0.0
+    for i, predicted_label in enumerate(predicted_labels):
+        predicted_center = clustering_algo.cluster_centers_[predicted_label, :]
+        new_error = np.sqrt(np.sum(np.square(predicted_center, input_strs[i])))
+        cluster_error += new_error
+
+    return clustering_algo, predicted_labels, cluster_error
+
+
+def view_clusters(predicted_clusters: np.ndarray) -> pd.DataFrame:
+    mappings = dict()
+    for predicted_cluster, movie in zip(predicted_clusters, genres):
+        curr_mapping = mappings.get(predicted_cluster, [])
+        curr_mapping.append(movie)
+        mappings[predicted_cluster] = curr_mapping
+
+    output_df = pd.DataFrame()
+    max_len = max([len(x) for x in mappings.values()])
+    max_cluster = max(predicted_clusters)
+
+    for i in range(max_cluster + 1):
+        new_column_name = f"cluster_{i}"
+        new_column_data = mappings[i]
+        new_column_data.extend([''] * (max_len - len(new_column_data)))
+        output_df[new_column_name] = new_column_data
+
+    return output_df
+
+
+def add_new_genre(
+        new_genre: str = "",
+        num_clusters: int = 5,
+        ) -> pd.DataFrame:
+    global genres
+    if new_genre != "":
+        genres.add(new_genre)
+    embedded_genres = embed_string()
+
+    _, cluster_centers, error = gen_clusters(embedded_genres, num_clusters)
+    ouput_df = view_clusters(cluster_centers)
+    return ouput_df, error
+
+
+if __name__ == "__main__":
+    with gr.Blocks() as demo:
+        current_encoder = gr.Radio(encoder_options, label="Encoder")
+        current_encoder.change(fn=update_models, inputs=current_encoder)
+        new_genre_input = gr.Textbox(value="", label="New Genre")
+        num_clusters_input = gr.Number(
+            value=5,
+            precision=0,
+            label="Clusters"
+            )
+
+        output_clustering = gr.DataFrame()
+        output_error = gr.Number(label="Clustering Error", interactive=False)
+
+        encode_button = gr.Button(value="Run")
+        encode_button.click(
+            fn=add_new_genre,
+            inputs=[new_genre_input, num_clusters_input],
+            outputs=[output_clustering, output_error],
+            )
+
+    demo.launch()

genre_clustering.ipynb

@@ -0,0 +1,220 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/Users/nathanluskey/opt/anaconda3/envs/ml_env/lib/python3.10/site-packages/tqdm/auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
+      "  from .autonotebook import tqdm as notebook_tqdm\n"
+     ]
+    }
+   ],
+   "source": [
+    "from transformers import DistilBertTokenizer, DistilBertModel, \\\n",
+    "                         BertTokenizer, BertModel, \\\n",
+    "                         RobertaTokenizer, RobertaModel, \\\n",
+    "                         AutoTokenizer, AutoModelForMaskedLM\n",
+    "import gradio as gr\n",
+    "import pandas as pd\n",
+    "import numpy as np\n",
+    "import torch\n",
+    "from typing import List, Tuple\n",
+    "from sklearn.cluster import KMeans"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# global variables\n",
+    "encoder_options = [\n",
+    "    'distilbert-base-uncased',\n",
+    "    'bert-base-uncased',\n",
+    "    'bert-base-cased'\n",
+    "    'roberta-base',\n",
+    "    'xlm-roberta-base',\n",
+    "    ]\n",
+    "\n",
+    "current_encoder = encoder_options[0]\n",
+    "tokenizer = None\n",
+    "model = None\n",
+    "\n",
+    "genres = pd.read_csv(\"./all_genres.csv\")\n",
+    "genres = genres[\"genre\"].to_list()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "Some weights of the model checkpoint at distilbert-base-uncased were not used when initializing DistilBertModel: ['vocab_projector.bias', 'vocab_layer_norm.weight', 'vocab_transform.weight', 'vocab_layer_norm.bias', 'vocab_transform.bias', 'vocab_projector.weight']\n",
+      "- This IS expected if you are initializing DistilBertModel from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
+      "- This IS NOT expected if you are initializing DistilBertModel from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n"
+     ]
+    }
+   ],
+   "source": [
+    "if current_encoder == 'distilbert-base-uncased':\n",
+    "    tokenizer = DistilBertTokenizer.from_pretrained('distilbert-base-uncased')\n",
+    "    model = DistilBertModel.from_pretrained('distilbert-base-uncased')\n",
+    "elif current_encoder == 'bert-base-uncased':\n",
+    "    tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\n",
+    "    model = BertModel.from_pretrained('bert-base-uncased')\n",
+    "elif current_encoder == 'bert-base-cased':\n",
+    "    tokenizer = BertTokenizer.from_pretrained('bert-base-cased')\n",
+    "    model = BertModel.from_pretrained('bert-base-cased')\n",
+    "elif current_encoder == 'roberta-base':\n",
+    "    tokenizer = RobertaTokenizer.from_pretrained('roberta-base')\n",
+    "    model = RobertaModel.from_pretrained('roberta-base')\n",
+    "elif current_encoder == 'xlm-roberta-base':\n",
+    "    tokenizer = AutoTokenizer.from_pretrained('xlm-roberta-base')\n",
+    "    model = AutoModelForMaskedLM.from_pretrained('xlm-roberta-base')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def embed_string() -> np.ndarray:\n",
+    "    output = []\n",
+    "    for text in genres:\n",
+    "        encoded_input = tokenizer(text, return_tensors='pt')\n",
+    "        # forward pass\n",
+    "        new_output = model(**encoded_input)\n",
+    "        to_append = new_output.last_hidden_state\n",
+    "        to_append = to_append[:, -1, :] #Take the last element\n",
+    "        to_append = to_append.flatten().detach().cpu().numpy()\n",
+    "        output.append(to_append)\n",
+    "    np_output = np.zeros((len(output), output[0].shape[0]))\n",
+    "    for i, vector in enumerate(output):\n",
+    "        np_output[i, :] = vector\n",
+    "    return np_output"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def gen_clusters(input_strs:np.ndarray, num_clusters:int) -> Tuple[KMeans, np.ndarray, float]:\n",
+    "    clustering_algo = KMeans(n_clusters=num_clusters)\n",
+    "    predicted_labels = clustering_algo.fit_predict(input_strs)\n",
+    "\n",
+    "    cluster_error = 0.0\n",
+    "    for i, predicted_label in enumerate(predicted_labels):\n",
+    "        predicted_center = clustering_algo.cluster_centers_[predicted_label, :]\n",
+    "        new_error = np.sqrt(np.sum(np.square(predicted_center, input_strs[i])))\n",
+    "        cluster_error += new_error\n",
+    "\n",
+    "    return clustering_algo, predicted_labels, cluster_error\n",
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 16,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def view_clusters(predicted_clusters:np.ndarray) -> pd.DataFrame:\n",
+    "    mappings = dict()\n",
+    "    for predicted_cluster, movie in zip(predicted_clusters, genres):\n",
+    "        curr_mapping = mappings.get(predicted_cluster, [])\n",
+    "        curr_mapping.append(movie)\n",
+    "        mappings[predicted_cluster] = curr_mapping\n",
+    "\n",
+    "    output_df = pd.DataFrame()\n",
+    "    max_len = max([len(x) for x in mappings.values()])\n",
+    "    max_cluster = max(predicted_clusters)\n",
+    "\n",
+    "    for i in range(max_cluster + 1):\n",
+    "        new_column_name = f\"cluster_{i}\"\n",
+    "        new_column_data = mappings[i]\n",
+    "        new_column_data.extend([''] * (max_len - len(new_column_data)))\n",
+    "        output_df[new_column_name] = new_column_data\n",
+    "\n",
+    "    return output_df"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 17,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def add_new_genre(clustering_algo:KMeans, new_genre:str, recompute:bool = False) -> pd.DataFrame:\n",
+    "    global genres\n",
+    "    genres.append(new_genre)\n",
+    "    embedded_genres = embed_string()\n",
+    "    if recompute:\n",
+    "        cluster_algo, cluster_centers, error = gen_clusters(embedded_genres, 5)\n",
+    "    else:\n",
+    "       cluster_centers = cluster_algo.predict(embedded_genres)\n",
+    "    \n",
+    "    ouput_df = view_clusters(cluster_centers)\n",
+    "    return ouput_df\n",
+    "    "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 18,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "embedded_genres = embed_string()\n",
+    "clustering_algo, predicted_labels, cluster_error = gen_clusters(embedded_genres, 5)\n",
+    "output_df = view_clusters(predicted_labels)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3.10.6 ('ml_env')",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.10.6"
+  },
+  "orig_nbformat": 4,
+  "vscode": {
+   "interpreter": {
+    "hash": "2434bee09bcd67f653a1f2d2df1f4f18cabf9d6c39b42950acaa6ef605d590bc"
+   }
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}