Add gradio app example (#26)

* Put all apps in an `apps` folder
* fix formatting for html separators

buster/apps/gradio_app.ipynb

@@ -0,0 +1,126 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "4a6b2b70",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import gradio as gr\n",
+    "\n",
+    "from buster.chatbot import Chatbot, ChatbotConfig\n",
+    "\n",
+    "hf_transformers_cfg = ChatbotConfig(\n",
+    "    documents_file=\"../data/document_embeddings_hf_transformers.tar.gz\",\n",
+    "    unknown_prompt=\"This doesn't seem to be related to the huggingface library. I am not sure how to answer.\",\n",
+    "    embedding_model=\"text-embedding-ada-002\",\n",
+    "    top_k=3,\n",
+    "    thresh=0.7,\n",
+    "    max_chars=3000,\n",
+    "    completion_kwargs={\n",
+    "        \"engine\": \"text-davinci-003\",\n",
+    "        \"max_tokens\": 500,\n",
+    "    },\n",
+    "    separator=\"<br>\",\n",
+    "    link_format=\"markdown\",\n",
+    "    text_after_response=\"I'm a bot 🤖 trained to answer huggingface 🤗 transformers questions. My answers aren't always perfect.\",\n",
+    "    text_before_prompt=\"\"\"You are a slack chatbot assistant answering technical questions about huggingface transformers, a library to train transformers in python.\n",
+    "    Make sure to format your answers in Markdown format, including code block and snippets.\n",
+    "    Do not include any links to urls or hyperlinks in your answers.\n",
+    "\n",
+    "    If you do not know the answer to a question, or if it is completely irrelevant to the library usage, simply reply with:\n",
+    "\n",
+    "    'This doesn't seem to be related to the huggingface library.'\n",
+    "\n",
+    "    For example:\n",
+    "\n",
+    "    What is the meaning of life for huggingface?\n",
+    "\n",
+    "    This doesn't seem to be related to the huggingface library.\n",
+    "\n",
+    "    Now answer the following question:\n",
+    "    \"\"\",\n",
+    ")\n",
+    "hf_transformers_chatbot = Chatbot(hf_transformers_cfg)\n",
+    "\n",
+    "def chat(question, history):\n",
+    "    history = history or []\n",
+    "    answer = hf_transformers_chatbot.process_input(question)\n",
+    "\n",
+    "    history.append((question, answer))\n",
+    "    print(history)\n",
+    "    return history, history\n",
+    "\n",
+    "\n",
+    "\n",
+    "block = gr.Blocks(css=\".gradio-container {background-color: lightgray}\")\n",
+    "\n",
+    "with block:\n",
+    "    with gr.Row():\n",
+    "        gr.Markdown(\"<h3><center>Buster 🤖: A Question-Answering Bot for Huggingface 🤗 Transformers </center></h3>\")\n",
+    "\n",
+    "\n",
+    "    chatbot = gr.Chatbot()\n",
+    "\n",
+    "    with gr.Row():\n",
+    "        message = gr.Textbox(\n",
+    "            label=\"What's your question?\",\n",
+    "            placeholder=\"What kind of model should I use for sentiment analysis?\",\n",
+    "            lines=1,\n",
+    "        )\n",
+    "        submit = gr.Button(value=\"Send\", variant=\"secondary\").style(full_width=False)\n",
+    "\n",
+    "    gr.Examples(\n",
+    "        examples=[\n",
+    "            \"What kind of models should I use for images and text?\",\n",
+    "            \"When should I finetune a model vs. training it form scratch?\",\n",
+    "            \"How can I deploy my trained huggingface model?\",\n",
+    "            \"Can you give me some python code to quickly finetune a model on my sentiment analysis dataset?\",\n",
+    "        ],\n",
+    "        inputs=message,\n",
+    "    )\n",
+    "\n",
+    "    gr.Markdown(\n",
+    "    \"\"\"This simple application uses GPT to search the huggingface 🤗 transformers docs and answer questions.\n",
+    "    For more info on huggingface transformers view the [full documentation.](https://huggingface.co/docs/transformers/index).\"\"\" \n",
+    "    )\n",
+    "\n",
+    "\n",
+    "    gr.HTML(\n",
+    "        \"️<center> Created with ❤️ by @jerpint and @hadrienbertrand\"\n",
+    "    )\n",
+    "\n",
+    "    state = gr.State()\n",
+    "    agent_state = gr.State()\n",
+    "\n",
+    "    submit.click(chat, inputs=[message, state], outputs=[chatbot, state])\n",
+    "    message.submit(chat, inputs=[message, state], outputs=[chatbot, state])\n",
+    "\n",
+    "\n",
+    "block.launch(debug=True)"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.9.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

app.py → buster/apps/slackbot.py

@@ -15,7 +15,7 @@ PYTORCH_CHANNEL = "C04MEK6N882"
 HF_TRANSFORMERS_CHANNEL = "C04NJNCJWHE"
 
 mila_doc_cfg = ChatbotConfig(
-    documents_file="buster/data/document_embeddings.csv",
+    documents_file="../data/document_embeddings.csv",
     unknown_prompt="This doesn't seem to be related to cluster usage.",
     embedding_model="text-embedding-ada-002",
     top_k=3,
@@ -51,7 +51,7 @@ mila_doc_cfg = ChatbotConfig(
 mila_doc_chatbot = Chatbot(mila_doc_cfg)
 
 orion_cfg = ChatbotConfig(
-    documents_file="buster/data/document_embeddings_orion.csv",
+    documents_file="../data/document_embeddings_orion.csv",
     unknown_prompt="This doesn't seem to be related to the orion library. I am not sure how to answer.",
     embedding_model="text-embedding-ada-002",
     top_k=3,
@@ -84,7 +84,7 @@ orion_cfg = ChatbotConfig(
 orion_chatbot = Chatbot(orion_cfg)
 
 pytorch_cfg = ChatbotConfig(
-    documents_file="buster/data/document_embeddings_pytorch.tar.gz",
+    documents_file="../data/document_embeddings_pytorch.tar.gz",
     unknown_prompt="This doesn't seem to be related to the pytorch library. I am not sure how to answer.",
     embedding_model="text-embedding-ada-002",
     top_k=3,
@@ -117,7 +117,7 @@ pytorch_cfg = ChatbotConfig(
 pytorch_chatbot = Chatbot(pytorch_cfg)
 
 hf_transformers_cfg = ChatbotConfig(
-    documents_file="buster/data/document_embeddings_hf_transformers.tar.gz",
+    documents_file="../data/document_embeddings_hf_transformers.tar.gz",
     unknown_prompt="This doesn't seem to be related to the huggingface library. I am not sure how to answer.",
     embedding_model="text-embedding-ada-002",
     top_k=3,

buster/chatbot.py

@@ -128,12 +128,12 @@ class Chatbot:
         names = matched_documents.name.to_list()
         similarities = matched_documents.similarity.to_list()
 
-        response += f"{sep}{sep}Here are the sources I used to answer your question:\n"
+        response += f"{sep}{sep}Here are the sources I used to answer your question:{sep}"
         for url, name, similarity in zip(urls, names, similarities):
             if format == "markdown":
-                response += f"{sep}[{name}]({url}){sep}"
+                response += f"[{name}]({url}), relevance: {similarity:2.3f}{sep}"
             elif format == "slack":
-                response += f"• <{url}|{name}>, score: {similarity:2.3f}{sep}"
+                response += f"• <{url}|{name}>, relevance: {similarity:2.3f}{sep}"
             else:
                 raise ValueError(f"{format} is not a valid URL format.")