Add gradio app example (#26)
Browse files* Put all apps in an `apps` folder
* fix formatting for html separators
- buster/__init__.py +0 -0
- buster/apps/gradio_app.ipynb +126 -0
- app.py β buster/apps/slackbot.py +4 -4
- buster/chatbot.py +3 -3
buster/__init__.py
ADDED
File without changes
|
buster/apps/gradio_app.ipynb
ADDED
@@ -0,0 +1,126 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "code",
|
5 |
+
"execution_count": null,
|
6 |
+
"id": "4a6b2b70",
|
7 |
+
"metadata": {},
|
8 |
+
"outputs": [],
|
9 |
+
"source": [
|
10 |
+
"import gradio as gr\n",
|
11 |
+
"\n",
|
12 |
+
"from buster.chatbot import Chatbot, ChatbotConfig\n",
|
13 |
+
"\n",
|
14 |
+
"hf_transformers_cfg = ChatbotConfig(\n",
|
15 |
+
" documents_file=\"../data/document_embeddings_hf_transformers.tar.gz\",\n",
|
16 |
+
" unknown_prompt=\"This doesn't seem to be related to the huggingface library. I am not sure how to answer.\",\n",
|
17 |
+
" embedding_model=\"text-embedding-ada-002\",\n",
|
18 |
+
" top_k=3,\n",
|
19 |
+
" thresh=0.7,\n",
|
20 |
+
" max_chars=3000,\n",
|
21 |
+
" completion_kwargs={\n",
|
22 |
+
" \"engine\": \"text-davinci-003\",\n",
|
23 |
+
" \"max_tokens\": 500,\n",
|
24 |
+
" },\n",
|
25 |
+
" separator=\"<br>\",\n",
|
26 |
+
" link_format=\"markdown\",\n",
|
27 |
+
" text_after_response=\"I'm a bot π€ trained to answer huggingface π€ transformers questions. My answers aren't always perfect.\",\n",
|
28 |
+
" text_before_prompt=\"\"\"You are a slack chatbot assistant answering technical questions about huggingface transformers, a library to train transformers in python.\n",
|
29 |
+
" Make sure to format your answers in Markdown format, including code block and snippets.\n",
|
30 |
+
" Do not include any links to urls or hyperlinks in your answers.\n",
|
31 |
+
"\n",
|
32 |
+
" If you do not know the answer to a question, or if it is completely irrelevant to the library usage, simply reply with:\n",
|
33 |
+
"\n",
|
34 |
+
" 'This doesn't seem to be related to the huggingface library.'\n",
|
35 |
+
"\n",
|
36 |
+
" For example:\n",
|
37 |
+
"\n",
|
38 |
+
" What is the meaning of life for huggingface?\n",
|
39 |
+
"\n",
|
40 |
+
" This doesn't seem to be related to the huggingface library.\n",
|
41 |
+
"\n",
|
42 |
+
" Now answer the following question:\n",
|
43 |
+
" \"\"\",\n",
|
44 |
+
")\n",
|
45 |
+
"hf_transformers_chatbot = Chatbot(hf_transformers_cfg)\n",
|
46 |
+
"\n",
|
47 |
+
"def chat(question, history):\n",
|
48 |
+
" history = history or []\n",
|
49 |
+
" answer = hf_transformers_chatbot.process_input(question)\n",
|
50 |
+
"\n",
|
51 |
+
" history.append((question, answer))\n",
|
52 |
+
" print(history)\n",
|
53 |
+
" return history, history\n",
|
54 |
+
"\n",
|
55 |
+
"\n",
|
56 |
+
"\n",
|
57 |
+
"block = gr.Blocks(css=\".gradio-container {background-color: lightgray}\")\n",
|
58 |
+
"\n",
|
59 |
+
"with block:\n",
|
60 |
+
" with gr.Row():\n",
|
61 |
+
" gr.Markdown(\"<h3><center>Buster π€: A Question-Answering Bot for Huggingface π€ Transformers </center></h3>\")\n",
|
62 |
+
"\n",
|
63 |
+
"\n",
|
64 |
+
" chatbot = gr.Chatbot()\n",
|
65 |
+
"\n",
|
66 |
+
" with gr.Row():\n",
|
67 |
+
" message = gr.Textbox(\n",
|
68 |
+
" label=\"What's your question?\",\n",
|
69 |
+
" placeholder=\"What kind of model should I use for sentiment analysis?\",\n",
|
70 |
+
" lines=1,\n",
|
71 |
+
" )\n",
|
72 |
+
" submit = gr.Button(value=\"Send\", variant=\"secondary\").style(full_width=False)\n",
|
73 |
+
"\n",
|
74 |
+
" gr.Examples(\n",
|
75 |
+
" examples=[\n",
|
76 |
+
" \"What kind of models should I use for images and text?\",\n",
|
77 |
+
" \"When should I finetune a model vs. training it form scratch?\",\n",
|
78 |
+
" \"How can I deploy my trained huggingface model?\",\n",
|
79 |
+
" \"Can you give me some python code to quickly finetune a model on my sentiment analysis dataset?\",\n",
|
80 |
+
" ],\n",
|
81 |
+
" inputs=message,\n",
|
82 |
+
" )\n",
|
83 |
+
"\n",
|
84 |
+
" gr.Markdown(\n",
|
85 |
+
" \"\"\"This simple application uses GPT to search the huggingface π€ transformers docs and answer questions.\n",
|
86 |
+
" For more info on huggingface transformers view the [full documentation.](https://huggingface.co/docs/transformers/index).\"\"\" \n",
|
87 |
+
" )\n",
|
88 |
+
"\n",
|
89 |
+
"\n",
|
90 |
+
" gr.HTML(\n",
|
91 |
+
" \"οΈ<center> Created with β€οΈ by @jerpint and @hadrienbertrand\"\n",
|
92 |
+
" )\n",
|
93 |
+
"\n",
|
94 |
+
" state = gr.State()\n",
|
95 |
+
" agent_state = gr.State()\n",
|
96 |
+
"\n",
|
97 |
+
" submit.click(chat, inputs=[message, state], outputs=[chatbot, state])\n",
|
98 |
+
" message.submit(chat, inputs=[message, state], outputs=[chatbot, state])\n",
|
99 |
+
"\n",
|
100 |
+
"\n",
|
101 |
+
"block.launch(debug=True)"
|
102 |
+
]
|
103 |
+
}
|
104 |
+
],
|
105 |
+
"metadata": {
|
106 |
+
"kernelspec": {
|
107 |
+
"display_name": "Python 3 (ipykernel)",
|
108 |
+
"language": "python",
|
109 |
+
"name": "python3"
|
110 |
+
},
|
111 |
+
"language_info": {
|
112 |
+
"codemirror_mode": {
|
113 |
+
"name": "ipython",
|
114 |
+
"version": 3
|
115 |
+
},
|
116 |
+
"file_extension": ".py",
|
117 |
+
"mimetype": "text/x-python",
|
118 |
+
"name": "python",
|
119 |
+
"nbconvert_exporter": "python",
|
120 |
+
"pygments_lexer": "ipython3",
|
121 |
+
"version": "3.9.12"
|
122 |
+
}
|
123 |
+
},
|
124 |
+
"nbformat": 4,
|
125 |
+
"nbformat_minor": 5
|
126 |
+
}
|
app.py β buster/apps/slackbot.py
RENAMED
@@ -15,7 +15,7 @@ PYTORCH_CHANNEL = "C04MEK6N882"
|
|
15 |
HF_TRANSFORMERS_CHANNEL = "C04NJNCJWHE"
|
16 |
|
17 |
mila_doc_cfg = ChatbotConfig(
|
18 |
-
documents_file="
|
19 |
unknown_prompt="This doesn't seem to be related to cluster usage.",
|
20 |
embedding_model="text-embedding-ada-002",
|
21 |
top_k=3,
|
@@ -51,7 +51,7 @@ mila_doc_cfg = ChatbotConfig(
|
|
51 |
mila_doc_chatbot = Chatbot(mila_doc_cfg)
|
52 |
|
53 |
orion_cfg = ChatbotConfig(
|
54 |
-
documents_file="
|
55 |
unknown_prompt="This doesn't seem to be related to the orion library. I am not sure how to answer.",
|
56 |
embedding_model="text-embedding-ada-002",
|
57 |
top_k=3,
|
@@ -84,7 +84,7 @@ orion_cfg = ChatbotConfig(
|
|
84 |
orion_chatbot = Chatbot(orion_cfg)
|
85 |
|
86 |
pytorch_cfg = ChatbotConfig(
|
87 |
-
documents_file="
|
88 |
unknown_prompt="This doesn't seem to be related to the pytorch library. I am not sure how to answer.",
|
89 |
embedding_model="text-embedding-ada-002",
|
90 |
top_k=3,
|
@@ -117,7 +117,7 @@ pytorch_cfg = ChatbotConfig(
|
|
117 |
pytorch_chatbot = Chatbot(pytorch_cfg)
|
118 |
|
119 |
hf_transformers_cfg = ChatbotConfig(
|
120 |
-
documents_file="
|
121 |
unknown_prompt="This doesn't seem to be related to the huggingface library. I am not sure how to answer.",
|
122 |
embedding_model="text-embedding-ada-002",
|
123 |
top_k=3,
|
|
|
15 |
HF_TRANSFORMERS_CHANNEL = "C04NJNCJWHE"
|
16 |
|
17 |
mila_doc_cfg = ChatbotConfig(
|
18 |
+
documents_file="../data/document_embeddings.csv",
|
19 |
unknown_prompt="This doesn't seem to be related to cluster usage.",
|
20 |
embedding_model="text-embedding-ada-002",
|
21 |
top_k=3,
|
|
|
51 |
mila_doc_chatbot = Chatbot(mila_doc_cfg)
|
52 |
|
53 |
orion_cfg = ChatbotConfig(
|
54 |
+
documents_file="../data/document_embeddings_orion.csv",
|
55 |
unknown_prompt="This doesn't seem to be related to the orion library. I am not sure how to answer.",
|
56 |
embedding_model="text-embedding-ada-002",
|
57 |
top_k=3,
|
|
|
84 |
orion_chatbot = Chatbot(orion_cfg)
|
85 |
|
86 |
pytorch_cfg = ChatbotConfig(
|
87 |
+
documents_file="../data/document_embeddings_pytorch.tar.gz",
|
88 |
unknown_prompt="This doesn't seem to be related to the pytorch library. I am not sure how to answer.",
|
89 |
embedding_model="text-embedding-ada-002",
|
90 |
top_k=3,
|
|
|
117 |
pytorch_chatbot = Chatbot(pytorch_cfg)
|
118 |
|
119 |
hf_transformers_cfg = ChatbotConfig(
|
120 |
+
documents_file="../data/document_embeddings_hf_transformers.tar.gz",
|
121 |
unknown_prompt="This doesn't seem to be related to the huggingface library. I am not sure how to answer.",
|
122 |
embedding_model="text-embedding-ada-002",
|
123 |
top_k=3,
|
buster/chatbot.py
CHANGED
@@ -128,12 +128,12 @@ class Chatbot:
|
|
128 |
names = matched_documents.name.to_list()
|
129 |
similarities = matched_documents.similarity.to_list()
|
130 |
|
131 |
-
response += f"{sep}{sep}Here are the sources I used to answer your question
|
132 |
for url, name, similarity in zip(urls, names, similarities):
|
133 |
if format == "markdown":
|
134 |
-
response += f"
|
135 |
elif format == "slack":
|
136 |
-
response += f"β’ <{url}|{name}>,
|
137 |
else:
|
138 |
raise ValueError(f"{format} is not a valid URL format.")
|
139 |
|
|
|
128 |
names = matched_documents.name.to_list()
|
129 |
similarities = matched_documents.similarity.to_list()
|
130 |
|
131 |
+
response += f"{sep}{sep}Here are the sources I used to answer your question:{sep}"
|
132 |
for url, name, similarity in zip(urls, names, similarities):
|
133 |
if format == "markdown":
|
134 |
+
response += f"[{name}]({url}), relevance: {similarity:2.3f}{sep}"
|
135 |
elif format == "slack":
|
136 |
+
response += f"β’ <{url}|{name}>, relevance: {similarity:2.3f}{sep}"
|
137 |
else:
|
138 |
raise ValueError(f"{format} is not a valid URL format.")
|
139 |
|