diff --git a/README.md b/README.md
index 4f20e80a64732f56211872f43bf31653ce761de5..004a5ca4a609644dcb91241c7fdd3b2168e11e62 100755
--- a/README.md
+++ b/README.md
@@ -20,7 +20,7 @@
## Installation
-You can install ConvLab-3 in one of the following ways according to your need. Higher versions of `torch` and `transformers` may also work.
+You can install ConvLab-3 in one of the following ways according to your need. We use `torch>=1.10.1,<=1.13` and `transformers>=4.17.0,<=4.24.0`. Higher versions of `torch` and `transformers` may also work.
### Git clone and pip install in development mode (Recommend)
@@ -65,6 +65,7 @@ docker exec -it CONTAINER_ID bash
## Tutorials
+- [Getting Started](https://github.com/ConvLab/ConvLab-3/blob/master/tutorials/Getting_Started.ipynb) (Have a try on [Colab](https://colab.research.google.com/github/ConvLab/ConvLab-3/blob/master/tutorials/Getting_Started.ipynb)!)
- [Introduction to Unified Data Format](https://github.com/ConvLab/ConvLab-3/tree/master/data/unified_datasets)
- [Utility functions for unified datasets](https://github.com/ConvLab/ConvLab-3/blob/master/convlab/util/unified_datasets_util.py)
- [RL Toolkit](https://github.com/ConvLab/ConvLab-3/tree/master/convlab/policy)
diff --git a/tutorials/Getting_Started.ipynb b/tutorials/Getting_Started.ipynb
index a4b28edc9fbbc06e7ac354b84d9948fb04f0d469..c672d92353a4da6729814ead1d1c8494b6f005f4 100644
--- a/tutorials/Getting_Started.ipynb
+++ b/tutorials/Getting_Started.ipynb
@@ -1,31 +1,16 @@
{
- "nbformat": 4,
- "nbformat_minor": 0,
- "metadata": {
- "colab": {
- "name": "Tutorial",
- "provenance": [],
- "collapsed_sections": [],
- "toc_visible": true
- },
- "kernelspec": {
- "name": "python3",
- "display_name": "Python 3",
- "language": "python"
- }
- },
"cells": [
{
"cell_type": "markdown",
"metadata": {
- "id": "x3qZ8fi1FAWS",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "x3qZ8fi1FAWS"
},
"source": [
"# Getting Started\n",
"\n",
"In this tutorial, you will know how to\n",
- "- use the models in **ConvLab-2** to build a dialog agent.\n",
+ "- use the models in **ConvLab-3** to build a dialog agent.\n",
"- build a simulator to chat with the agent and evaluate the performance.\n",
"- try different module combinations.\n",
"- use analysis tool to diagnose your system.\n",
@@ -36,27 +21,27 @@
{
"cell_type": "markdown",
"metadata": {
- "id": "NvXG3UyBFFS3",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "NvXG3UyBFFS3"
},
"source": [
"## Environment setup\n",
- "Run the command below to install ConvLab-2. Then restart the notebook and skip this commend."
+ "Run the command below to install ConvLab-3. Then restart the notebook and skip this commend."
]
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "qPqTs6EP0wxO",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "qPqTs6EP0wxO"
},
+ "outputs": [],
"source": [
- "# first install ConvLab-2 and restart the notebook\n",
- "! git clone https://github.com/thu-coai/ConvLab-2.git && cd ConvLab-2 && pip install -e ."
- ],
- "execution_count": null,
- "outputs": []
+ "# first install ConvLab-3 and restart the notebook\n",
+ "! git clone --depth 1 https://github.com/ConvLab/ConvLab-3.git && cd ConvLab-3 && pip install -e ."
+ ]
},
{
"cell_type": "code",
@@ -68,18 +53,11 @@
"!python -m spacy download en_core_web_sm"
]
},
- {
- "cell_type": "code",
- "execution_count": null,
- "metadata": {},
- "outputs": [],
- "source": []
- },
{
"cell_type": "markdown",
"metadata": {
- "id": "1TPEtcEpFiCU",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "1TPEtcEpFiCU"
},
"source": [
"## build an agent\n",
@@ -91,11 +69,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "OqzFlmCBCXEV",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "OqzFlmCBCXEV"
},
+ "outputs": [],
"source": [
"# common import: convlab.$module.$model.$dataset\n",
"from convlab.nlu.jointBERT.multiwoz import BERTNLU\n",
@@ -109,15 +89,13 @@
"import random\n",
"import numpy as np\n",
"import torch"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "N-18Q6YKGEzY",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "N-18Q6YKGEzY"
},
"source": [
"Then, create the models and build an agent:"
@@ -125,11 +103,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "-GPQgUD6GjHv",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "-GPQgUD6GjHv"
},
+ "outputs": [],
"source": [
"# go to README.md of each model for more information\n",
"# BERT nlu\n",
@@ -142,15 +122,13 @@
"sys_nlg = TemplateNLG(is_user=False)\n",
"# assemble\n",
"sys_agent = PipelineAgent(sys_nlu, sys_dst, sys_policy, sys_nlg, name='sys')"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "FvqIiO22H0Od",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "FvqIiO22H0Od"
},
"source": [
"That's all! Let's chat with the agent using its response function:"
@@ -158,100 +136,100 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "9LYnDLysH1nX",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "9LYnDLysH1nX"
},
+ "outputs": [],
"source": [
"sys_agent.response(\"I want to find a moderate hotel\")"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "dAvi-IzsH474",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "dAvi-IzsH474"
},
+ "outputs": [],
"source": [
"sys_agent.response(\"Which type of hotel is it ?\")"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "f0A6Pt4tH60c",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "f0A6Pt4tH60c"
},
+ "outputs": [],
"source": [
"sys_agent.response(\"OK , where is its address ?\")"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "8VH6yO6PH8Ua",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "8VH6yO6PH8Ua"
},
+ "outputs": [],
"source": [
"sys_agent.response(\"Thank you !\")"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "wto-PxgNH9gp",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "wto-PxgNH9gp"
},
+ "outputs": [],
"source": [
"sys_agent.response(\"Try to find me a Chinese restaurant in south area .\")"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "MJ9C1ZhbH_JY",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "MJ9C1ZhbH_JY"
},
+ "outputs": [],
"source": [
"sys_agent.response(\"Which kind of food it provides ?\")"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "DpaQNA_HIAmT",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "DpaQNA_HIAmT"
},
+ "outputs": [],
"source": [
"sys_agent.response(\"Book a table for 5 , this Sunday .\")"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "n6uuuRonIHvW",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "n6uuuRonIHvW"
},
"source": [
"## Build a simulator to chat with the agent and evaluate\n",
@@ -263,11 +241,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "pAMAJZSF7D5w",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "pAMAJZSF7D5w"
},
+ "outputs": [],
"source": [
"# MILU\n",
"user_nlu = MILU()\n",
@@ -279,15 +259,13 @@
"user_nlg = TemplateNLG(is_user=True)\n",
"# assemble\n",
"user_agent = PipelineAgent(user_nlu, user_dst, user_policy, user_nlg, name='user')"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "Rghl-V2AJhRY",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "Rghl-V2AJhRY"
},
"source": [
"\n",
@@ -298,23 +276,23 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "saUoLUUGJqDa",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "saUoLUUGJqDa"
},
+ "outputs": [],
"source": [
"evaluator = MultiWozEvaluator()\n",
"sess = BiSession(sys_agent=sys_agent, user_agent=user_agent, kb_query=None, evaluator=evaluator)"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "kevGJZhFJzTU",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "kevGJZhFJzTU"
},
"source": [
"Let's make this two agents chat! The key is `next_turn` method of `BiSession` class."
@@ -322,11 +300,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "FIV_qkE49LzE",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "FIV_qkE49LzE"
},
+ "outputs": [],
"source": [
"def set_seed(r_seed):\n",
" random.seed(r_seed)\n",
@@ -354,15 +334,13 @@
"print('final goal:')\n",
"pprint(sess.evaluator.goal)\n",
"print('='*100)"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "CKOQs1l8LpTR",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "CKOQs1l8LpTR"
},
"source": [
"## Try different module combinations\n",
@@ -383,11 +361,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "G-9G0VYUNYOI",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "G-9G0VYUNYOI"
},
+ "outputs": [],
"source": [
"# available NLU models\n",
"from convlab.nlu.svm.multiwoz import SVMNLU\n",
@@ -413,15 +393,13 @@
"# available E2E models\n",
"from convlab.e2e.sequicity.multiwoz import Sequicity\n",
"from convlab.e2e.damd.multiwoz import Damd"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "6TS2_Tp1Nzvq",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "6TS2_Tp1Nzvq"
},
"source": [
"NLU+RuleDST or Word-DST:"
@@ -429,11 +407,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "DZMk9wAlONrz",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "DZMk9wAlONrz"
},
+ "outputs": [],
"source": [
"# NLU+RuleDST:\n",
"sys_nlu = BERTNLU()\n",
@@ -445,15 +425,13 @@
"# sys_nlu = None\n",
"# sys_dst = SUMBT()\n",
"# sys_dst = TRADE()"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "gUUYsDMJPJRl",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "gUUYsDMJPJRl"
},
"source": [
"Policy+NLG or Word-Policy:"
@@ -461,11 +439,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "PTJ95x9UPHt4",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "PTJ95x9UPHt4"
},
+ "outputs": [],
"source": [
"# Policy+NLG:\n",
"sys_policy = RulePolicy()\n",
@@ -481,15 +461,13 @@
"# sys_policy = HDSA()\n",
"# sys_policy = MDRGWordPolicy()\n",
"# sys_nlg = None"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "s9lGIv0oPupn",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "s9lGIv0oPupn"
},
"source": [
"Assemble the Pipeline system agent:"
@@ -497,22 +475,22 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "IvLx7HUkPyZ5",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "IvLx7HUkPyZ5"
},
+ "outputs": [],
"source": [
"sys_agent = PipelineAgent(sys_nlu, sys_dst, sys_policy, sys_nlg, 'sys')"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "hR4A8WbZP2lc",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "hR4A8WbZP2lc"
},
"source": [
"Or Directly use an end-to-end model:"
@@ -520,23 +498,23 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "8VdUmcxoP6ej",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "8VdUmcxoP6ej"
},
+ "outputs": [],
"source": [
"# sys_agent = Sequicity()\n",
"# sys_agent = Damd()"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "_v-eoBtnP9J9",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "_v-eoBtnP9J9"
},
"source": [
"Config an user agent similarly:"
@@ -544,11 +522,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "UkHpjvR5QezN",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "UkHpjvR5QezN"
},
+ "outputs": [],
"source": [
"user_nlu = BERTNLU()\n",
"# user_nlu = MILU()\n",
@@ -559,15 +539,13 @@
"user_nlg = TemplateNLG(is_user=True)\n",
"# user_nlg = SCLSTM(is_user=True)\n",
"user_agent = PipelineAgent(user_nlu, user_dst, user_policy, user_nlg, name='user')"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "VJTBxEDhSAqc",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "VJTBxEDhSAqc"
},
"source": [
"## Use analysis tool to diagnose the system\n",
@@ -578,11 +556,13 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "Hu2q3lQiSMDy",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "Hu2q3lQiSMDy"
},
+ "outputs": [],
"source": [
"from convlab.util.analysis_tool.analyzer import Analyzer\n",
"\n",
@@ -591,15 +571,13 @@
"\n",
"set_seed(20200131)\n",
"analyzer.comprehensive_analyze(sys_agent=sys_agent, model_name='sys_agent', total_dialog=100)"
- ],
- "execution_count": null,
- "outputs": []
+ ]
},
{
"cell_type": "markdown",
"metadata": {
- "id": "AB-mDm0plQWd",
- "colab_type": "text"
+ "colab_type": "text",
+ "id": "AB-mDm0plQWd"
},
"source": [
"To compare several models:"
@@ -607,17 +585,41 @@
},
{
"cell_type": "code",
+ "execution_count": null,
"metadata": {
- "id": "GKe_DNJUlWzh",
+ "colab": {},
"colab_type": "code",
- "colab": {}
+ "id": "GKe_DNJUlWzh"
},
+ "outputs": [],
"source": [
"set_seed(20200131)\n",
"analyzer.compare_models(agent_list=[sys_agent1, sys_agent2], model_name=['sys_agent1', 'sys_agent2'], total_dialog=100)"
- ],
- "execution_count": null,
- "outputs": []
+ ]
+ }
+ ],
+ "metadata": {
+ "colab": {
+ "collapsed_sections": [],
+ "name": "Tutorial",
+ "provenance": [],
+ "toc_visible": true
+ },
+ "kernelspec": {
+ "display_name": "Python 3.8.12 ('py38')",
+ "language": "python",
+ "name": "python3"
+ },
+ "language_info": {
+ "name": "python",
+ "version": "3.8.12"
+ },
+ "vscode": {
+ "interpreter": {
+ "hash": "0f9333403d680bc010aa5ce5a2f27ba398c9e47e92ba3724506306aa234cd07d"
+ }
}
- ]
-}
\ No newline at end of file
+ },
+ "nbformat": 4,
+ "nbformat_minor": 0
+}