Few-Shot Bot: Prompt-Based Learning for Dialogue Systems

This repository includes the dataset, experiments results, and code for the paper:

Few-Shot Bot: Prompt-Based Learning for Dialogue Systems PDF.

Authors: Andrea Madotto, Zhaojiang Lin, Genta Indra Winata, Pascale Fung

Abstract

Learning to converse using only a few examples is a grand challenge in Conversational AI. The current best conversational models, which are either good chit-chatters (e.g., BlenderBot) or goal-oriented systems (e.g., MinTL), are language models (LMs) fine-tuned on large conversational datasets. Training these models is expensive, both in terms of computational resources and time, and it is hard to keep these models up to date with new conversational skills. A simple yet unexplored solution is prompt-based few-shot learning (Brown et al. 2020) which does not require gradient-based fine-tuning but instead uses a few examples in the LM context as the only source of learning. In this paper, we explore prompt-based few-shot learning in dialogue tasks. We benchmark LMs of different sizes in 9 response generation tasks, which include a variety of knowledge-grounded tasks, task-oriented generations, general open-chat, and controlled stylistic generation, and 5 conversational parsing tasks, which include dialogue state tracking, graph path generation, persona information extraction, and document retrieval. The current largest, released, LM (GPT-J-6B) achieves competitive performance to full-training state-of-the-art models by using the prompt-based few-shot learning, thus no training. Moreover, we proposed a novel perplexity-based classifier, that also does not require any fine-tuning, to select the most appropriate prompt given a dialogue history, as to create an all-in-one model with multiple dialogue skills. Finally, by combining the power of prompt-based few-shot learning and the skill selector, we create an end-to-end chatbot named the Few-Shot Bot, which automatically selects the most appropriate conversational skill, queries different KBs or the internet, and uses it to generate a human-like response, all by using only one dialogue example per skill.

Installation

In this repo, we load all the validation and test sets used in the evaluation. For running the experiments and the demo, you should install the following requirements:

pip install -r requirements.txt

Basic Running

Reproducing the results and plots

The generation folder stores the generated responses of the experiments in all datasets. To generate the tables and the plots in the paper, run

python generate_plots_tables.py

This script loads all the files and computes the mean between different runs and it generates the plots. Note that this script is very custum for each datasets, but it can serve as guide line for future extentions.

Running the experiments

There are three main files to run 1) response generation (main_response_generation.py), 2) conversational parsing (main_conversational_parsing.py), and 3) skill-selector (main_skill_selector.py). In these files, we load the necessary prompt (load_prefix) and we run the generation (generate_response) for each sample in the test set. Since each dialogue skill require a different template, as shown in the paper, we create a function that converts structured data into the correct shot prompt. An example of this function can be found in prompts/persona_chat.py, and in generic_prompts.py we store the generation functions.

In each main file there is configuration object (mapper) which specify meta-information about the task (i.e., number of shots, generation length, decoding type, prompt converter). Expecially for conversational parsing, there are different decoding type. For example, in MWOZ the model generates the dialogue state, which is further looped into the next turn.

How to run?

For example, to run the persona chat experiments (0, 1, k-shots), you can use the following command:

python main_response_generation.py --model_checkpoint EleutherAI/gpt-j-6B --dataset persona --gpu 0

In case your GPU has less that 16GB, then you could add --multigpu to spawn 4 GPUs (e.g., 1080Ti) and do inference in parallel. Similarly, for conversational parsing tasks, you could use:

python main_conversational_parsing.py --model_checkpoint EleutherAI/gpt-j-6B --dataset wow-parse --gpu 0

Notice that some parsing task requires a knowledge base (e.g., dialKG-parse requires the KG in neo4j). Finally, to run the skill-selector task, you could use:

python main_skill_selector.py --model_checkpoint EleutherAI/gpt-j-6B --shots_k 6 --repetition 1 --gpu 0

where repetition is the seed for selecting random samples in the prompts.

Runners

In the runners folder, we provide a rudimental runner to run all the experiments and reproduce the results in the paper.

Few-Shot Bot

There are two modes for the FSB such as 1) controlled style generation and 2) full-model. Currently we support the controlled style generation model. Check the FSB-CG.ipynb to try to interact with FSB in your local machine, or try directly in colab at https://colab.research.google.com/drive/15hQv1V3Cs5kQVfLOE_FZc1VCWQ3YpWVd?usp=sharing (Remeber to select the enviroment with GPU).