Back to the Future: Unsupervised Backprop-based Decoding for Counterfactual and Abductive Commonsense Reasoning

  title={Back to the Future: Unsupervised Backprop-based Decoding for Counterfactual and Abductive Commonsense Reasoning},
  author={Lianhui Qin and Vered Shwartz and Peter West and Chandra Bhagavatula and Jena D. Hwang and Ronan Le Bras and Antoine Bosselut and Yejin Choi},
Abductive and counterfactual reasoning, core abilities of everyday human cognition, require reasoning about what might have happened at time t, while conditioning on multiple contexts from the relative past and future. However, simultaneous incorporation of past and future contexts using generative language models (LMs) can be challenging, as they are trained either to condition only on the past context or to perform narrowly scoped text-infilling. In this paper, we propose DeLorean, a new… 

Figures and Tables from this paper

Conditional Generation of Temporally-ordered Event Sequences

A single model is proposed that addresses both temporal ordering, sorting given events into the order they occurred, and event infilling, predicting new events which fit into an existing temporally-ordered sequence.

GENIE: A Leaderboard for Human-in-the-Loop Evaluation of Text Generation

This work introduces GENIE, an extensible human evaluation leaderboard, which brings the ease of leaderboards to text generation tasks and provides formal granular evaluation metrics and identifies areas for future research.

Transferable Persona-Grounded Dialogues via Grounded Minimal Edits

Focusing on personas, Grounded Minimal Editor (GME), which learns to edit by disentangling and recombining persona-related and persona-agnostic parts of the response, is proposed, and experimental results show that GME outperforms competitive baselines by a large margin.

Computational Storytelling and Emotions: A Survey

Storytelling has always been vital for human nature. From ancient times, humans have used stories for several objectives including entertainment, advertisement, and education. Various analyses have

Achieving Conversational Goals with Unsupervised Post-hoc Knowledge Injection

This paper proposes a post-hoc knowledge-injection technique where a diverse set of relevant knowledge snippets are retrieved conditioned on both the dialog history and an initial response from an existing dialog model, and shows that knowledge-augmentation promotes success in achieving conversational goals in both experimental settings.

The Perils of Using Mechanical Turk to Evaluate Open-Ended Text Generation

It is shown that AMT worker judgments improve when they are shown model-generated output alongside human-generated references, which enables the workers to better calibrate their ratings, and interviews with the English teachers provide deeper insights into the challenges of the evaluation process.

Improving Automated Evaluation of Open Domain Dialog via Diverse Reference Augmentation

This work proposes a novel technique for automatically expanding a human generated reference to a set of candidate references, and fetch plausible references from knowledge sources, and adapt them so that they are more fluent in context of the dialog instance in question.

Coalescing Global and Local Information for Procedural Text Understanding

A new model that builds entity- and timestep-aware input representations (local input) considering the whole context (global input) and jointly model the entity states with a structured prediction objective (global output) is proposed, which optimizes for both precision and recall.

Composable Text Control Operations in Latent Space with Ordinary Differential Equations

A new efficient approach for composable text operations in the compact latent space of text based on ordinary differential equations given arbitrary plug-in operators, which permits diverse control operators acquired using any relevant data from different domains.

Can Language Models perform Abductive Commonsense Reasoning?

This report reviews over some of the methodologies that were at-tempted to solve theductive Reasoning challenge, re-implement the baseline models, and analyzesSome of the weaknesses that current approaches have.



Abductive Commonsense Reasoning

This study introduces a challenge dataset, ART, that consists of over 20k commonsense narrative contexts and 200k explanations, and conceptualizes two new tasks -- Abductive NLI: a multiple-choice question answering task for choosing the more likely explanation, and Abduction NLG: a conditional generation task for explaining given observations in natural language.

Cosmos QA: Machine Reading Comprehension with Contextual Commonsense Reasoning

This paper introduces Cosmos QA, a large-scale dataset of 35,600 problems that require commonsense-based reading comprehension, formulated as multiple-choice questions, and proposes a new architecture that improves over the competitive baselines.

ATOMIC: An Atlas of Machine Commonsense for If-Then Reasoning

Experimental results demonstrate that multitask models that incorporate the hierarchical structure of if-then relation types lead to more accurate inference compared to models trained in isolation, as measured by both automatic and human evaluation.

Counterfactual Story Reasoning and Generation

This paper proposes Counterfactual Story Rewriting: given an original story and an intervening counterfactual event, the task is to minimally revise the story to make it compatible with the given counterfactually event.

BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

A new language representation model, BERT, designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers, which can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks.

Bidirectional Beam Search: Forward-Backward Inference in Neural Sequence Models for Fill-in-the-Blank Image Captioning

This work develops the first approximate inference algorithm for 1-Best (and M-Best) decoding in bidirectional neural sequence models by extending Beam Search to reason about both forward and backward time dependencies, and introduces a novel Fill-in-the-Blank Image Captioning task which requires reasoning about both past and future sentence structure to reconstruct sensible image descriptions.

Improving a Neural Semantic Parser by Counterfactual Learning from Human Bandit Feedback

This work is the first to show that semantic parsers can be improved significantly by counterfactual learning from logged human feedback data, and devise an easy-to-use interface to collect human feedback on semantic parses.

Interpretation as Abduction

An approach to abductive inference, called “weighted abduction”, that has resulted in a significant simplification of how the problem of interpreting texts is conceptualized, can be combined with the older view of “parsing as deduction” to produce an elegant and thorough integration of syntax, semantics, and pragmatics.

Conversing by Reading: Contentful Neural Conversation with On-demand Machine Reading

A new end-to-end approach to contentful neural conversation that jointly models response generation and on-demand machine reading is presented, allowing for more focused integration of external knowledge than has been possible in prior approaches.

What Happens Next? Event Prediction Using a Compositional Neural Network Model

It is found that recent work on learning vector-space embeddings to capture word meaning can be effectively applied to this task, including simple incorporation of a verb's arguments in the representation by vector addition.