Papers by Jason Weston
Queens are Powerful too: Mitigating Gender Bias in Dialogue Generation (2020.emnlp-main)
Copied to clipboard
| Challenge: | Social biases present in data are often directly reflected in the predictions of models trained on that data. |
| Approach: | They analyze gender bias in dialogue data and propose techniques to mitigate it . they use counterfactual data augmentation, targeted data collection, and bias controlled training . |
| Outcome: | The proposed techniques mitigate gender bias by balancing genderedness of generated dialogue utterances. |
Multi-Dimensional Gender Bias Classification (2020.emnlp-main)
Copied to clipboard
| Challenge: | a novel framework decomposes gender bias in text along several pragmatic and semantic dimensions . language is a primary means by which people communicate, express identities and categorize themselves . unwanted gender biases can affect downstream applications, leading to poor user experiences . |
| Approach: | They propose a framework that decomposes gender bias in text along several dimensions . they annotate eight large scale datasets with gender information and collect a benchmark . |
| Outcome: | The proposed framework decomposes gender bias in text along several pragmatic and semantic dimensions. |
Am I Me or You? State-of-the-Art Dialogue Models Cannot Maintain an Identity (2022.findings-naacl)
Copied to clipboard
| Challenge: | State-of-the-art open-domain dialogue models fail to maintain character identity throughout discourse . despite improvements in accuracy and self-contradiction, agents take on the role of interlocutor . |
| Approach: | They formalize and quantify the deficiency in character identity modeling by using human evaluations. |
| Outcome: | The proposed models reduce mistaken identity issues by nearly 65% according to human annotators while improving engagingness. |
Beyond Goldfish Memory: Long-Term Open-Domain Conversation (2022.acl-long)
Copied to clipboard
| Challenge: | Despite recent improvements in open-domain dialogue models, state of the art models are trained and evaluated on short conversations with little context. |
| Approach: | They propose to use retrieval-augmented methods to summarize and recall past conversations to improve their models. |
| Outcome: | The proposed models outperform the current state-of-the-art models on human-human chat sessions in both automatic and human evaluations. |
Chain-of-Verification Reduces Hallucination in Large Language Models (2024.findings-acl)
Copied to clipboard
Shehzaad Dhuliawala, Mojtaba Komeili, Jing Xu, Roberta Raileanu, Xian Li, Asli Celikyilmaz, Jason Weston
| Challenge: | Large language models can generate plausible but incorrect factual information, termed hallucinations, but they can still fail on lesser known facts. |
| Approach: | They develop a method that allows language models to deliberate on the responses they give in order to correct their errors. |
| Outcome: | The proposed method decreases hallucinations across a variety of tasks, including list-based questions, closed book MultiSpanQA and longform text generation. |
How to Motivate Your Dragon: Teaching Goal-Driven Agents to Speak and Act in Fantasy Worlds (2021.naacl-main)
Copied to clipboard
| Challenge: | a recent improvement in the quality of natural language processing and generation (NLG) is needed for goal-oriented ML driven agents. |
| Approach: | They propose a reinforcement learning system that integrates large-scale language modeling and commonsense reasoning-based pre-training to imbue the agent with relevant priors. |
| Outcome: | The proposed system is able to act and talk naturally with respect to their motivations. |
Recipes for Building an Open-Domain Chatbot (2021.eacl-main)
Copied to clipboard
Stephen Roller, Emily Dinan, Naman Goyal, Da Ju, Mary Williamson, Yinhan Liu, Jing Xu, Myle Ott, Eric Michael Smith, Y-Lan Boureau, Jason Weston
| Challenge: | Existing work shows that scaling models in the number of parameters and the size of the data they are trained on gives improved results, but other factors are important. |
| Approach: | They propose to build open-domain chatbots that can be scaled to improve their performance . they use a blend of cognitive and cognitive skills to build a model that combines these skills . |
| Outcome: | The proposed models outperform existing approaches in multi-turn dialogue on engagingness and humanness measurements. |
Image-Chat: Engaging Grounded Conversations (2020.acl-main)
Copied to clipboard
| Challenge: | In order for machines to communicate with humans, they must understand the natural things that humans say about the world they live in and respond in kind. |
| Approach: | They propose to fuse a set of neural architectures using image and text representations to achieve this goal. |
| Outcome: | The proposed model performs well on the Image-Chat task and humans prefer it 47.7% of the time. |
The CRINGE Loss: Learning what language not to model (2023.acl-long)
Copied to clipboard
| Challenge: | Standard language model training uses gold human documents or human-human interaction data and treats all training data as positive examples. |
| Approach: | They propose a procedure to train with negative examples using the "CRINGE" loss technique and use it to train models with such data. |
| Outcome: | The proposed procedure outperforms multiple strong baselines and is simple to train and implement. |
Multi-Modal Open-Domain Dialogue (2021.emnlp-main)
Copied to clipboard
| Challenge: | Recent work in open-domain conversational agents has demonstrated that significant improvements in humanness and user preference can be achieved via massive scaling in both pre-training data and model size. |
| Approach: | They combine open-domain dialogue agents with vision models to investigate human preferences and humanness. |
| Outcome: | The proposed model outperforms existing models in multi-modal dialogue while performing as well as its predecessor (text-only) BlenderBot. |
Reason first, then respond: Modular Generation for Knowledge-infused Dialogue (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Large language models can produce fluent dialogue but often hallucinate factual inaccuracies. |
| Approach: | They propose a modular model for incorporating knowledge into conversational agents that generates a knowledge sequence and then attends to its own generated knowledge sequence. |
| Outcome: | The proposed model hallucinates less in knowledge-grounded dialogue tasks and has advantages in terms of interpretability and modularity. |
The ART of LLM Refinement: Ask, Refine, and Trust (2024.naacl-long)
Copied to clipboard
Kumar Shridhar, Koustuv Sinha, Andrew Cohen, Tianlu Wang, Ping Yu, Ramakanth Pasunuru, Mrinmaya Sachan, Jason Weston, Asli Celikyilmaz
| Challenge: | Large Language Models (LLMs) have demonstrated remarkable generative abilities, but can they judge the quality of their own generations and self-improve? |
| Approach: | They propose a reasoning with a refinement strategy called *ART: Ask, Refine, and Trust* that asks necessary questions to decide when an LLM should refine its output and uses it to affirm or deny trust. |
| Outcome: | The proposed reasoning with a refinement strategy achieves a performance gain of +5 points over baselines on two multistep reasoning tasks. |
System-Level Natural Language Feedback (2024.eacl-long)
Copied to clipboard
| Challenge: | Existing studies on NL feedback focus on instance-level approaches to refine specific examples, but we present a framework for system-level use of NL. |
| Approach: | They propose a framework for system-level use of natural language feedback . they use feedback to formalize system-design decisions in a human-in-the-loop-process . |
| Outcome: | The proposed framework improves search query and dialog response generation and human written instance-level feedback brings further gains over GPT-3.5 written feedback. |
Don’t Say That! Making Inconsistent Dialogue Unlikely with Unlikelihood Training (2020.acl-main)
Copied to clipboard
| Challenge: | Unlikelihood is a technique developed for removal of repetition in language model completions . it allows for a model to be generalized to solve a number of problems . |
| Approach: | They extend the unlikelihood objective to generate generations that contain repetitions . they show that such an objective can be used to improve logical consistency . |
| Outcome: | The proposed approach can be applied to a number of dialogue tasks. |
Adversarial NLI: A New Benchmark for Natural Language Understanding (2020.acl-main)
Copied to clipboard
| Challenge: | a new large-scale NLI benchmark dataset is presented to test models on a variety of popular NLIs. |
| Approach: | They propose a large-scale NLI benchmark dataset that is iteratively compared with a human-and-model-in-the-loop procedure. |
| Outcome: | The proposed method can be applied in a never-ending learning scenario, becoming a moving target for NLU, rather than a static benchmark that will quickly saturate. |
Recommendation as a Communication Game: Self-Supervised Bot-Play for Goal-oriented Dialogue (D19-1)
Copied to clipboard
| Challenge: | Traditional recommendation systems produce static rather than interactive recommendations invariant to a user’s specific requests, clarifications, or current mood. |
| Approach: | They use a goal-driven recommendation dialogue dataset to develop an end-to-end dialogue system that can simultaneously converse and recommend. |
| Outcome: | The proposed system can converse and recommend movies to humans without considering the task goal itself. |
Internet-Augmented Dialogue Generation (2022.acl-long)
Copied to clipboard
| Challenge: | Large language models are known to hallucinate facts when generating dialogue, and are unable to encode the knowledge in the model at the point of training. |
| Approach: | They propose an approach that generates an internet search query based on the context and conditions on the results to generate a response. |
| Outcome: | The proposed model generates an internet search query and then conditions on the results to generate a response. |
Dialogue in the Wild: Learning from a Deployed Role-Playing Game with Humans and Bots (2021.findings-acl)
Copied to clipboard
| Challenge: | prevailing paradigm in natural language processing research is to build a fixed dataset and freeze it, without any ability for the model to interact with humans using language at training time at all. |
| Approach: | They build and deploy a role-playing game where players converse with learning agents situated in an open-domain fantasy world. |
| Outcome: | The proposed game enables human players to learn from human conversations and improves on their models. |
Finding Generalizable Evidence by Learning to Convince Q&A Models (D19-1)
Copied to clipboard
| Challenge: | a system that finds the strongest supporting evidence for a given answer is proposed . a study using passage-based question-answering (QA) shows that agents select evidence that generalizes . |
| Approach: | They propose a system that finds the strongest supporting evidence for a given answer . they use passage-based question-answering (QA) as a testbed to train evidence agents . |
| Outcome: | The proposed system improves QA in a robust manner by using agent-selected evidence. |
What makes a good conversation? How controllable attributes affect human judgments (N19-1)
Copied to clipboard
| Challenge: | Existing work on dialogue models for conversational quality is incompletely understanding the relationship between quality and individual attributes. |
| Approach: | They propose to use conditional training and weighted decoding to control four attributes for chit-chat dialogue: repetition, specificity, response-relatedness and question-asking. |
| Outcome: | The proposed methods improve human quality judgments by controlling combinations of these variables. |
The Dialogue Dodecathlon: Open-Domain Knowledge and Image Grounded Conversational Agents (2020.acl-main)
Copied to clipboard
| Challenge: | a set of 12 tasks that measure if a conversational agent can communicate engagingly with personality and empathy, ask questions, answer questions by utilizing knowledge resources, and perceive and converse about images. |
| Approach: | They propose a set of 12 tasks that measure if a conversational agent can communicate engagingly with personality and empathy . they use large dialogue datasets to multi-task and obtain state-of-the-art results . |
| Outcome: | The proposed model improves over a BERT pre-trained model on large dialogue datasets and provides state-of-the-art results on many of the tasks. |
Can You Put it All Together: Evaluating Conversational Agents’ Ability to Blend Skills (2020.acl-main)
Copied to clipboard
| Challenge: | Existing work has focused on learning specific qualities of conversational agents, but it remains unclear how to combine them. |
| Approach: | They propose to combine models trained towards isolated capabilities with multi-task training to improve conversation performance. |
| Outcome: | The proposed dataset compares models trained towards isolated capabilities with models trained on a single skill. |
TOOLVERIFIER: Generalization to New Tools via Self-Verification (2024.findings-emnlp)
Copied to clipboard
Dheeraj Mekala, Jason Weston, Jack Lanchantin, Roberta Raileanu, Maria Lomeli, Jingbo Shang, Jane Dwivedi-Yu
| Challenge: | Existing tools and APIs present a challenge for generalization, despite frequent parameter updates and the daily introduction of new tools. |
| Approach: | They propose a method which distinguishes between close candidates by self-asking contrastive questions during tool selection and parameter generation. |
| Outcome: | Experiments on 4 tasks from the ToolBench benchmark show an improvement of 22% over few-shot baselines. |
Retrieval Augmentation Reduces Hallucination in Conversation (2021.findings-emnlp)
Copied to clipboard
| Challenge: | State-of-the-art dialogue models suffer from factual incorrectness and hallucination of knowledge. |
| Approach: | They propose to use neural-retrieval-in-the-loop architectures to optimize knowledge-grounded dialogue by retrieving, ranking, and encoder-decoders. |
| Outcome: | The proposed architectures exhibit open-domain conversational capabilities and generalize effectively to scenarios not within the training data. |
Learning to Speak and Act in a Fantasy Text Adventure Game (D19-1)
Copied to clipboard
Jack Urbanek, Angela Fan, Siddharth Karamcheti, Saachi Jain, Samuel Humeau, Emily Dinan, Tim Rocktäschel, Douwe Kiela, Arthur Szlam, Jason Weston
| Challenge: | Existing studies on grounded dialogue use only statistical regularities of text data, without explicit understanding of the world that the text describes. |
| Approach: | They propose a large-scale crowdsourced text adventure game as a research platform for studying grounded dialogue. |
| Outcome: | The proposed game allows agents to perceive, emote, and act whilst conducting dialogue with other agents. |
I like fish, especially dolphins: Addressing Contradictions in Dialogue Modeling (2021.acl-long)
Copied to clipboard
| Challenge: | Recent advances on neural approaches to natural language processing have triggered a resurgent interest on building intelligent open-domain chatbots. |
| Approach: | They propose a dialoguE COntradiction DEtection task and a conversational dataset . they show that their best contradiction detection model correlates well with human judgments . |
| Outcome: | The proposed model is more robust and generalizes well on analysis and out-of-distribution dialogues than standard (unstructured) Transformer models that explicitly hinge on utterance structures are more robust, the study shows . |
Build it Break it Fix it for Dialogue Safety: Robustness from Adversarial Human Attack (D19-1)
Copied to clipboard
| Challenge: | Detecting offensive language in the context of a dialogue is an increasingly important application of natural language processing. |
| Approach: | They propose to train a model to be robust to such attacks by iterative build it, break it, fix it scheme with humans and models in the loop. |
| Outcome: | The proposed model is significantly more robust to such human attacks than previous systems. |
When Life Gives You Lemons, Make Cherryade: Converting Feedback from Bad Responses into Good Labels (2024.naacl-long)
Copied to clipboard
| Challenge: | Existing dialogue models are primarily trained on human-human conversations . thumb ups/downs and gold corrections are often sparse in real-life deployment settings . |
| Approach: | They propose a framework to make use of binary and free-form textual human feedback. |
| Outcome: | The proposed framework improves the final dialogue model by using model-corrected replies. |
Language Models that Seek for Knowledge: Modular Search & Generation for Dialogue and Prompt Completion (2022.findings-emnlp)
Copied to clipboard
| Challenge: | Language models (LMs) have been shown to generate more factual responses by employing modularity in combination with retrieval. |
| Approach: | They extend the recent approach of Adolphs et al. (2021) to include internet search as a module. |
| Outcome: | The proposed method outperforms the state-of-the-art model BlenderBot 2 on open-domain knowledge-grounded conversations for the same number of parameters. |
Branch-Solve-Merge Improves Large Language Model Evaluation and Generation (2024.naacl-long)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) are often used for multi-faceted language generation and evaluation tasks that require complex user constraints or taking into account multiple aspects and criteria. |
| Approach: | They propose a Large Language Model program that consists of branch, solve, and merge modules that are parameterized with specific prompts to the base LLM. |
| Outcome: | The proposed program improves the evaluation correctness and consistency for each LLM by up to 26%, reducing length and pairwise position biases by up 50%, and allowing LLaMA-2-chat to match or outperform GPT-4 on most domains. |
Better Alignment with Instruction Back-and-Forth Translation (2024.findings-emnlp)
Copied to clipboard
| Challenge: | et al., 2023) proposes a method to improve instruction-tuning data . e.g., we generate synthetic instructions using the backtranslation approach . |
| Approach: | They propose a method to improve instruction-tuning data using web-based inputs . they generate synthetic instructions using the backtranslation approach and filter the generated data . |
| Outcome: | The proposed method improves the quality of instruction-tuning data based on preprocessed texts . it yields better AlpacaEval win rates than direct distillation . |
Bot-Adversarial Dialogue for Safe Conversational Agents (2021.naacl-main)
Copied to clipboard
| Challenge: | a new method for evaluating chatbot safety is proposed to mimic human-generated data . a bot-adversarial dialogue model learns undesirable features from this data, a study finds . |
| Approach: | They propose a human-and-model-in-the-loop framework for evaluating toxicity of chatbots . they propose two methods for safe conversational agents by either training on data or ”baking-in” safety to the generative model itself. |
| Outcome: | The proposed methods are safer than existing models while maintaining usability metrics, the authors say . they show that the proposed methods can be used to make safer models with human-model interactions . |
Learning from Dialogue after Deployment: Feed Yourself, Chatbot! (P19-1)
Copied to clipboard
| Challenge: | a majority of conversations a dialogue agent sees over its lifetime occur after it has already been trained and deployed, leaving a vast store of potential training signal untapped. |
| Approach: | They propose a self-feeding chatbot that extracts new training examples from conversations it participates in. |
| Outcome: | The proposed chatbot extracts training examples from conversations it participates in and predicts user satisfaction in its responses. |
Leveraging Implicit Feedback from Deployment Data in Dialogue (2024.eacl-short)
Copied to clipboard
| Challenge: | Xu et al., 2023) and Bai ed., 2019) use crowdworkers to collect signals from natural dialogue episodes. |
| Approach: | They use the publicly released BlenderBot deployment data to extract signals from conversations to implicitly measure the quality of a machine-generated utterance. |
| Outcome: | The proposed model improves over baseline models, but some proxy signals can lead to undesirable generations. |
Personalizing Dialogue Agents: I have a dog, do you have pets too? (P18-1)
Copied to clipboard
| Challenge: | chit-chat models lack specificity, do not display a consistent personality and are often not very captivating. |
| Approach: | They propose to train chit-chat models to condition on profile information and profile information about the interlocutors. |
| Outcome: | The proposed model can predict profile information about the interlocutors based on the data . the proposed model is able to generate meaningful responses in a chit-chat setting . |
Learning New Skills after Deployment: Improving open-domain internet-driven dialogue with human feedback (2023.acl-long)
Copied to clipboard
| Challenge: | Frozen models trained to mimic static datasets can never improve their performance. |
| Approach: | They propose to use binary quality measurements and free-form text feedback to improve conversational skills in a conversational learning framework. |
| Outcome: | The proposed model improves on the DIRECTOR model, which is based on binary quality measurements and free-form text feedback, and shows that iterative retraining and redeployment can improve the model. |
Director: Generator-Classifiers For Supervised Language Modeling (2022.aacl-main)
Copied to clipboard
| Challenge: | Current language models achieve low perplexity but their resulting generations still suffer from toxic responses, repetitiveness, and contradictions. |
| Approach: | They propose a new language model architecture that uses a language modeling and a classification head for each output token. |
| Outcome: | The proposed model outperforms existing model guiding approaches in terms of accuracy and efficiency. |
Dialogue Natural Language Inference (P19-1)
Copied to clipboard
| Challenge: | Consistency is a long standing issue faced by dialogue models. |
| Approach: | They propose to frame the consistency of dialogue agents as natural language inference and create a new natural language dataset called Dialogue NLI. |
| Outcome: | The proposed model can improve the consistency of a dialogue model with human evaluation and automatic metrics on a suite of evaluation sets designed to measure the model’s consistency. |
ELI5: Long Form Question Answering (P19-1)
Copied to clipboard
| Challenge: | Existing question answering datasets provide extractive or short answers, but less attention has been paid to open-ended questions that require explanations. |
| Approach: | They present a large-scale corpus for long form question answering . they use a Reddit forum to provide elaborate answers to open-ended questions . |
| Outcome: | The proposed model outperforms Seq2Seq, language modeling, and other models in human evaluations. |