Interactive Fiction Games: A Colossal Adventure

By Matthew J. Hausknecht, Prithviraj Ammanabrolu, Marc-Alexandre Côté, Xingdi Yuan. 2019

261 citations recorded by the source corpus.

A hallmark of human intelligence is the ability to understand and communicate with language. Interactive Fiction games are fully text-based simulation environments where a player issues text commands to effect change in the environment and progress through the story. We argue that IF games are an excellent testbed for studying language-based autonomous agents. In particular, IF games combine challenges of combinatorial action spaces, language understanding, and commonsense reasoning. To facilitate rapid development of language-based agents, we introduce Jericho, a learning environment for man-made IF games and conduct a comprehensive study of text-agents across a rich set of games, highlighting directions in which agents can improve.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Is Reinforcement Learning (Not) for Natural Language Processing?: Benchmarks, Baselines, and Building Blocks for Natural Language Policy Optimization (2022, neutral): “, 2017), question generation (Pang & He, 2021), text-games (Narasimhan et al., 2015; Hausknecht et al., 2020), and more (Ranzato et al.”
WebShop: Towards Scalable Real-World Web Interaction with Grounded Language Agents (2022, background): “These include: 1) generation of good search queries [22, 59] and reformulation [37, 51], 2) strategic exploration for navigating through the website [55, 56, 29], 3) robust language understanding for textual state and action spaces [3, 7, 17, 44], and 4) long-term”
Language Models are Few-Shot Butlers (2021, background): “However, the application of modern language models to action generation in text-based environments (Côté et al., 2018; Hausknecht et al., 2019; Shridhar et al., 2021) remains largely unexplored (Yao et al., 2020).”
ALFWorld: Aligning Text and Embodied Environments for Interactive Learning (2020, methodology): “Interactive Text-Only Environments : We build on the work of text-based environments like TextWorld (Côté et al., 2018) and Jericho (Hausknecht et al., 2020).”
Keep CALM and Explore: Language Models for Action Generation in Text-based Games (2020, methodology): “We demonstrated that CALM can generate high-quality, contextuallyrelevant actions even for games unseen in its training set, and when paired with a DRRN agent, outperforms previous approaches on the Jericho benchmark (Hausknecht et al., 2019a) by as much as 69% in terms of average normalized score.”
Aligning AI With Shared Human Values (2020, background): “Eventually work should apply human value models in multimodal and sequential decision making environments (Hausknecht et al., 2019).”
Quark: Controllable Text Generation with Reinforced Unlearning (2022, methodology): “Finally, prior work has used RL techniques to generate language grounded in text-based narrative games [23, 4, 3].”
Graph Constrained Reinforcement Learning for Natural Language Action Spaces (2020, methodology): “Importantly, TDQN uses the same set of handicaps as KG-A2C allowing a fair comparison between these two algorithms.”
Reading and Acting while Blindfolded: The Need for Semantics in Text Game Agents (2021, methodology): “DRRN Baseline Our baseline RL agent DRRN (He et al., 2016) learns a Q-network Qφ(o, a) parametrized by φ.”
Generative Agents: Interactive Simulacra of Human Behavior (2023, background): “They have not yet addressed the challenge of creating believable agents in an open world [40, 74, 91].”
AgentBench: Evaluating LLMs as Agents (2023, methodology): “Historically, text-based game environments [57; 17; 30; 82] have been employed for language agent evaluation.”
Aligning to Social Norms and Values in Interactive Narratives (2022, methodology): “, 2021b) is a set of 25 text adventure games—drawn from the Jericho suite (Hausknecht et al., 2020)— that contains morality annotations for possible action-outcome in a particular world state.”
Learning Dynamic Belief Graphs to Generalize on Text-Based Games (2020, background): “Various deep learning agents have been explored [29, 17, 14, 51, 20, 3, 52, 47].”
Cognitive Architectures for Language Agents (2023, background): “This includes interacting with games (Hausknecht et al., 2020; Côté et al., 2019; Shridhar et al., 2020; Wang et al., 2022a; Liu et al., 2022), APIs (Schick et al.”
Situated Dialogue Learning through Procedural Environment Generation (2021, methodology): “Formally, we adapt the definition of text-based games as seen in (Côté et al., 2018; Hausknecht et al., 2020) to LIGHT.”