arXiv subjects

Andrzej Banburski-Fahey

Publications and source records attributed to Andrzej Banburski-Fahey.

12 recordsLinked to original sources

arXiv2026

PuzzleJAX: A Benchmark for Reasoning and Learning

We introduce PuzzleJAX, a GPU-accelerated puzzle game engine and description language designed to support rapid benchmarking of tree search, reinforcement learning, and LLM reasoning abilities. Unlike existing GPU-accelerated learning environments that provide hard-coded implementations of fixed sets of games, PuzzleJAX allows dynamic compilation of any game expressible in its domain-specific language (DSL). This DSL follows PuzzleScript, which is a popular and accessible online game engine for designing puzzle games. In this paper, we validate in PuzzleJAX several hundred of the thousands of games designed in PuzzleScript by both professional designers and casual creators since its release in 2013, thereby demonstrating PuzzleJAX's coverage of an expansive, expressive, and human-relevant space of tasks. By analyzing the performance of search, learning, and language models on these games, we show that PuzzleJAX can naturally express tasks that are both simple and intuitive to understand, yet often deeply challenging to master, requiring a combination of control, planning, and high-level insight.

cs.AI↗

arXiv2026

Targeting Pivotal Decisions for Credit Assignment in Agentic Reinforcement Learning

Group Relative Policy Optimization (GRPO) has become a promising approach for training large language model agents. However, its uniform assignment of trajectory-level advantages to all policy tokens fails to distinguish consequential decisions from less relevant ones, obscuring which intermediate decisions contributed to success. We introduce ProVer, a framework that targets potentially pivotal decisions for fine-grained credit assignment in agentic reinforcement learning. Given a rollout group, an agentic judge contrasts successful and failed trajectories to propose a segment potentially responsible for their divergent outcomes. Rather than directly trusting the judge's assessment, ProVer verifies the proposed segment by estimating its advantage from the difference in terminal success rates between current-policy continuations sampled before and after the segment. Positive estimates are then incorporated into the GRPO advantages of policy tokens within the proposed segment. By using model judgment only to select where to verify, ProVer grounds local credit in observed outcomes without exhaustively evaluating every intermediate state. Across ALFWorld, WebShop, and SearchQA, ProVer achieves the strongest average performance at both model scales, with relative improvements over GRPO of 9.91% and 7.12% for Qwen3.5-2B and Qwen3.5-4B, respectively. Further analyses demonstrate that informed segment selection improves policy training with modest additional generation overhead, even without a frontier-scale judge model, highlighting the effectiveness and efficiency of selectively targeting pivotal decisions for fine-grained credit assignment in agentic reinforcement learning.

cs.CL↗

arXiv2026

Certified Long-Horizon Code Agent Evolution via Validation-Gated Skill Optimization

Long horizon agent self-evolution without model weight updates is essential for enabling deployed agents to accumulate reusable skills and improve over time. Prior self-evolution work has focused primarily on short-horizon tasks, while repository-level software engineering remains unexplored despite being an ideal testbed for long-horizon adaptation. In this setting, agents are required to solve streams of sequential tasks, navigate complex dependencies with evolving repositories and persistently store and reuse experience. Text-based skill optimization offers an efficient, non-parametric approach for such adaptation. However, existing methods often suffer from unstable updates, performance drawdown, and agent collapse over extended deployments. In this paper, we formalize the concept of in-context self-evolution and introduce VALVE, a validated-gated framework for long-horizon skill optimization. We establish finite convergence, provide theoretical guarantees for future-task gain and drawdown, and derive the validation and evaluation holdout sizes required for a prescribed tolerance, with leading-order scaling Empirically, our pipeline, VALVE achieves stable self-improvement over evolution horizon spanning more than 1,000 SWE tasks, with average final and peak gains of $14.9$ and $16.5$ points across three frontier models (GPT-5.5, Claude-4.6 and MiniMax-M2.7). The validation gate reduces average drawdown by 75% and produces an 11x more compact skill bank than ungated evolution. We further present extensive ablations identifying the design choices most critical to long-horizon skill evolution.

cs.AI↗

arXiv2026

Model-Aware Data Selection from In-and-Out Information Interplay

LLMs are effective representations that assimilate vast amounts of knowledge during pretraining, but post-training is necessary for models to reliably access this knowledge and "know what they know." We observe an interesting rank equilibrium between knowledge stored in the weights and the data stream passing through the model. Across all model layers, we find that the hidden states (data stream) follow a U-shaped pattern, showing substantial compression in early layers and a steep rise during the late-layer decoding phase. In contrast, the weight rank follows an inverted U-shaped pattern, with very low rank in the early and late layers and high rank in the middle. We interpret this as an in-and-out information interplay: intermediate activations do not need to carry content that the weights can supply later, so they primarily preserve what the weights cannot provide. Motivated by this observation, we propose a model-aware data selection method, CAP (Counterfactual Assimilation Profile), which can determine whether a data candidate contains information accessible to the current model by utilizing the divergence gap in early- and late-layer representations between model-generated and reference responses. Across math, code, and science domains, CAP delivers 35.4% greater average improvement over the base model than the strongest baseline under different selection budgets. With only 10% of the data pool, CAP surpasses or matches full-pool training on math and science. We further show that CAP transfers to multimodal data selection and is robust to response horizon and noise.

cs.AI↗

arXiv2026

DataDignity: Training Data Attribution for Large Language Models

Auditing language-model outputs often requires more than judging correctness: an auditor may need to identify which source document most likely supports the knowledge expressed in a response. We study this as pinpoint provenance: given a prompt, a target-model response, and a candidate corpus, rank the documents that best support the response. We introduce FakeWiki, a controlled benchmark of 3,537 fabricated Wikipedia-style articles designed to preserve ground-truth provenance while weakening lexical shortcuts. FakeWiki includes QA probes, source-preserving paraphrases, retro-generated variants, hard anti-documents that remain topically similar while removing answer-critical facts, and five query conditions: clean prompting plus four jailbreak-inspired transformations. We evaluate seven retrieval baselines, a training-free activation-steering retrieval-fusion method, SteerFuse, and a supervised contrastive provenance ranker, ScoringModel. ScoringModel maps response and document features into a shared space and is trained with InfoNCE using in-batch, retrieval-mined, and anti-document negatives. Across nine open-weight instruction-tuned LLMs and five query conditions, ScoringModel improves mean Recall@10 from 35.0 for the strongest retrieval baseline to 52.2, without inference-time fusion, and wins 41/45 model-by-condition cells. SteerFuse is usually second-best despite requiring no supervised training, showing that activation-space evidence can efficiently complement text retrieval. On jailbreak-inspired transformed queries, ScoringModel improves Recall@10 by 15.7 points on average over the best baseline. Overall, our work shows that robust training data attribution requires evaluation settings that separate true answer support from topical or lexical resemblance.

cs.AI↗

arXiv2025

DreamGarden: A Designer Assistant for Growing Games from a Single Prompt

Coding assistants are increasingly leveraged in game design, both generating code and making high-level plans. To what degree can these tools align with developer workflows, and what new modes of human-computer interaction can emerge from their use? We present DreamGarden, an AI system capable of assisting with the development of diverse game environments in Unreal Engine. At the core of our method is an LLM-driven planner, capable of breaking down a single, high-level prompt -- a dream, memory, or imagined scenario provided by a human user -- into a hierarchical action plan, which is then distributed across specialized submodules facilitating concrete implementation. This system is presented to the user as a garden of plans and actions, both growing independently and responding to user intervention via seed prompts, pruning, and feedback. Through a user study, we explore design implications of this system, charting courses for future work in semi-autonomous assistants and open-ended simulation design.

cs.HC↗

arXiv2025

ScriptDoctor: Automatic Generation of PuzzleScript Games via Large Language Models and Tree Search

There is much interest in using large pre-trained models in Automatic Game Design (AGD), whether via the generation of code, assets, or more abstract conceptualization of design ideas. But so far this interest largely stems from the ad hoc use of such generative models under persistent human supervision. Much work remains to show how these tools can be integrated into longer-time-horizon AGD pipelines, in which systems interface with game engines to test generated content autonomously. To this end, we introduce ScriptDoctor, a Large Language Model (LLM)-driven system for automatically generating and testing games in PuzzleScript, an expressive but highly constrained description language for turn-based puzzle games over 2D gridworlds. ScriptDoctor generates and tests game design ideas in an iterative loop, where human-authored examples are used to ground the system's output, compilation errors from the PuzzleScript engine are used to elicit functional code, and search-based agents play-test generated games. ScriptDoctor serves as a concrete example of the potential of automated, open-ended LLM-based workflows in generating novel game content.

cs.AI↗

arXiv2024

Social Conjuring: Multi-User Runtime Collaboration with AI in Building Virtual 3D Worlds

Generative artificial intelligence has shown promise in prompting virtual worlds into existence, yet little attention has been given to understanding how this process unfolds as social interaction. We present Social Conjurer, a framework for AI-augmented dynamic 3D scene co-creation, where multiple users collaboratively build and modify virtual worlds in real-time. Through an expanded set of interactions, including social and tool-based engagements as well as spatial reasoning, our framework facilitates the creation of rich, diverse virtual environments. Findings from a preliminary user study (N=12) provide insight into the user experience of this approach, how social contexts shape the prompting of spatial environments, and perspective on social applications of prompt-based 3D co-creation. In addition to highlighting the potential of AI-supported multi-user world creation and offering new pathways for AI-augmented creative processes in VR, this article presents a set of implications for designing human-centered interfaces that incorporate AI models into 3D content generation.

cs.HC↗

arXiv2024

Reprompting: Automated Chain-of-Thought Prompt Inference Through Gibbs Sampling

We introduce Reprompting, an iterative sampling algorithm that automatically learns the Chain-of-Thought (CoT) recipes for a given task without human intervention. Through Gibbs sampling, Reprompting infers the CoT recipes that work consistently well for a set of training samples by iteratively sampling new recipes using previously sampled recipes as parent prompts to solve other training problems. We conduct extensive experiments on 20 challenging reasoning tasks. Results show that Reprompting outperforms human-written CoT prompts substantially by +9.4 points on average. It also achieves consistently better performance than the state-of-the-art prompt optimization and decoding algorithms.

cs.LG↗

arXiv2024

LLMR: Real-time Prompting of Interactive Worlds using Large Language Models

We present Large Language Model for Mixed Reality (LLMR), a framework for the real-time creation and modification of interactive Mixed Reality experiences using LLMs. LLMR leverages novel strategies to tackle difficult cases where ideal training data is scarce, or where the design goal requires the synthesis of internal dynamics, intuitive analysis, or advanced interactivity. Our framework relies on text interaction and the Unity game engine. By incorporating techniques for scene understanding, task planning, self-debugging, and memory management, LLMR outperforms the standard GPT-4 by 4x in average error rate. We demonstrate LLMR's cross-platform interoperability with several example worlds, and evaluate it on a variety of creation and modification tasks to show that it can produce and edit diverse objects, tools, and scenes. Finally, we conducted a usability study (N=11) with a diverse set that revealed participants had positive experiences with the system and would use it again.

cs.HC↗

arXiv2024

Real-time Animation Generation and Control on Rigged Models via Large Language Models

We introduce a novel method for real-time animation control and generation on rigged models using natural language input. First, we embed a large language model (LLM) in Unity to output structured texts that can be parsed into diverse and realistic animations. Second, we illustrate LLM's potential to enable flexible state transition between existing animations. We showcase the robustness of our approach through qualitative results on various rigged models and motions.

cs.GR↗

arXiv2022

Steps towards prompt-based creation of virtual worlds

Large language models trained for code generation can be applied to speaking virtual worlds into existence (creating virtual worlds). In this work we show that prompt-based methods can both accelerate in-VR level editing, as well as can become part of gameplay rather than just part of game development. As an example, we present Codex VR Pong which shows non-deterministic game mechanics using generative processes to not only create static content but also non-trivial interactions between 3D objects. This demonstration naturally leads to an integral discussion on how one would evaluate and benchmark experiences created by generative models - as there are no qualitative or quantitative metrics that apply in these scenarios. We conclude by discussing impending challenges of AI-assisted co-creation in VR.

cs.HC↗