PhantomEnvironments: Training LLM Agents in Fictional Worlds
Training LLM agents with reinforcement learning (RL) is bottlenecked by environments, which must provide verifiable rewards, support long-horizon interaction, and scale cheaply.
Every paper with a repository link. On a card, when the archive marks a repository official for the paper, the Syntology line starts with that repository's state (“official” is the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded). Each page shows two streams, newest first within each, counted separately: papers newer than the archive snapshot come from Syntology's graph Syntology; the rest are archive rows archive 2025-07-28. The two are never added together.
Cards 121–135 of 18,637 graph papers newer than 2025-07-28; this feed shows the newest 150, newest arXiv id first. Dates and the abstract sentence are from arXiv's metadata (CC0) for 18,637 of 18,637. Of the 18,637: 6,651 with a sample that ran: 5,550 with a run with no instrument failure, 1,101 where every run was a failure of Syntology's instrument.
Training LLM agents with reinforcement learning (RL) is bottlenecked by environments, which must provide verifiable rewards, support long-horizon interaction, and scale cheaply.
Language agents are expected to solve increasingly complex tasks, creating a growing need for continual improvement.
Power-law learning curves are often treated as fixed properties of a model and its data, although learning-rate and batch-size schedules can change the observed loss.
Bias has long been studied in social psychology and cognitive science, where decades of research have produced a body of validated interventions that reduce stereotypical thinking and prejudiced responses in humans.
Motivated by the computational challenges of large-scale Cox regression, we study stochastic minimization of LogSumExp objectives over large sets.
As LLM capabilities advance, agents are tackling increasingly complex tasks over longer horizons.
Continued pretraining enables language models to adapt to new domains and knowledge, but often at the cost of forgetting previously acquired capabilities.
Estimating causal effects from observational data is central to science and policy, but the effects are not identified when confounders are unmeasured.
Digital image watermarking is increasingly critical in media contexts, as emerging regulations and industry practices require marking AI-generated content and ensuring traceable sources to prevent manipulation or misuse.
Omni-modal large language models (OmniLLMs) enable unified audio-video understanding, but their long multimodal token sequences make deployment computationally expensive.
Vision-language models face a fundamental scaling bottleneck: the number of visual tokens grows with both temporal duration and spatial resolution, making long-video understanding expensive for the vision encoder and…
Parallel search may generate a correct answer that final-answer voting fails to select.
Scientific software presents a demanding test for computer-using agents based on visual language models (VLMs): completing a research workflow requires interpreting specialized interfaces, manipulating scientific…
Different from natural videos, Screen Content Videos (SCVs) are characterized by abrupt motion, scene switches, and high-frequency details such as text and graphics.
Role prompting elicits specialized behavior from large language models through an expert identity, offering a lightweight way to guide reasoning on demanding tasks.
Cards 121–135 of 218,469 archive papers with a code link; this feed shows the newest 150, archive date first (newest archive date 2025-09-24). Within a month, dated rows come first, then the 49,179 undated rows placed by the month in their arXiv id. 4 archive rows carry a date after the snapshot and are placed by that date. 382 archive papers with neither a date nor an arXiv id cannot be placed and are not listed. Of the 218,469: 38,710 with a sample that ran: 31,633 with a run with no instrument failure, 7,077 where every run was a failure of Syntology's instrument. 1 code-linked slug has no paper row in the archive and is not listed (so 218,469 listed + 382 unplaceable + 1 = 218,852 papers with code).
Language Model (LM) pipelines can dynamically refine their outputs against programmatic constraints.
Recent research has focused on applying speech large language model (SLLM) to improve speech emotion recognition (SER).
We introduce Ibex, a pan-immunoglobulin structure prediction model that achieves state-of-the-art accuracy in modeling the variable domains of antibodies, nanobodies, and T-cell receptors.
BrainLesion Suite is a versatile toolkit for building modular brain lesion image analysis pipelines in Python.
Cancer research is increasingly driven by the integration of diverse data modalities, spanning from genomics and proteomics to imaging and clinical factors.
Recent years have witnessed remarkable progress in developing Vision-Language Models (VLMs) capable of processing both textual and visual inputs.
Due to the excellent performance in yielding high-quality, zero-shot segmentation, Segment Anything Model (SAM) and its variants have been widely applied in diverse scenarios such as healthcare and intelligent…
Digital Twins (DT) have the potential to transform traffic management and operations by creating dynamic, virtual representations of transportation systems that sense conditions, analyze operations, and support…
Transformer models are computationally costly on long sequences since regular attention has quadratic O(n²) time complexity.
3D Semantic Scene Completion (SSC) has gained increasing attention due to its pivotal role in 3D perception.
Syntology
official (archive's flag): 7 ran
7 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 1 honoured, 0 violated, 1 with no contract checked; 5 where Syntology's instrument failed) · 1 unverified
“instrument failed”: failures of Syntology's instrument, not of the code
licence: 8 of 8 pointer-only
Medical image retrieval is a valuable field for supporting clinical decision-making, yet current methods primarily support 2D images and require fully annotated queries, limiting clinical flexibility.
Document image dewarping remains a challenging task in the deep learning era.
Large language models (LLMs) are increasingly used to support creative tasks such as research idea generation.
Enhancing the mathematical reasoning of Large Language Models (LLMs) is a pivotal challenge in advancing AI capabilities.
Syntology
official (archive's flag): 6 ran
6 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 0 honoured, 0 violated, 1 with no contract checked; 5 where Syntology's instrument failed) · 0 unverified
“instrument failed”: failures of Syntology's instrument, not of the code
licence: 6 of 6 pointer-only
In this paper, we investigate the applicability of the CLIP-EBC framework, originally designed for crowd counting, to car object counting using the CARPK dataset.
The feed is static: 10 pages of up to 15 cards per stream, rebuilt with the site. Older papers are reachable from task, dataset and method pages and from search. No repository stars are tracked and nothing here is ranked by popularity. Machine-readable twin: JSON.