Storage Is Not Memory: A Retrieval-Centered Architecture for Agent Recall
arXiv:2605.04897v1 Announce Type: new Abstract: Extraction at ingestion is the wrong primitive for agent memory: content discarded before the query is known cannot be recovered at retrieval time. We propose True Memory, a six-layer architecture that shifts the center of the system from a storage sc
Rethinking Local Learning: A Cheaper and Faster Recipe for LLM Post-Training
arXiv:2605.04913v1 Announce Type: new Abstract: LLM post-training typically propagates task gradients through the full depth of the model. Although this end-to-end structure is simple and general, it couples task adaptation to full-depth activation storage, long-range backward dependencies and dire
Unintended Negative Impacts of Promotional Language in Patent Evaluation
arXiv:2605.04926v1 Announce Type: new Abstract: Promotional language has been increasingly used to aid the communication of innovative ideas in science. Yet, less is known about its role in the context of technological innovation. Here, we use a validated and domain-diagnosed lexicon of 135 promoti
UFAL-CUNI at SemEval-2026 Task 11: An Efficient Modular Neuro-symbolic Method for Syllogistic Reasoning
arXiv:2605.04941v1 Announce Type: new Abstract: This paper describes our system submitted to SemEval-2026 Task 11: Disentangling Content and Formal Reasoning in Large Language Models. We present an efficient modular neuro-symbolic approach, combining a symbolic prover with small reasoning LLMs (4B
Adapting Large Language Models to a Low-Resource Agglutinative Language: A Comparative Study of LoRA and QLoRA for Bashkir
arXiv:2605.04948v1 Announce Type: new Abstract: This paper presents a comparative study of parameter-efficient fine-tuning (PEFT) methods, including LoRA and QLoRA, applied to the task of adapting large language models to the Bashkir language, a low-resource agglutinative language of the Turkic fam
TabEmbed: Benchmarking and Learning Generalist Embeddings for Tabular Understanding
arXiv:2605.04962v1 Announce Type: new Abstract: Foundation models have established unified representations for natural language processing, yet this paradigm remains largely unexplored for tabular data. Existing methods face fundamental limitations: LLM-based approaches lack retrieval-compatible ve
Why Expert Alignment Is Hard: Evidence from Subjective Evaluation
arXiv:2605.04972v1 Announce Type: new Abstract: Aligning large language models with expert judgment is especially difficult in subjective evaluation tasks, where experts may disagree, rely on tacit criteria, and change their judgments over time. In this paper, we study expert alignment as a way to
Misaligned by Reward: Socially Undesirable Preferences in LLMs
arXiv:2605.05003v1 Announce Type: new Abstract: Reward models are a key component of large language model alignment, serving as proxies for human preferences during training. However, existing evaluations focus primarily on broad instruction-following benchmarks, providing limited insight into whet
Research
How an exoplanet odd couple survived by traveling in from the cold together
By probing the atmosphere of a mini-Neptune exoplanet, the James Webb Space Telescope has found that it formed much farther from its star than it is today, possibly explaining the origin of many other mini-Neptunes in the process.
Research
Happy 100th birthday, David Attenborough! 13 surprising facts about the famous naturalist
As Sir David Attenborough turns 100, here are 13 surprising facts about the beloved broadcaster and environmental advocate whose voice has shaped how generations see the natural world.
Research
Europe's 1st reusable spacecraft 'Space Rider' clears key hurdles on the road to launch
Space Rider, a novel spacecraft concept from the European Space Agency, is passing key milestones in its tests and qualifications to bring the vehicle to flight readiness.
Research
The night sky could get three times brighter as new satellites launch — all but ruining the Vera C. Rubin Observatory's survey of the universe
Extremely bright satellites and megaconstellations could make the night sky up to three times brighter than it is now, a new study warns. This would seriously hinder astronomical imaging devices, like the enormous camera at the Vera C. Rubin Observatory.
Research
The brain's memory center doesn't start as a blank slate, study suggests
Early in life, neural networks in the brain's memory center are highly connected, and they are only later refined into precise systems, a mouse study finds.
Research
'Whatever Russia is testing, it's sophisticated': 2 Russian satellites get within 10 feet of each other in orbit
Two Russian satellites got within 10 feet (3 meters) of each other recently, demonstrating a sophisticated set of orbital maneuvering skills.
Research
3 puzzles of our universe could be solved with this new dark matter theory
A new recipe of dark matter that interacts with itself could be the solution to three separate and vastly different cosmic puzzles.
Research
'Thor' hammered down 15 years ago, and it's still the only Marvel movie to do the God of Thunder justice
In 2011, Thor proved himself more than worthy in his roaring MCU debut. Since then, the thunder has turned into laughter.
Research
Sci-fi RPG 'Exodus''s combat looks and feels like 'Mass Effect', but the dialogue system could be a big upgrade (video)
Jun Aslan and Phaedra share a quiet chat in Archetype Entertainment's latest Community Update.
Research
Nikon Action 7x50 binoculars review
Nikon's new Action 7x50 binoculars: A good buy for astronomy novices
Research
Why does NASA's Curiosity rover have a 'lucky penny' on Mars? | Space photo of the day for May 6, 2026
Providing good luck to the first Martian to find it.
Research
Quantum battery charges in a quadrillionth of a second with a laser — larger prototypes could last for years after charging for just a minute
Quantum batteries can be charged remotely and could allow for far better energy density than conventional batteries used in devices today.