No reviews yet. Be the first to share your experience!
Latest Posts
Just links
May 21, 2026, 06:00 AM
Natural Language Autoencoders Produce Unsupervised Explanations of LLM Activations https://transformer-circuits.pub/2026/nla/index.html
2,030
0
0
Just links
May 21, 2026, 06:00 AM
https://lean-lang.org/eval/
1,680
1
0
Just links
May 21, 2026, 06:00 AM
SPEC CPU: The Next Generation https://arxiv.org/abs/2605.01575
1,510
0
0
Just links
May 21, 2026, 06:00 AM
Soohak: A Mathematician-Curated Benchmark for Evaluating Research-level Math Capabilities of LLMs https://arxiv.org/abs/2605.09063
1,200
0
0
Just links
May 21, 2026, 06:00 AM
Incompressible Knowledge Probes: Estimating Black-Box LLM Parameter Counts via Factual Capacity https://arxiv.org/abs/2604.24827
via https://t.me/seeallochnaya
2,240
3
0
Just links
May 21, 2026, 06:00 AM
Report of the 5th PVUW Challenge: Towards More Diverse Modalities in Pixel-Level Understanding https://arxiv.org/abs/2604.26031
1,840
0
0
Just links
May 21, 2026, 06:00 AM
📷 Photo
How do we measure 3D spatial intelligence?
We show an agent ~20 photos from inside an apartment and ask it to produce the floor plan. It has to identify rooms, work out connections, and keep scale consistent. It does this for 50 apartments, with a notepad to learn across them.
Read more: https://andonlabs.com/evals/blueprint-bench-2
2,160
2
Just links
May 21, 2026, 06:00 AM
Correlated Phase Error Bursts in a Gap-Engineered Superconducting Qubit Array https://journals.aps.org/prx/abstract/10.1103/1bl4-b2f7
2,360
0
0
Just links
May 21, 2026, 06:00 AM
Too Sharp, Too Sure: When Calibration Follows Curvature https://arxiv.org/abs/2604.20614
2,550
2
0
Just links
May 21, 2026, 06:00 AM
Where the goblins came from https://openai.com/index/where-the-goblins-came-from/
2,070
4
0
Just links
May 6, 2026, 06:59 PM
Продолжаем геммапропаганду. В прошлом году у NVIDIA вышла https://arxiv.org/abs/2510.27055 о том, как ловить людей, которые доливают тест в трейн. CoDeC – нормализованный показатель перплексии, где для тестсета бенчмарка считают изменения в перплексии с дополнительными примерами из того же бенчмарка. Для неконтаминированных моделек мы ожидаем, что дополнительные примеры не будут сбивать модель с толку, а в лучшем случае помогут. С другой стороны, если модель запомнила текст из теста, дополнительные примеры собьют её с толку и уверенность модели в ответе упадёт. Шкала нормализована от 0 до 100, где ~80% значит, что примеры из теста модель видела буквально, ~40% – в перефразированном виде. Товарищ с твиттера https://x.com/bnjmn_marie/status/2041540879165403527?s=52 CoDeC для Gemma 4 и сравнил с Qwen 3.5 – почему-то у наших китайских коллег модель почти запоминает примеры из теста.
When Can LLMs Learn to Reason with Weak Supervision? https://salmanrahman.net/rlvr-weak-supervision
1,900
3
0
Just links
May 6, 2026, 06:59 PM
Erdős Problem #1196
https://www.erdosproblems.com/forum/thread/1196
2,800
0
0
Just links
May 6, 2026, 06:59 PM
Test of the Gravitational Force Law on Cosmological Scales Using the Kinematic Sunyaev-Zeldovich Effect https://journals.aps.org/prl/abstract/10.1103/rk8v-rcm3
2,040
2
0
Just links
May 6, 2026, 06:59 PM
Offline Materials Optimization with CliqueFlowmer https://arxiv.org/abs/2603.06082
2,090
0
0
Just links
May 6, 2026, 06:59 PM
Preparing 100-qubit symmetry-protected topological order on a digital quantum computer https://arxiv.org/abs/2603.06325
1,990
3
0
Just links
May 6, 2026, 06:59 PM
A Fuzzy Sphere Journey in Critical Phenomena https://www.annualreviews.org/content/journals/10.1146/annurev-conmatphys-031424-020256
2,410
0
0
Just links
May 6, 2026, 06:59 PM
Topologically shadowed quantum criticality: A non-compact conformal manifold https://arxiv.org/abs/2604.05391
2,710
2
0
Just links
May 6, 2026, 06:59 PM
Riemann-Bench: A Benchmark for Moonshot Mathematics https://arxiv.org/abs/2604.06802
2,570
3
0
Just links
May 6, 2026, 06:59 PM
📷 Photo
В эту https://calendar.google.com/calendar/event?action=TEMPLATE&tmeid=NHRydnJmcjAwOXRmNGZsaWE1Z3Q5NGZ1M2ggYW5kcmVpQHBhbmZlcm92Lm9yZw&[email protected] рассказываем про NVFP4 претрен на GPU Mode.
Show Me When and Where: Towards Referring Video Object Segmentation in the Wild https://arxiv.org/abs/2603.14300
2,160
0
0
Just links
May 1, 2026, 07:16 AM
Thinking—Fast, Slow, and Artificial: How AI is Reshaping Human Reasoning and the Rise of Cognitive Surrender https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6097646
The 1st Winner for 5th PVUW MeViS-Text Challenge: Strong MLLMs Meet SAM3 for Referring Video Object Segmentation https://arxiv.org/abs/2604.00404
1,700
0
0
Just links
Apr 19, 2026, 07:17 PM
AgentRVOS: Reasoning over Object Tracks for Zero-Shot Referring Video Object Segmentation https://arxiv.org/abs/2603.23489
2,290
0
0
Just links
Apr 19, 2026, 07:17 PM
NASA Artemis II moon mission live launch broadcast (🔥 Score: 156+ in 1 hour)
Link: https://readhacker.news/s/6R5Kj
Comments: https://readhacker.news/c/6R5Kj
1,800
11
0
Just links
Apr 19, 2026, 07:17 PM
GrandCode: Achieving Grandmaster Level in Competitive Programming via Agentic Reinforcement Learning https://github.com/deepreinforce-ai/grandcode/blob/main/grandcode.pdf
2,070
2
0
Just links
Apr 19, 2026, 07:17 PM
Intrinsic Error Thresholds in Nearly Critical Toric Codes https://arxiv.org/abs/2603.14098
2,200
2
0
Just links
Apr 19, 2026, 07:17 PM
Revisiting Tree Search for LLMs: Gumbel and Sequential Halving for Budget-Scalable Reasoning https://arxiv.org/abs/2603.21162
