🚀 Check out this awesome post from Hacker News đź“– đź“‚ Category: 📌 Key idea: ACM Classic: Reflections on Trusting Trust Reflections on Trusting TrustKen Thompson Reprinted from Communication of the ACM, Vol. 27, No. 8, August 1984, pp. 761-763. Copyright © 1984, Association for Computing Machinery, Inc. Also appears in ACM Turing Award Lectures: The First Twenty Years 1965-1985 Copyright © 1987 by the ACM press and Computers Under Attack: Intruders, Worms, and Viruses Copyright © 1990 by the ACM press. I copied this page from the ACM, in fear that it would someday turn stale. Introduction I thank…
🔥 Discover this trending post from WIRED đź“– đź“‚ Category: Gear,Gear / Reviews,Gear / Products / Home Office,Product Review đź’ˇ Key idea: Omni looks nice. It's a step up from your average office chair design, with a bit of sophistication and a design language on the backrest that echoes the spine-like look of the Herman Miller Embody. It comes in midnight black or space gray (creative names), and my unit is the latter. The company says the Omni can support people with up to ÂŁ300.I don't think I've ever sat in an office chair with softer padding than the Omni.…
🚀 Read this trending post from BBC Sport đź“– đź“‚ Category: 📌 Key idea: Alexander-Arnold was one of three senior Liverpool players - along with captain Virgil van Dijk and talisman Mohamed Salah - who entered the final year of their contracts last season.Van Dijk and Salah ended up signing new two-year contracts in April, after which the spotlight turned solely to the English defender. Posters were placed outside the stadium by disgruntled fans, questioning his loyalty to the club where he won two league titles, as well as the Champions League, Club World Cup, FA Cup and Carabao Cup.Alexander-Arnold…
đź’Ą Explore this insightful post from BBC Sport đź“– đź“‚ Category: 📌 Here’s what you’ll learn: Pereira signed a new three-year contract just 45 days ago after guiding Wolves to safety last season. He replaced Gary O'Neill last December - after O'Neill managed another 10-match winless streak - and oversaw a six-game winning streak between March and April. This was the club's best run in the top flight since 1970. It looked like a perfect match – Pereira was nominated for Premier League Coach of the Year – but cracks began to appear in the summer, which have grown to…
✨ Explore this insightful post from Culture | The Guardian đź“– đź“‚ Category: Television,Television & radio,Culture,Spice Girls,All Saints,Little Mix,Sugababes đź’ˇ Key idea: AIn a pop culture moment, the girl group boom at the turn of the millennium was not bathed in formal evaluation and analysis. Wisely, this fantastically entertaining three-part documentary doesn't attempt to correct that. Instead, Girlbands Forever reminisces in a way that is equal parts flesh and foam. And yes, it's often as nauseating as this combination sounds.At the heart of this series — the female-focused follow-up to 2024's Boybands Forever — is a lot of old ground.…
🚀 Discover this trending post from Investopedia | Expert Financial Advice and Markets News đź“– đź“‚ Category: Retirement Planning,Personal Finance đź’ˇ Here’s what you’ll learn: Key takeaways Sebring and surrounding Highlands County offer a lower cost of living than larger cities in Florida and many other retirement destinations in the Sunbelt.Known for its auto racing track, Sebring has many other attractions, including museums, art galleries and a historic downtown.People who love the outdoors will enjoy the parks, lakes, and golf courses.Sebring County offers more than 200 healthcare facilities and is within driving distance of many major cities, including Orlando and…
đź’Ą Check out this insightful post from Hacker News đź“– đź“‚ Category: âś… Key idea: GITHUB HUGGINGFACE MODELSCOPE SHOWCASEFrom Chatbot to Autonomous Agent#We are proud to present Tongyi DeepResearch, the first fully open‑source Web Agent to achieve performance on par with OpenAI’s DeepResearch across a comprehensive suite of benchmarks. Tongyi DeepResearch demonstrates state‑of‑the‑art results, scoring 32.9 on the academic reasoning task Humanity’s Last Exam (HLE), 43.4 on BrowseComp and 46.7 on BrowseComp‑ZH in extremely complex information‑seeking tasks, and achieving a score of 75 on the user‑centric xbench‑DeepSearch benchmark, systematically outperforming all existing proprietary and open‑source Deep Research agents.Beyond the model, we share a complete and battle‑tested methodology for creating such advanced agents. Our contribution details a novel data synthesis solution applied across the entire training pipeline, from Agentic Continual Pre‑training (CPT) and Supervised Fine‑Tuning (SFT) for cold‑starting, to the final Reinforcement Learning (RL) stage.  For RL, we provide a full‑stack solution, including algorithmic innovations, automated data curation, and robust infrastructure. For inference, the vanilla ReAct framework showcases the model’s powerful intrinsic capabilities without any prompt engineering, while the advanced Heavy Mode (test‑time‑scaling) demonstrates the upper limits of its complex reasoning and planning potential.Continual Pre‑training and Post‑training Empowered by Fully Synthetic Data#Continual Pre‑training Data#We introduce Agentic CPT to deep research agent training, creating powerful agentic foundation models for post‑training. We propose AgentFounder, a systematic and scalable solution for large‑scale data synthesis that creates a data flywheel with data from the post‑training pipeline.Data Reorganization and Question Construction. We continuously collect data from various sources, including documents, publicly available crawled data, knowledge graphs, and historical trajectories and tool invocation records (e.g., search results with links). As shown in the figure, these diverse data sources are restructured into an entity‑anchored open‑world knowledge memory. Based on randomly sampled entities and their corresponding knowledge, we generate multi‑style (question,answer) pairs.Action Synthesis.  Based on diverse problems and historical trajectories, we construct first‑order action synthesis data and higher‑order action synthesis data. Our method enables large‑scale and comprehensive exploration of the potential reasoning‑action space within offline environments, thereby thereby eliminating the need for additional commercial tool API calls. Specifically, for the higher‑order action synthesis, we remodel trajectories as multi‑step decision‑making processes to enhance the model’s decision‑making capabilities.Post-training Data#High-quality synthetic QA pairsWe develop an end‑to‑end solution for synthetic data generation. This fully automated process requires no human intervention to construct super‑human quality datasets, designed to push the boundaries of AI agent performance. Through long‑term exploration and iteration‑from early methods like reverse‑engineering QA pairs from clickstreams (WebWalker) to the more systematic graph‑based synthesis (WebSailor and WebSailor‑V2), then the formalized task modeling (WebShaper)‑our approach ensures both exceptional data quality and massive scalability, breaking through the upper limits of model capabilities.To address complex, high‑uncertainty questions, we synthesize web‑based QA data through a novel pipeline. The process begins by constructing a highly interconnected knowledge graph via random walks and isomorphic tables towards tabular data fusion from real‑world websites , ensuring a realistic information structure. We then sample subgraphs and subtables to generate initial questions and answers. The crucial step involves intentionally increasing difficulty by strategically obfuscating or blurring information within the question. This practical approach is grounded in a complete theoretical framework, where we formally model QA difficulty as a series of controllable “atomic operations” (e.g., merging entities with similar attributes) on entity relationships, allowing us to systematically increase complexity.To further reduce inconsistencies between the organized information structure and the reasoning structure of QA, enable more controllable difficulty and structure scaling of reasoning, we proposed a formal modeling of the information‑seeking problem based on set theory. With this formalization, we developed agents that expands the problem in a controlled manner, and minimizes reasoning shortcuts and structural redundancy, leading to further improved QA quality. Moreover, this formal modeling also allows for efficient verification of QA correctness, effectively addressing the challenge of validating synthetic information‑seeking data for post‑training.Furthermore, we have developed an automated data engine to scale up the creation of PhD‑level research questions. This engine begins with a multi‑disciplinary knowledge base, generating “seed” QA pairs that require multi‑source reasoning. Each seed then enters a self‑guided loop of “iterative complexity upgrades”, where a question‑crafting agent is equipped with a powerful toolset including web search, academic retrieval, and a Python execution environment. In each iteration, the agent expands knowledge boundaries, deepens conceptual abstraction, and even constructs computational tasks, creating a virtuous cycle where the output of one round becomes the more complex input for the next, ensuring a controllable and systematic escalation of task difficulty.Unleashing Agent Capabilities with Diverse Reasoning PatternTo bootstrap the model’s initial capabilities, we constructed a set of trajectories via rejection sampling, based on the ReAct and IterResearch frameworks (for details, see below). On one hand, ReAct, as a classic and foundational multi-turn reasoning format, instills rich reasoning behaviors and reinforces the model’s ability to adhere to structured formats.On the other hand, we introduce IterResearch, an innovative agent paradigm (detailed below). It unleashes the model’s full reasoning potential by…
✨ Check out this awesome post from WIRED đź“– đź“‚ Category: Gear,Gear / Products,Gear / Buying Guides,Gear / Products / Online Services,Buying Guide âś… Here’s what you’ll learn: I won't mince words: Sling TV is confusing. It has, by far, the most confusing array of plans and add-ons of any of the live TV services I've tested. There are a few basic plans, none of which include the range of Hulu Live TV, YouTube TV, or DirecTV, plus about a half-dozen add-ons to bring the channel lineup up to par. This standardized approach is annoying when shopping, although it also…
🚀 Discover this insightful post from BBC Sport đź“– đź“‚ Category: âś… Main takeaway: Once the referee stops the match due to a head injury, a doctor or physiotherapist will assess the player on the field.The player must then leave the field of play for further medical assessment and remain off the field for at least 30 seconds after the match resumes. If a player is judged likely to have suffered a concussion, he or she must not return to activity for at least 24 hours.Medical evaluations can include looking for obvious signs of a concussion such as a blank…
đź’Ą Read this awesome post from Sportskeeda đź“– đź“‚ Category: Cricket đź’ˇ Main takeaway: India registered a comfortable five-wicket win over Australia at Bellerive Oval, Hobart, on Sunday, November 2, in the third T20I of the ongoing five-match series. The Men in Blue bounced back superbly from a four-wicket defeat in the second match to level the series 1-1. After being asked to bat first, Australia finished on 186/6 after 20 overs. Arshdeep Singh helped India get off to a flying start, dismissing Travis Head (6 off 4) and Josh Inglis (1 off 7) early. Tim David rescued the hosts…
