Hamidah Oderinwale
89 followers · 66 following · 6017 views
on the atlas — 156
- The Rationality Community Sucks28 savers
- Longitudinal Science – Cross-disciplinary road-mapping and analysis of science and technology topics1 savers
- O-ring theory of economic development6 savers
- Jonny Kim1 savers
- Compact2 savers
- homwolf/Magic-Sand: Magic-Sand is a software for operating an augmented reality sandbox ·1 savers
- oakwood studio4 savers
- caenopy1 savers
- What African philosophy says about our duties to oneself | Aeon Essays1 savers
- Elsevier — my part in its downfall | Gowers's Weblog2 savers
- State-space representation1 savers
- A World Without Illusions | Celso Amorim1 savers
- Alexa Tartaglini | about2 savers
- Readme | Zhengdong10 savers
- Townscaper3 savers
- Don’t Learn Value From Society9 savers
- Notes on Tajikistan – Matt Lakeman4 savers
- Distill is dedicated to making machine learning clear and dynamic5 savers
- Incremental Database Computations2 savers
- The Dial2 savers
- The Choice to Become a Mother Alone — The Dial2 savers
- ekzhang/jute: A native Jupyter notebook frontend with local + remote kernels, reactive cells, and IDE features, implemented in Rust1 savers
- Why you should share your data - by Sergey Frolov1 savers
- arXivLabs: An invitation to collaborate - arXiv info1 savers
- Conservation Ecology: Bridging the Gap Between Economics and Ecology1 savers
- A century of statistical Ecology - Gilbert - 2024 - Ecology - Wiley Online Library1 savers
- sylvainhalle/textidote: Spelling, grammar and style checking on LaTeX documents1 savers
- HypotheSAEs2 savers
- Backing Up US Federal Databases | Azimuth1 savers
- Home - Loud and Clear3 savers
- A Framework for Standardizing Similarity Measures in a Rapidly Evolving Field | OpenReview1 savers
- Saerch.ai - a Hugging Face Space by charlieoneill1 savers
- The Data Cards Playbook - Data Cards Playbook1 savers
- The Overedge Catalog: The Future of Research Organizations7 savers
- Bag-of-words model - Wikipedia2 savers
- Game theory as an engine for large-scale data analysis - Google DeepMind1 savers
- The Gap Map34 savers
- Federal Government Academy, Suleja - Wikipedia1 savers
- Paper Ideas1 savers
- Irydium1 savers
- reStructuredText - Wikipedia1 savers
- Inventing the Future: Postcapitalism and a World Without Work | Verso Books1 savers
- The Great Silence - Electric Literature2 savers
- Likelihood-ratio test2 savers
- Complex Systems Won’t Survive the Competence Crisis2 savers
- The Mountain To Climb | Oliver W. Kim3 savers
- ARIA-Safeguarded-AI-Programme-Thesis-V1.pdf3 savers
- Topological Data Analysis: An Overview2 savers
- Music in Human Evolution | Melting Asphalt8 savers
- Better Living Through Group Chemistry—Asterisk2 savers
- Why transformative artificial intelligence is really, really hard to achieve13 savers
- Idea List6 savers
- Staring into the abyss as a core life skill85 savers
- Theory of Change (Aaron Swartz's Raw Thought)83 savers
- Quit Your Job58 savers
- Don’t Surround Yourself With Smarter People54 savers
- Taste is a guide for what is worthwhile49 savers
- Andy Matuschak44 savers
- To Enjoy UWaterloo40 savers
- Learn In Public30 savers
- choosing - by vincent huang - a slice of my mind29 savers
- A Gentle Introduction to Graph Neural Networks26 savers
- Multi-layered calendars « julian.digital26 savers
- Maggie Appleton26 savers
- Cognitive load is what matters25 savers
- School Is Not Enough25 savers
- How to Optimize a CUDA Matmul Kernel for cuBLAS-like Performance: a Worklog24 savers
- on taking yourself home - by jennifer tsai - jenn's journal24 savers
- June Huh, High School Dropout, Wins the Fields Medal | Quanta Magazine24 savers
- The Missing Semester of Your CS Education · the missing semester of your cs education19 savers
- Nadia Asparouhova | Explaining tech's notion of talent scarcity17 savers
- Not a Number17 savers
- 12ft Ladder16 savers
- Bret Victor, beast of burden16 savers
- Distill — Latest articles about machine learning16 savers
- The Best Tacit Knowledge Videos on Every Subject — LessWrong15 savers
- Using Artificial Intelligence to Augment Human Intelligence15 savers
- Howl by Allen Ginsberg | Poetry Foundation14 savers
- Ronald McDonald just chased me thru the play place and I spilled my apple juice13 savers
- Read Something Great13 savers
- A Vision of Metascience13 savers
- You And Your Research · Gwern.net13 savers
- EC05-Chen.dvi12 savers
- Montreal in Motion12 savers
- Don't End The Week With Nothing11 savers
- Welcome to The Creative Independent11 savers
- About | Dan Wang, Gavekal Dragonomics and Bloomberg contributor11 savers
- NYSRG11 savers
- Exorcising us of the Primer | Andy Matuschak10 savers
- Pandoc - About pandoc9 savers
- Explorable Explanations9 savers
- Frontier9 savers
- About me | devonzuegel.com9 savers
- 100 things you can do on your personal website | James' Coffee Blog9 savers
- Tools, Computing and Research Labs9 savers
- Orbit8 savers
- A Vision of Metascience8 savers
- top 10 (university) life hacks8 savers
- Dunbar's number8 savers
- Building data-centric apps with a reactive relational database8 savers
highlights — 342
In addition to doing the work of developing a novel evaluation framework and challenging dataset to test world knowledge, we lay out best practices for evaluation data use: our framework and dataset are licensed CC-BY 4.0 with Terms of Use (1) requiring attribution if used to train models, and (2) prohibiting plain-text release of EWoK-generated data on the internet to prevent accidental inclusion in large training corpora. We would like to encourage a community standard of clearly reporting what models are trained on, and appropriately providing attribution where it is due---both, for careful…
EWoK: Elements of World KnowledgeSearchable columns# Let's create a new column fts inside the books table to store the searchable index of the title and description columns. We can use a special feature of Postgres called Generated Columns to ensure that the index is updated any time the values in the title and description columns change.
Full Text Search | Supabase DocsAll rights reserved ©, but not for long: We are retaining the rights while we develop a license that fits our mission. Stay tuned.
About Usnon-textual file formats remain second-class in Git.
Local-first software: You own your data, in spite of the cloudnon-textual file formats remain second-class in Git.
Local-first software: You own your data, in spite of the cloudFollow the guide to implement FTS with GIN-indexes and semantic search with pgvector (also known as bi-encoder dense retrieval).
Postgres as a search engine — any blockers?Making it easier for foreign researchers and AI professionals to work in the United States would likely both lower post-grad outflow, by allowing them to stay in the United States after completing their degrees, and increase post-grad inflow, by making it more attractive for researchers who are educated outside the United States to come to the United States for work. Targeted immigration reform measures could help increase the supply of top talent in the United States.
AI Talent Report | CEA | The White HouseMaking it easier for foreign researchers and AI professionals to work in the United States would likely both lower post-grad outflow, by allowing them to stay in the United States after completing their degrees, and increase post-grad inflow, by making it more attractive for researchers who are educated outside the United States to come to the United States for work. Targeted immigration reform measures could help increase the supply of top talent in the United States.
AI Talent Report | CEA | The White HouseIn July 2020, GitHub stored a February archive of the site[60] in an abandoned mountain mine in Svalbard, Norway, part of the Arctic World Archive and not far from the Svalbard Global Seed Vault. The archive contained the code of all active public repositories, as well as that of dormant but significant public repositories. The 21TB of data was stored on piqlFilm archival film reels as matrix (2D) barcode (Boxing barcode), and is expected to last 500–1,000 years.[109][110][111][112] The GitHub Archive Program is also working with partners on Project Silica, in an attempt to store all public re…
GitHubI can't reproduce unfortunately, even when running with the same versions of Node and NPM as you. :(
`dev` cells throws module import error · Issue #21 · ekzhang/percivalIf we publish runtime.build to NPM, we can import Percival and run datalog within ObservableHQ notebooks. But, we won't have syntax highlighting :(
JS in Percival / Percival in JS · Issue #18 · ekzhang/percivalAt the regional level, maintain an up-to-date database of cables traversing Southeast Asia including details on ownership as well as countries’ policies and regulations on cables. The academic community has created a basis for regulation summaries that could be built upon with the necessary support.81 Adding information on ownership and cable ship bases in the region would aggregate regional data in a single location and provide transparency for researchers and policymakers on the future of cables in Southeast Asia.
Subsea Communication Cables in Southeast Asia: A Comprehensive Approach Is Needed | Carnegie Endowment for International PeaceSTGs have been interfaced with various HDLs, see for example links with VHDL[35] (1996) and Verilog[36] (2000) with the aim to support asynchronous design. Placed into the synthesis flow from VHDL, STGs and Petri nets have been shown instrumental,[37] and likewise with Verilog,[38] where a tool VERISYN was developed.[39] More recently STGs have been connected with notations that are believed to be easier for practical hardware designers, hence the emergence of the model of waveform-transition graphs (WTG).[40] Likewise, realising that the model of finite state machine (FSM) can be easier for d…
Signal transition graphsBut once you’re an expert, it’s 6 years later, you sit at the frontiers of human knowledge — and nothing you write can be understood by 99% of people. Your work is alien scribbles. The world has no empathy for experts, and that’s hard to take.
Rekindling Is RefiningIn the early 2000s, Stassun co-founded the Fisk-Vanderbilt Master's-to-PhD Bridge Program (FVBP). The program increases the diversity of students receiving advanced degrees in STEM fields by serving as a stepping stone for promising students at Fisk University to gain the experience and skills needed to go directly into a doctorate program. Since it began 18 years ago, 172 students have enrolled in the FVBP, 137 master's degrees have been awarded, 120 students have bridged to doctoral programs and 52 students have earned their doctorate. Today, Fisk University is the number one producer of Afr…
Four NSF-supported researchers awarded MacArthur Genius Grants | NSF - National Science FoundationI hope the 2013 version is otherwise error-free. But (sigh) it probably isn't. Therefore I will gratefully deposit 0x$1.00 ($2.56) to the account of the first person who finds and reports anything that remains technically, historically, typographically, or politically incorrect. Please send suggested corrections to knuth-bug@cs.stanford.edu, or send snail mail to Prof. D. Knuth, Computer Science Department, Gates Building 1B, Stanford University, Stanford, CA 94305-9015 USA. In either case please include your postal address, so that I can mail an official certificate of deposit as a token of t…
Knuth: Literate Programming"I take responsibility for my crimes, for my charges," he said. "I want to say sorry to my family, my mom, my mom's got 11 kids, I can't say all their names, my managers, my kids."
Is Young Thug Getting Released From Jail Today? Here's All We Know - NewsweekThere are more features available in KGL, such as filtering and a depth-first search-powered network analyzer to find how two nodes in a graph connect. These are documented in the project language reference.
Designing a knowledge graph query language | James' Coffee Blogggest neural representations and computational components are reused across tasks. Shared representational subspaces could speed learning by allowing knowledge to generalize across tasks
Building compositional tasks with shared neural subspaces - PMCneural representations and computational components are reused across tasks. Shared representational subspaces could speed learning by allowing knowledge to generalize across tasks
Building compositional tasks with shared neural subspaces - PMCunderstand open continuous-time Markov chains
A Compositional Framework for Markov Processes | AzimuthBoth categorical and category-free representations can be low or high-dimensional (Figure 2, right). Interestingly, it is also possible to determine the dimensionality of the representations in the condition space: linear algebra dictates that the row rank of the matrix is the same as the column rank. However, in the categorical case, the number of clusters can limit the maximal dimensionality in the same way that the total number of points can. Note that the study of the distribution of points in the condition space depends less on experimenter choices than it does in many types of analysis. …
The implications of categorical and category-free mixed selectivity on representational geometries - ScienceDirectThis may or may not have positive values for all the 𝑥 𝑖 ; if it does not, then there is no stable attractor for which the populations of all species are positive. If there is a fixed point with all positive populations the Jacobian matrix in a neighbourhood of the fixed point 𝑥 is given by diag ( 𝑥 ) 𝐴 .
Generalized Lotka–Volterra equation - WikipediaOne solution to a fast-growing monorepo is to consider it as if it was a time-series database: the changes over time are important, so what if it shards based on time instead of based on the worktree? When performing a time-based shard, first determine a point in time where the existing monorepo can be paused and all movement on the trunk branch can be blocked. Pausing work on a monorepo is very unusual, so should be done with extreme care and preparation. After pausing the changes to the monorepo’s trunk, create a new repository with the same root tree as the current trunk of the old monorepo…
Git's database internals V: scalability - The GitHub BlogJames Bernat, a neurologist at Dartmouth, proposed a rationale for why brain death and biological death could actually be one and the same. He claimed that when patients have the devastating brain injury that represents brain death, it turns out that they always have a cardiac arrest within a very short period of time, no more than a few days.
Harvard ethicist Robert Truog on ambiguities of brain death — Harvard Gazette7: Anyone with access to the internet can publish their thoughts online on a pretentious minimalist website. It’s just that most people don’t, which leads to a kind of fallacy where it can seem like the words I publish here are inherently more special than some random comment your Uber driver made yesterday, or the weird plotline of last night’s dream. They are not (necessarily).
Disclaimer - Quarter Mile7: Anyone with access to the internet can publish their thoughts online on a pretentious minimalist website. It’s just that most people don’t, which leads to a kind of fallacy where it can seem like the words I publish here are inherently more special than some random comment your Uber driver made yesterday, or the weird plotline of last night’s dream. They are not (necessarily).
Disclaimer - Quarter MileSecond: As much as I like to see myself as independent, so much of me is the people around me. So much of me is the culture we live in. And while this isn’t necessarily bad, I could use some space to reflect; unstructured time to live and learn apart from the world I’m embedded in. Ava, a writer whose words I love: If your judgement is always derivative, ripped from your community or your friends, how can you ever trust yourself? How can you ever sit in a dark room and know what you want when all sources of validation are stripped away?
O n Taking a Gap YearIn contrast, deterministic database systems can exploit the application-specific fact that the order of non-conflicting transactions does not need to be constrained, even if they acquire the same mutex(es) internally, because reordering them does not affect externally visible state.
An Overview of Deterministic Database Systems – Communications of the ACMRecovery from the input log. Database state recovery is typically much simpler in deterministic database systems relative to traditional nondeterministic database systems. Deterministic database systems can simply load a checkpoint23 and play the input log forward from there in order to recover state at the time of the failure (and continue from there).19,26,31 In contrast, the input request log is not sufficient to recover state in non-deterministic systems. Instead, they have to replay history, which typically involves physically reloading every page that was modified from a “redo log.”20
An Overview of Deterministic Database Systems – Communications of the ACMopen continuous-time Markov chains
A Compositional Framework for Markov Processes | The n-Category CaféFor instance, if a 20-something got in a fight with another 20-something and the cops got involved, the police would most likely contact the families and let the parents and brothers sort that shit out with a warning not to cause trouble again.
Notes on Tajikistan – Matt LakemanThen he said a few broken words to me in English for five minutes which concluded with asking me how much a visa to the US costs.
Notes on Tajikistan – Matt LakemanThe countryside is dotted with half-finished concrete homes which I was told were the long-term retirement plans of Tajiks working abroad; they build their structures for a few months each year over many years or decades using trickles of funds earned from abroad. When the structures are finished, they prepare them to be their permanent homes once they retire from migrant work.
Notes on Tajikistan – Matt LakemanIn 2014, Tajikistan was the most-remittance based economy on earth, amounting to almost half of its GDP.
Notes on Tajikistan – Matt LakemanThe work was strengthened in [28] to study the 2000 and 2008 based on persistence landscapes, which was introduced in [29] to quantify the persistence homology of complexes.
Empirical study of financial crises based on topological data analysis - ScienceDirectIn January 2016, he published a topological data analysis on the Donald Trump presidential campaign, 2016 and was able to outline the reach potential of Trump's messages in the mind of skeptical voters.[10]
Gunnar Carlsson - WikipediaSituations with systematic feedback (e.g., users tag their friends in photos or mark emails as spam) or self-supervision approaches like BERT [D19], SimCLR [CK20], and DINO [C21] can generate an unbounded amount of data.
Growing or Compressing Datasets · Introduction to Data-Centric AIThe COVID-19 pandemic showed weaknesses in the IHR, including that it did not mention “pandemic”. Member states submitted over 300 proposed amendments to the Working Group on Amendments to the IHR (WGIHR), chaired by New Zealand’s Dr Ashley Bloomfield and Dr Abdullah Asiri of Saudi Arabia.
'The World Has Won': New International Regulations To Protect Against Pandemics Finally Approved - Health Policy WatchCharles Gore, executive director of the Medicines Patent Pool, is confident that the vaccine will be successful and that the South African government will order it. “What I'm less confident about is whether other countries in the region will support this and that's what we need. We need all 54 African countries buying the vaccine, not just the six where the spokes are ideally to be supporting this project,” he adds. The vaccine must be priced well compared with alternatives for it to be an attractive option.
First African-made mRNA vaccine, a test platform for the futureMany Chinese nationals with AI expertise have chosen to work at Kaust because they have been prevented from studying and working in the US after graduating from Chinese universities on the US entity list, according to two Kaust sources.
Saudi Arabia and UAE race to buy Nvidia chips to power AI ambitionsLarge Language Models as Simulated Economic Agents: What can we learn from Homo Silicus?
Research | John J. HortonIf you know you need a specific precision after the decimal/binary point, then fixed point allows you to maintain accuracy of those digits without truncation even as the value grows, up to the limits of the range. If you need more range, you have to add more bits. Hence decimal128 becomes important for some users.
Implementing High-Precision Decimal Arithmetic with CUDA int128 | NVIDIA Technical BlogHowever, rounding high precision results to a lower precision is not equivalent to performing the entire computation in lower precision.
Floating Point and IEEE 754Because of these issues, guaranteeing a specific precision level on the CPU can sometimes be tricky
Floating Point and IEEE 754On 32-bit x86 targets without SSE it can be helpful to declare variables using volatile and force floating point values to be stored to memory (/Op in Visual Studio and -ffloat-store in gcc). This moves results from extended precision registers into memory, where the precision is precisely single or double precision. Alternately, the x87 control word can be updated to set the precision to 24 or 53 bits using the assembly instruction fldcw or a compiler option such as -mpc32 or-mpc64 in gcc.
Floating Point and IEEE 754Current generations of the NVIDIA architecture such as Tesla Kxx, GTX 8xx, and GTX 9xx, support both single and double precision with IEEE 754 precision and include hardware support for fused multiply-add in both single and double precision. In CUDA, the features supported by the GPU are encoded in the compute capability number. The runtime library supports a function call to determine the compute capability of a GPU at runtime; the CUDA C++ Programming Guide also includes a table of compute capabilities for many different devices
Floating Point and IEEE 754Microdosing Lucidity
ykumar.orgNietzsche contra Wagner
ykumar.orgThe default hash algorithm used by RDFC-1.0, namely, SHA-256 [FIPS-180-4]. Implementations MUST support a parameter to define the hash algorithm, MUST support SHA-256 and SHA-384 [FIPS-180-4], and SHOULD support the ability to specify other hash algorithms. Using a different hash algorithm will generally result in different output than using the default.
RDF Dataset Canonicalization