Hey Hackernoon Readers, I’m , one of the managing partners at blockchain venture fund . -- First of all, a huge thank you to David, and entire Hackernoon community and staff for awarding me : Christian Kameir Sustany Capital five writing awards in 2021 Contributor of the Year - Money Contributor of the Year - VC NFT Wizard of 2021 Contributor of the Year - Art Contributor of the Year - Internet As someone who has focused on the decentralization of the internet since 2008, I think that the most exciting technology of the present is Web 3.0 because we can return agency to the users of the technology again, rather than allowing large companies to social-engineer users to cater to corporate objectives. Learn more about my thoughts and opinions on Web 3.0 and my journey in the tech industry via the interview below. 1. What do you do and why do you do it? As managing partners at Sustany Capital, we focus on operating a thesis-driven venture capital firm investing in Web 3.0, decentralized finance, and legal technologies. I have been a tech entrepreneur since the internet dial-up days and led the merger of two of the earliest internet service providers in Europe. In 2000 I moved to California to start a technology venture fund. I  studied classical literature and graduated from Muenster's School of Law. I am a board member at , co-chair of the Banking and Financial Services Interest Group at the , and lecture a class at . I write and speak on topics ranging from high-tech investing to decentralized finance, identity, and blockchain. Our current book project - working title ' ' - will be published in Q2/2022. Open Travel Decentralized Identity Foundation Harvard Streaming Money 2. Tell us more about the things you write about! I have been writing on ideas to “fix the web” since 2008. A core objective to me is to fix information distribution on the world wide web to do the most good. As such it is essential to foster solutions that contribute to creating consensus, and reverse the current paradigms that seek engagement for divisive content. Internet users should not be treated as NPCs - gamer-speak for non-player characters - but improve their own experiences daily by network technology that serves the interest of the individual. To me, advertising is not a suitable business model on top of a shared resource (commons). 3. How did you end up on your current career path? I started in software development in 1985, and I am as fascinated with the opportunities of network technology now, as I was back then. Realizing that I would never be a great coder, but understanding software principles well enough to explain them to others, I decided to join a startup as their general counsel. At the time - in 1997 - we were selling dial-up internet services, and we had to explain why this technology was relevant and interesting. The same is still true for the upcoming fundamental changes to the internet (Web 3.0), today. 4. What tech are you most excited or about right now and why? Returning control over bytes back to users: regardless if this refers to money or any other digitally expressed value or thought. Blockchain-based solutions are a central part of that, and many other cryptographic primitives that remove friction from our daily lives, and allow individuals around the world to engage in peer-to-peer interactions. Why this may sound corny: these are true ! freedom technologies 5. What tech are you most worried about right now and why? Blockchain-based solutions can both enable self-sovereignty or be ultimate surveillance tools for entities and governments which value individual freedoms less than the control of their subjects (see my article on ). The Folly of Digital Fiat 6. If we gave you 10 million dollars to invest in something today, what would you invest in and why? Disruptive finance solutions that replace the middleman, and Web 3.0 primitives that make centralized institutions unnecessary (read more ). here 7. What are you currently learning? We are currently writing a book on . As strict observers of the scientific methods, we started out aggregating the current financial landscape three years ago. We are learning something new about the evolving financial landscape daily. However, we are interested in a wide spectrum at the intersection of finance, technology, law, and socio-economic issues. Streaming Money 8. What’s the best advice you’ve ever given someone? Never choose unhappiness over uncertainty! 9. What’s the best advice you’ve ever received? A. The meaning of your communication is the response you get. B. There is no failure, only feedback. C. The map is not the territory. About HackerNoon’s 2021 Noonie Awards The annual Noonie Awards celebrate the best and brightest of the tech industry, bringing together all who are making the Internet and the world of tech what it is today. Please be sure to , nominate, and vote for the people and companies who you think are making the biggest impact on the tech industry today. check out our award categories

"There is no Failure, only Feedback"  with 5x Noonies Winner Christian Kameir

Negative sentiment around Trump lifted slightly when he was diagnosed — and dropped when he returned to the White House. During the September 29th, 2020 Presidential debate held in Cleveland, OH, Donald Trump said to Joe Biden: “They give you good press, they give me bad press because that’s the way it is, unfortunately” — Donald Trump Until now, this assertion by Trump has been hard-to-impossible to prove or disprove. But a new media study released by Import.io has shown that the POTUS is correct in this assertion — to a degree. The analyzed 49,682 news stories from the web sites of 1,571 U.S.-based, English-language news organizations. This huge data set represents coverage through September 1st through October 7th that mentions Trump, Biden, or Trump and Biden. The data was measured in terms of both volume and tone. study by Import.io Here’s what they found: Trump & Biden Coverage — Not Overwhelmingly Positive Neither candidate has predominately positive coverage, but Trump’s is certainly more negative in tone. The timeline shows Biden in Blue and Trump in Red. Certain events like Trump praising SCOTUS Ruth Bader Ginsburg and Trump’s own diagnosis with COVID-19 did create a temporary lift in sentiment toward Trump — but it was short-lived. Media sentiment timeline — Trump vs. Biden TL;DR analysis — Neither candidate has greatly positive coverage but Trump’s coverage is more negative and swings more wildly around key events. Trump gets far more coverage… Since September, there have been 5x the number of stories about only Trump vs. stories about only Biden. If you factor in stories that mentioned both candidates, Trump still gets x2 the amount of coverage. Volume of coverage — Trump vs. Biden Trump may not evince great sentiment from media, but he is great at being talked about. TL;DR analysis — More facts to consider: The media sentiment towards both candidates took a negative turn immediately following the first Presidential debate.When Trump was diagnosed with COVID-19 on Oct. 2, it was the biggest day in terms of media coverage of the 2020 US Presidential election, followed by September 30th, the day after the first Presidential debate.After the announcement that Trump contracted COVID-19, coverage spiked for Trump (but not for Trump/Biden or Biden alone) and Trump’s sentiment score reached an all-time high.The elevated media sentiment score for Trump continued in the days during his hospitalization but then dropped again upon his return to the White House Conclusion As pundits on both sides marvel at the mass appeal of this highly polarizing President, these figures beg the question — is it true that It remains to be seen, and the remaining weeks running up to the campaign are likely to be full of surprises. all news is good news? Import.io extracts and analyzes web data at scale for global companies. To learn more about the methodology of the study, . check out the company’s blog entry *Article uses original composite image made using campaign web site imagery.

50k+ News Story Study Finds Trump Gets 5x the Coverage, but Not Much Love

Photo by on Sanjeevan SatheesKumar Unsplash Recently my team have started working on the first big React Native app. Quite soon I had to implement a page with filters list: Problem In the initial version of the page, a single click on checkbox element initiated a noticeable delay. takes most of FlatList[update] 250ms There are 75 elements in the list and only 11 visible at a single moment of time. So I was expecting (which has a virtualization feature out of the box) to rerender only a single changed element (in the worst case — only visible elements). But it . FlatList rerenders all items To show a selectable list of categories I retrieve from redux state and create a new array that contains Category data and flag : Original Approach Categories extendedCategories isShown Solution After the small change, render time went down to 56ms. To , I had to stop mutating data source for FlatList and rerender items ONLY if they have changed. cut down render time 5x times Instead of creating a new data source object after each rerender, I use existing array to map over. And calculate on the level. Now the data source is always the same object and prop will get new value only when checkbox changes it’s value. categories isShown FlatList isShown 2. Use for PureComponent renderItem is similar to . The difference between them is that doesn’t implement , but implements it with a shallow prop and state comparison. ’s only shallowly compares the objects. If these contain complex data structures, it may produce false-negatives for deeper differences. PureComponent Component Component shouldComponentUpdate() PureComponent PureComponent shouldComponentUpdate() Redux State — an array of all meta info about categories Categories — an array with only selected categories {id, name} SelectedCategories After clicking on the category, it is added to array, and object is always the same, without mutations. If the user clicks on an already selected category, it is being removed from the array. SelectedCategories Categories SelectedCategories Profiling From my experience, unlike React, extra rerender in React Native affects user experience much worse. The fastest way to find extra renders is old in render function. After you found an issue, use Chrome Profiler tab in developer tools to dig into it. Or (coming in 16.5). console.log new React Profiler Want to play with the code? Check out the GitHub repo! _Contribute to ProductCrafters/rn_perf development by creating an account on GitHub._github.com ProductCrafters/rn_perf P.S. If you enjoyed this article and want more like these, please clap and share with friends that may need it. 🚀 . We help US startup founders to bring their ideas to life. If you need some help, send a message ✉️ My team uses JS and React to build production apps over 3 years oleg@productcrafters.io

How to improve React Native list performance 5x times

Here’s a quick(ish) tip on how to use non-recurring revenue services (training, consulting, etc.) to 2.5x (or more) the recurring revenue you get from a customer. This tip requires that you have a baseline understanding of Customer Success . Let’s say you have a customer at $10k/ARR and they’re humming along just fine. If you leave them to grow on their own, it might take 18 months before they’re ready for the next pricing tier or add-on that would take them to $25k/ARR. By knowing where the customer is on their journey, you can intelligently offer services that can short-circuit their path to the next Success Milestone that has that logical expansion opportunity attached to it, and achieve account expansion much faster than if you left them to grow organically. Let’s say they hit a at 6 months in (remember, each customer will likely hit that milestone on their own cadence), and you know that a training program would not only help their current active users get more value from your product but could increase the number of users in their company, so you offer that up for a fee, but maybe only to break even. Success Milestone And that training program would get them to that point in 3 months, meaning they’re ready to move up to the next pricing tier 9 months earlier than they would organically. In businesses with professional investors and a board of directors, a large percentage of non-recurring revenue is often frowned upon since it’s not predictable and valued lower by said investors or potential acquirers. HOLD UP! SaaS So generally in these types of companies you want the “attachment rate” of non-RR to RR to be low (whatever that means to your investors, board, etc.). If that’s you, don’t immediately dismiss what I’m saying here… in fact, keep reading. BTW, in bootstrapped or other businesses, revenue is revenue so don’t worry about attachment rates (until you go to sell, I suppose). So let’s say you have a training program that you can offer a customer to short-circuit their success, but you have to charge $2.5k just to break even on it. If the customer is paying $10k, that’s a 25% attachment rate, which is probably higher than most investors/boards want to see. BUT… after the training, the most logical next step for the customer is to buy additional seats, add-ons, etc. and they move to $25k/ARR, now that attachment rate is only 10%… now you’re cookin’ with evil gas. The moral of this story is… know where your customers are on their journey toward their Desired Outcome, proactively offer them things to help them grow, and when they take that AND grow, offer them the thing they’ve grown into. Oh, and since they expanded 9 months earlier than they would have organically, where might they be in the next 9 months? You might be able to 5x+ an account value in that first 18 months. Hmm…. ….. Lincoln Murphy is a world-renowned Architect, Consultant, Author, and Keynote Speaker. As founder of — and previously leader of Customer Success Evangelism at Gainsight — he’s used Customer Success to drive growth across the entire customer lifecycle for more than 400 SaaS and enterprise software companies over the last decade. Growth Sixteen Ventures Follow Lincoln on Twitter

How to Quickly 2.5x Your Recurring Revenue

From OpenAI to Closed AI: Custom Chips Are Closing The Doors—What’s Next? 

Ад OpenAI да закрытага AI: Карыстальніцкія мікрасхемы зачыняюць дзверы - што далей? 

De l'OpenAI a la IA tancada: els xips personalitzats estan tancant les portes: què passa? 

Von OpenAI zu Closed AI: Kundenspezifische Chips schließen die Türen – was kommt als Nächstes?

Από το OpenAI στο Closed AI: Τα προσαρμοσμένα τσιπ κλείνουν τις πόρτες—Τι ακολουθεί; 

De la IA abierta a la IA cerrada: los chips personalizados están cerrando las puertas. ¿Qué sigue?

Az OpenAI-tól a zárt mesterséges intelligenciaig: Az egyedi chipek bezárják az ajtókat – mi a következő lépés? 

Dari OpenAI ke AI Tertutup: Chip Kustom Menutup Pintu—Apa Selanjutnya?

OpenAI からクローズド AI へ: カスタム チップが扉を閉ざす — 次は何が起こるのか?

Kobanda na OpenAI tii na AI oyo ekangami: Ba puces personnalisées ezali kokanga ba portes —What’s Next? 

له OpenAI څخه تړل شوي AI ته: دودیز چپس دروازې بندوي — بیا څه دي؟ 

От открытого ИИ к закрытому ИИ: кастомные чипы закрывают двери — что дальше?

Від відкритого штучного інтелекту до закритого штучного інтелекту: спеціальні мікросхеми закривають двері — що далі? 

I think OpenAI is not being honest about the diminishing returns of scaling AI with data and compute alone. I think they are also putting a lot of the economy, the world and this entire industry in jeopardy by not talking more openly about the topic. At first, I believed what they told us, that all you need to do is add more compute power and more data, and LLMs as well as other models will simply get better. That this relationship between the models, their compute and data could grow linearly until the end of time. The leap from GPT-3 and GPT-3.5 was immense. The leap from GPT-3.5 to GPT-4 seemed like clear evidence that this presumption was correct. But then things got weird. Instead of releasing a model called GPT-5 or even GPT-4.5, they released GPT-4-turbo. GPT-4-turbo is not as intelligent as GPT-4 but it is much faster and it's cheaper. That all makes sense. But then, this trend kept going. After GPT-4-turbo, OpenAI's next release was GPT-4o. GPt-4o is more or less just as intelligent as GPT-4-turbo, but it is even faster and even cheaper. The functionality that really sold us, however, was its ability to talk and understand things via audio and its speed. However, take note, at this point in our story, GPT-4-turbo is not more intelligent than GPT-4 and GPT-4o is not more intelligent than GPT-4-turbo. And none of them are more intelligent than GPT-4. Their next and most recent release was GPT-o1. GPT-o1 can perform better than GPT-4 onsome tasks. But that's because o1 is not really a single model. GPT-o1 is actually a black box of multiple lightweight LLM models working together. Perhaps o1 is even better described as software or middleware than it is an actual model. You give it a question, it comes up with an answer, and then it repeatedly uses other models tasked with checking the answer to make sure it’s right, and it disguises all of these operations. It does all of this very, very quickly. Why not just make an LLM that's more powerful than GPT-4? Why resort to such cloak-and-dagger techniques to achieve new releases? GPT-4 came out 2 years ago, we should be well beyond its capabilities by now. Well, Noam Brown, a researcher at OpenAI had something to say on why they went this route with o1 at TED AI. He said “It turned out that having a bot think for just 20 seconds in a hand of poker got the same boosting performance as scaling up the model by 100,000x and training it for 100,000 times longer,” Now stop and really think about what is being said there. A bot thinking for 20 seconds is as good as a bot trained 100,000 times longer with 100,000 times more computing power.  If the scaling laws are infinite, that math is impossible. Something is either wrong here or someone is lying. Why does all of this matter? OpenAI is worth 150 billion dollars and the majority of that market cap is based on projections that depend on the improvement of models over time. If AI is only as good as it is today, that’s still an interesting future, but that’s not what’s being sold to investors by AI companies whose entire IP is their model. That also changes the product roadmap of many other companies that depend on the continued advancement of their LLMs to build their own products. OpenAI’s goal and ambitions of AGI are severely delayed if this is all true. A Hypothesis The reason LLMs are so amazing is because of a higher level philosophical phenomenon that we never considered, that language inherently possesses an extremely large amount of context and data about the world within even small sections of text. Unlike pixels in a picture or video, words in a sentence implicitly describe one another. A completely cohesive sentence is by definition, “rational”. Whether or not it’s true is a very different story and a problem that transcends language alone. No matter how much text you consume, “truth” and “falsehoods” are not simply linguistic concepts. You can say something is completely rational but in no way “true”. It is at this point LLMs will consistently hit a brick wall. Over the last 12 months, I’d like to formally speculate that behind closed doors there have been no huge leaps in LLMs at OpenAI, GrokAI, or at Google. To be specific I don’t think anyone, anywhere has made any LLM that is even 1.5X better than GPT-4. At OpenAI it seems that high-level staff are quitting. Right now they’re saying it’s because of safety but I’m going to put my tinfoil hat on now and throw an idea out there. They are aware of this issue and they’re jumping ship before it’s too late. Confirmation I started discussing this concern with friends 3 months ago. I was called many names haha. But in the last 3 weeks, a lot of the press has begun to smell something fishy: OpenAI is no longer releasing Orion (GPT-5) because it did not meet expected performance benchmarks and it is seeing diminishing returns. (https://www.theinformation.com/articles/openai-shifts-strategy-as-rate-of-gpt-ai-improvements-slows)


Bloomberg reports that OpenAI, Google and Anthropic are all having struggles making more advanced AI. (https://www.bloomberg.com/news/articles/2024-11-13/openai-google-and-anthropic-are-struggling-to-build-more-advanced-ai) What can we do about it? It’s hard to recommend a single solution. The tech behind o1 is proof that even low-performance models can be repurposed to do complicated operations. But that is not a solution to the problem of AI scaling. I think there needs to be substantial investment and rapid testing of new model architectures. We also have run out of data and need new ways of extrapolating usable data for LLMs to be trained on. Perhaps using multidimensional labeling that helps guide its references for truthful information directly. Another good idea could be to simply continue fine-tuning LLMs for specific use cases like math, science, and healthcare running and using AI agent workflows, similar to o1. It might give a lot of companies wiggle room until a new architecture arises. This problem is really bad but I think that the creativity in machine learning and software development it will inspire will be immense. Once we get over this hurdle, we’ll certainly be well on schedule for AGI and perhaps ASI. I think OpenAI is not being honest about the diminishing returns of scaling AI with data and compute alone. I think they are also putting a lot of the economy, the world and this entire industry in jeopardy by not talking more openly about the topic. At first, I believed what they told us, that all you need to do is add more compute power and more data, and LLMs as well as other models will simply get better. That this relationship between the models, their compute and data could grow linearly until the end of time. The leap from GPT-3 and GPT-3.5 was immense. The leap from GPT-3.5 to GPT-4 seemed like clear evidence that this presumption was correct. But then things got weird. Instead of releasing a model called GPT-5 or even GPT-4.5, they released GPT-4-turbo. GPT-4-turbo is not as intelligent as GPT-4 but it is much faster and it's cheaper. That all makes sense. But then, this trend kept going. After GPT-4-turbo, OpenAI's next release was GPT-4o. GPt-4o is more or less just as intelligent as GPT-4-turbo, but it is even faster and even cheaper. The functionality that really sold us, however, was its ability to talk and understand things via audio and its speed. However, take note, at this point in our story, GPT-4-turbo is not more intelligent than GPT-4 and GPT-4o is not more intelligent than GPT-4-turbo. And none of them are more intelligent than GPT-4. Their next and most recent release was GPT-o1. GPT-o1 can perform better than GPT-4 on some tasks. But that's because o1 is not really a single model. GPT-o1 is actually a black box of multiple lightweight LLM models working together. Perhaps o1 is even better described as software or middleware than it is an actual model. You give it a question, it comes up with an answer, and then it repeatedly uses other models tasked with checking the answer to make sure it’s right, and it disguises all of these operations. It does all of this very, very quickly. some Why not just make an LLM that's more powerful than GPT-4? Why resort to such cloak-and-dagger techniques to achieve new releases? GPT-4 came out 2 years ago, we should be well beyond its capabilities by now. Well, Noam Brown, a researcher at OpenAI had something to say on why they went this route with o1 at TED AI. He said “It turned out that having a bot think for just 20 seconds in a hand of poker got the same boosting performance as scaling up the model by 100,000x and training it for 100,000 times longer,” Now stop and really think about what is being said there. A bot thinking for 20 seconds is as good as a bot trained 100,000 times longer with 100,000 times more computing power.  If the scaling laws are infinite, that math is impossible. Something is either wrong here or someone is lying. Why does all of this matter? OpenAI is worth 150 billion dollars and the majority of that market cap is based on projections that depend on the improvement of models over time. If AI is only as good as it is today, that’s still an interesting future, but that’s not what’s being sold to investors by AI companies whose entire IP is their model. That also changes the product roadmap of many other companies that depend on the continued advancement of their LLMs to build their own products. OpenAI’s goal and ambitions of AGI are severely delayed if this is all true. A Hypothesis The reason LLMs are so amazing is because of a higher level philosophical phenomenon that we never considered, that language inherently possesses an extremely large amount of context and data about the world within even small sections of text. Unlike pixels in a picture or video, words in a sentence implicitly describe one another. A completely cohesive sentence is by definition, “rational”. Whether or not it’s true is a very different story and a problem that transcends language alone. No matter how much text you consume, “truth” and “falsehoods” are not simply linguistic concepts. You can say something is completely rational but in no way “true”. It is at this point LLMs will consistently hit a brick wall. Over the last 12 months, I’d like to formally speculate that behind closed doors there have been no huge leaps in LLMs at OpenAI, GrokAI, or at Google. To be specific I don’t think anyone, anywhere has made any LLM that is even 1.5X better than GPT-4. At OpenAI it seems that high-level staff are quitting. Right now they’re saying it’s because of safety but I’m going to put my tinfoil hat on now and throw an idea out there. They are aware of this issue and they’re jumping ship before it’s too late. Confirmation Confirmation I started discussing this concern with friends 3 months ago. I was called many names haha. But in the last 3 weeks, a lot of the press has begun to smell something fishy: OpenAI is no longer releasing Orion (GPT-5) because it did not meet expected performance benchmarks and it is seeing diminishing returns. (https://www.theinformation.com/articles/openai-shifts-strategy-as-rate-of-gpt-ai-improvements-slows) Bloomberg reports that OpenAI, Google and Anthropic are all having struggles making more advanced AI. (https://www.bloomberg.com/news/articles/2024-11-13/openai-google-and-anthropic-are-struggling-to-build-more-advanced-ai) OpenAI is no longer releasing Orion (GPT-5) because it did not meet expected performance benchmarks and it is seeing diminishing returns. (https://www.theinformation.com/articles/openai-shifts-strategy-as-rate-of-gpt-ai-improvements-slows) OpenAI is no longer releasing Orion (GPT-5) because it did not meet expected performance benchmarks and it is seeing diminishing returns. ( https://www.theinformation.com/articles/openai-shifts-strategy-as-rate-of-gpt-ai-improvements-slows ) OpenAI is no longer releasing Orion (GPT-5) because it did not meet expected performance benchmarks and it is seeing diminishing returns. https://www.theinformation.com/articles/openai-shifts-strategy-as-rate-of-gpt-ai-improvements-slows Bloomberg reports that OpenAI, Google and Anthropic are all having struggles making more advanced AI. (https://www.bloomberg.com/news/articles/2024-11-13/openai-google-and-anthropic-are-struggling-to-build-more-advanced-ai) Bloomberg reports that OpenAI, Google and Anthropic are all having struggles making more advanced AI. ( https://www.bloomberg.com/news/articles/2024-11-13/openai-google-and-anthropic-are-struggling-to-build-more-advanced-ai ) Bloomberg reports that OpenAI, Google and Anthropic are all having struggles making more advanced AI. https://www.bloomberg.com/news/articles/2024-11-13/openai-google-and-anthropic-are-struggling-to-build-more-advanced-ai What can we do about it? It’s hard to recommend a single solution. The tech behind o1 is proof that even low-performance models can be repurposed to do complicated operations. But that is not a solution to the problem of AI scaling. I think there needs to be substantial investment and rapid testing of new model architectures. We also have run out of data and need new ways of extrapolating usable data for LLMs to be trained on. Perhaps using multidimensional labeling that helps guide its references for truthful information directly. Another good idea could be to simply continue fine-tuning LLMs for specific use cases like math, science, and healthcare running and using AI agent workflows, similar to o1. It might give a lot of companies wiggle room until a new architecture arises. This problem is really bad but I think that the creativity in machine learning and software development it will inspire will be immense. Once we get over this hurdle, we’ll certainly be well on schedule for AGI and perhaps ASI. software development

Is AI Progress Slowing? The Scaling Debate OpenAI Doesn’t Want to Have

¿Se está ralentizando el progreso de la IA? El debate sobre la escalabilidad que OpenAI no quiere tener

האם התקדמות הבינה המלאכותית מואטת? הדיון בקנה מידה ש-OpenAI לא רוצה שיהיה

Apakah Kemajuan AI Melambat? Perdebatan Skala yang Tidak Diinginkan OpenAI

Il progresso dell'intelligenza artificiale sta rallentando? Il dibattito sulla scalabilità che OpenAI non vuole avere

AIの進歩は鈍化しているのか？OpenAIが望まないスケーリングの議論

Bokóli ya AI ezali kokita? Débat ya Échelle OpenAI Elingi Ezala Te

Vertraagt de vooruitgang van AI? Het schaaldebat dat OpenAI niet wil hebben

Spomaľuje sa pokrok AI? Debata o škálovaní OpenAI nechce mať

Saktar AI-utvecklingen ner? Skalningsdebatten som OpenAI inte vill ha

AI ösüşi haýalmy? Giňeldilen jedel OpenAI islemeýär

Чи сповільнюється розвиток ШІ? Дебати щодо масштабування, яких OpenAI не хоче мати

Ingabe Intuthuko ye-AI Iyahamba Kancane? I-Scaling Debate OpenAI Ayifuni Ukuba Nayo

Web Scraping Optimization: Tips for Faster, Smarter Scrapers

Web Scraping Optimization: Daha Sürətli, Daha Ağıllı Scrapers üçün Məsləhətlər

Optimización del web scraping: consejos para que los scrapers sean más rápidos e inteligentes

Web スクレイピングの最適化: より高速でスマートなスクレイパーのためのヒント

Žiniatinklio grandymo optimizavimas: greitesnių ir išmanesnių grandiklių patarimai

Fanamafisana ny fakosoka amin'ny Internet: Toro-hevitra ho an'ny mpikirakira haingana kokoa

वेब स्क्र्यापिंग अप्टिमाइजेसन: छिटो, स्मार्ट स्क्र्यापरहरूको लागि सुझाव

Optymalizacja Web Scraping: Wskazówki dotyczące szybszych i inteligentniejszych Scraperów

Gukuramo Urubuga Optimisiyoneri: Inama zo Kwihuta, Byoroheje

Оптимизација веб-стругања: савети за брже и паметније стругаче

Оптимизатсияи веб скрепинг: Маслиҳатҳо барои скреперҳои тезтар ва оқилтар

Web Scraping Optimization: Mga Tip para sa Mas Mabilis, Mas Matalino na mga Scraper

I-Web Scraping Optimization: Amathiphu Okwenza Ama-Scrapers Asheshayo, Ahlakaniphile

The "Smart Person" Trap and How to Avoid It

La trampa de la “persona inteligente” y cómo evitarla

"Pertsona adimenduna" tranpa eta nola saihestu

מלכודת "האדם החכם" וכיצד להימנע ממנה

«Խելացի մարդու» ծուղակը և ինչպես խուսափել դրանից

「賢い人」の罠とその回避方法

«Ақылды адам» тұзағы және одан қалай құтылуға болады

Стапица „паметна личност“ и како да ја избегнете

Sefero sa "Motho yo bohlale" le Kamoo o ka se Phemago

Ловушка «умного человека» и как ее избежать

Kurthi i "Personit të zgjuar" dhe si ta shmangni atë

กับดักของ “คนฉลาด” และวิธีหลีกเลี่ยง

"سمارٹ پرسن" کا جال اور اس سے کیسے بچنا ہے۔

Hey Hackers, Siete has been nominated in HackerNoon's annual Startups of The Year awards in Mexico. Please vote for us here: https://hackernoon.com/startups/industry/business-development?stup=66f31bf46ea4677e23db4fd2&embedable=true Read more about us below to understand why we deserve your vote. Meet Siete Siete is a B2B lead generation agency that specializes in helping businesses connect with qualified prospects through a data-driven, multi-channel approach. We streamline the prospecting process, delivering high-quality leads that drive sales and growth for our clients across various industries. Our goal is simple: to help companies achieve scalable, consistent growth by taking the pain out of lead generation. How Siete achieved product-market fit At Siete, we know we have achieved product-market-fit because in two years we have scale to more than 8 countries growing 30% MoM in revenue and new clients with little to none churn. All of that 100% bootstrapped. 4.5x Growth on Revenue (2023 vs 2024).
3x Growth on EBITDA (2023 vs 2024).
4.8/5 of client Satisfaction.
6x Growth on Headcount. The evolution of Siete Building Siete has been a journey of continuous iteration. We started with a simple model of lead generation with a flat fee and now we have two huge services (Staff Augmentation for comercial teams and Lead Generation 360). Both services are for different targets of companies and different budgets.
We were only serving enterprise clients with our lead gen 360 service and a lot of smaller companies where talking with us because they also needed a similar service but more flexible and thats why we launched staffing. Today staffing is 40% of Siete's revenue and growing rapidly.
Our clients’ feedback has been instrumental in shaping our priorities. For example, many businesses expressed a need for deeper analytics and reporting, which led us to develop more robust tracking and performance monitoring tools. Listening to our clients has helped us stay agile and adaptable. Milestones and achievements Generating over 10,000 qualified leads for our clients in the last year.
Expanding our team and operations to serve clients in over eight countries.
Being recognized as one of the top agencies for B2B lead generation in Latam. Lessons learned One of our biggest challenges has been navigating the rapidly changing landscape of digital marketing and prospecting. Early on, we learned the importance of diversifying our lead generation strategies and staying ahead of emerging trends. Adapting quickly has been essential for our success, and the lesson we’ve learned is to always keep innovating, especially when the market shifts. What Startups of The Year means to us Participating in HackerNoon’s Startups of The Year is an incredible opportunity for us to showcase the hard work and dedication that our team puts into building a service that truly adds value. It’s also a chance to gain visibility in the global startup ecosystem and connect with other innovative companies. Winning this recognition would solidify our position as a leader in the B2B lead generation space, while also opening new doors for growth and collaboration. Conclusion At Siete, we’re passionate about helping businesses grow by connecting them with the right opportunities. Our journey has been one of relentless focus on results and constant evolution based on feedback and market needs. We’re thrilled to be nominated, and with your support, we can take our next big step forward. Vote for us today! Welcome to HackerNoon’s Startups of The Year 2024 interview series, spotlighting startups from around the world that are changing the way we live, one solution at a time. If you’ve been nominated, create a brand profile and answer these questions here.
Want to nominate a startup? Follow the steps here.Looking for inspiration? Check out how other startups responded to last year’s Startups of The Year interview questions and examples of company responses here, here, and here. Hey Hackers, Siete has been nominated in HackerNoon's annual Startups of The Year awards in Mexico. Please vote for us here: Please vote for us here: https://hackernoon.com/startups/industry/business-development?stup=66f31bf46ea4677e23db4fd2&embedable=true https://hackernoon.com/startups/industry/business-development?stup=66f31bf46ea4677e23db4fd2&embedable=true Read more about us below to understand why we deserve your vote. Meet Siete Siete is a B2B lead generation agency that specializes in helping businesses connect with qualified prospects through a data-driven, multi-channel approach. We streamline the prospecting process, delivering high-quality leads that drive sales and growth for our clients across various industries. Our goal is simple: to help companies achieve scalable, consistent growth by taking the pain out of lead generation. How Siete achieved product-market fit At Siete, we know we have achieved product-market-fit because in two years we have scale to more than 8 countries growing 30% MoM in revenue and new clients with little to none churn. All of that 100% bootstrapped. 4.5x Growth on Revenue (2023 vs 2024). 3x Growth on EBITDA (2023 vs 2024). 4.8/5 of client Satisfaction. 6x Growth on Headcount. 4.5x Growth on Revenue (2023 vs 2024). 3x Growth on EBITDA (2023 vs 2024). 4.8/5 of client Satisfaction. 6x Growth on Headcount. The evolution of Siete Building Siete has been a journey of continuous iteration. We started with a simple model of lead generation with a flat fee and now we have two huge services (Staff Augmentation for comercial teams and Lead Generation 360). Both services are for different targets of companies and different budgets. We were only serving enterprise clients with our lead gen 360 service and a lot of smaller companies where talking with us because they also needed a similar service but more flexible and thats why we launched staffing. Today staffing is 40% of Siete's revenue and growing rapidly. Our clients’ feedback has been instrumental in shaping our priorities. For example, many businesses expressed a need for deeper analytics and reporting, which led us to develop more robust tracking and performance monitoring tools. Listening to our clients has helped us stay agile and adaptable. Building Siete has been a journey of continuous iteration. We started with a simple model of lead generation with a flat fee and now we have two huge services (Staff Augmentation for comercial teams and Lead Generation 360). Both services are for different targets of companies and different budgets. We were only serving enterprise clients with our lead gen 360 service and a lot of smaller companies where talking with us because they also needed a similar service but more flexible and thats why we launched staffing. Today staffing is 40% of Siete's revenue and growing rapidly. Our clients’ feedback has been instrumental in shaping our priorities. For example, many businesses expressed a need for deeper analytics and reporting, which led us to develop more robust tracking and performance monitoring tools. Listening to our clients has helped us stay agile and adaptable. Milestones and achievements Generating over 10,000 qualified leads for our clients in the last year. Expanding our team and operations to serve clients in over eight countries. Being recognized as one of the top agencies for B2B lead generation in Latam. Generating over 10,000 qualified leads for our clients in the last year. Expanding our team and operations to serve clients in over eight countries. Being recognized as one of the top agencies for B2B lead generation in Latam. Lessons learned One of our biggest challenges has been navigating the rapidly changing landscape of digital marketing and prospecting. Early on, we learned the importance of diversifying our lead generation strategies and staying ahead of emerging trends. Adapting quickly has been essential for our success, and the lesson we’ve learned is to always keep innovating, especially when the market shifts. One of our biggest challenges has been navigating the rapidly changing landscape of digital marketing and prospecting. Early on, we learned the importance of diversifying our lead generation strategies and staying ahead of emerging trends. Adapting quickly has been essential for our success, and the lesson we’ve learned is to always keep innovating, especially when the market shifts. What Startups of The Year means to us Participating in HackerNoon’s Startups of The Year is an incredible opportunity for us to showcase the hard work and dedication that our team puts into building a service that truly adds value. It’s also a chance to gain visibility in the global startup ecosystem and connect with other innovative companies. Winning this recognition would solidify our position as a leader in the B2B lead generation space, while also opening new doors for growth and collaboration. Participating in HackerNoon’s Startups of The Year is an incredible opportunity for us to showcase the hard work and dedication that our team puts into building a service that truly adds value. It’s also a chance to gain visibility in the global startup ecosystem and connect with other innovative companies. Winning this recognition would solidify our position as a leader in the B2B lead generation space, while also opening new doors for growth and collaboration. Conclusion At Siete, we’re passionate about helping businesses grow by connecting them with the right opportunities. Our journey has been one of relentless focus on results and constant evolution based on feedback and market needs. We’re thrilled to be nominated, and with your support, we can take our next big step forward. Vote for us today! Vote for us today ! Vote for us today Welcome to HackerNoon’s Startups of The Year 2024 interview series, spotlighting startups from around the world that are changing the way we live, one solution at a time. If you’ve been nominated, create a brand profile and answer these questions here. Want to nominate a startup? Follow the steps here. Looking for inspiration? Check out how other startups responded to last year’s Startups of The Year interview questions and examples of company responses here, here, and here. Welcome to HackerNoon’s Startups of The Year 2024 interview series, spotlighting startups from around the world that are changing the way we live, one solution at a time. If you’ve been nominated, create a brand profile and answer these questions here . Startups of The Year 2024 create a brand profile create a brand profile create a brand profile here here Want to nominate a startup? Follow the steps here . here here Looking for inspiration? Check out how other startups responded to last year’s Startups of The Year interview questions and examples of company responses here, here, and here. Looking for inspiration? Check out how other startups responded to last year’s Startups of The Year interview questions and examples of company responses here , here , and here . last year’s Startups of The Year interview questions last year’s Startups of The Year interview questions here here here here here here

Meet Siete, Startups of the Year 2024 Nominee

Table of Links Abstract and 1. Introduction 2 Architectural details and 2.1 Sparse Mixture of Experts 3 Results 3.1 Multilingual benchmarks, 3.2 Long range performance, and 3.3 Bias Benchmarks 4 Instruction Fine-tuning 5 Routing analysis 6 Conclusion, Acknowledgements, and References 3 Results We compare Mixtral to Llama, and re-run all benchmarks with our own evaluation pipeline for fair comparison. We measure performance on a wide variety of tasks categorized as follow: • Commonsense Reasoning (0-shot): Hellaswag [32], Winogrande [26], PIQA [3], SIQA [27], OpenbookQA [22], ARC-Easy, ARC-Challenge [8], CommonsenseQA [30] • World Knowledge (5-shot): NaturalQuestions [20], TriviaQA [19] • Reading Comprehension (0-shot): BoolQ [7], QuAC [5] • Math: GSM8K [9] (8-shot) with maj@8 and MATH [17] (4-shot) with maj@4 • Code: Humaneval [4] (0-shot) and MBPP [1] (3-shot) • Popular aggregated results: MMLU [16] (5-shot), BBH [29] (3-shot), and AGI Eval [34] (3-5-shot, English multiple-choice questions only) Detailed results for Mixtral, Mistral 7B and Llama 2 7B/13B/70B and Llama 1 34B[2] are reported in Table 2. Figure 2 compares the performance of Mixtral with the Llama models in different categories. Mixtral surpasses Llama 2 70B across most metrics. In particular, Mixtral displays a superior performance in code and mathematics benchmarks. Size and Efficiency. We compare our performance to the Llama 2 family, aiming to understand Mixtral models’ efficiency in the cost-performance spectrum (see Figure 3). As a sparse Mixtureof-Experts model, Mixtral only uses 13B active parameters for each token. With 5x lower active parameters, Mixtral is able to outperform Llama 2 70B across most categories. Note that this analysis focuses on the active parameter count (see Section 2.1), which is directly proportional to the inference compute cost, but does not consider the memory costs and hardware utilization. The memory costs for serving Mixtral are proportional to its sparse parameter count, 47B, which is still smaller than Llama 2 70B. As for device utilization, we note that the SMoEs layer introduces additional overhead due to the routing mechanism and due to the increased memory loads when running more than one expert per device. They are more suitable for batched workloads where one can reach a good degree of arithmetic intensity. Comparison with Llama 2 70B and GPT-3.5. In Table 3, we report the performance of Mixtral 8x7B compared to Llama 2 70B and GPT-3.5. We observe that Mixtral performs similarly or above the two other models. On MMLU, Mixtral obtains a better performance, despite its significantly smaller capacity (47B tokens compared to 70B). For MT Bench, we report the performance of the latest GPT-3.5-Turbo model available, gpt-3.5-turbo-1106. Evaluation Differences. On some benchmarks, there are some differences between our evaluation protocol and the one reported in the Llama 2 paper: 1) on MBPP, we use the hand-verified subset 2) on TriviaQA, we do not provide Wikipedia contexts. This paper is available on arxiv under CC 4.0 license. [2] Since Llama 2 34B was not open-sourced, we report results for Llama 1 34B. Authors:
(1) Albert Q. Jiang;
(2) Alexandre Sablayrolles;
(3) Antoine Roux;
(4) Arthur Mensch;
(5) Blanche Savary;
(6) Chris Bamford;
(7) Devendra Singh Chaplot;
(8) Diego de las Casas;
(9) Emma Bou Hanna;
(10) Florian Bressand;
(11) Gianna Lengyel;
(12) Guillaume Bour;
(13) Guillaume Lample;
(14) Lélio Renard Lavaud;
(15) Lucile Saulnier;
(16) Marie-Anne Lachaux;
(17) Pierre Stock;
(18) Sandeep Subramanian;
(19) Sophia Yang;
(20) Szymon Antoniak;
(21) Teven Le Scao;
(22) Théophile Gervet;
(23) Thibaut Lavril;
(24) Thomas Wang;
(25) Timothée Lacroix;
(26) William El Sayed. Table of Links Abstract and 1. Introduction Abstract and 1. Introduction 2 Architectural details and 2.1 Sparse Mixture of Experts 2 Architectural details and 2.1 Sparse Mixture of Experts 3 Results 3 Results 3.1 Multilingual benchmarks, 3.2 Long range performance, and 3.3 Bias Benchmarks 3.1 Multilingual benchmarks, 3.2 Long range performance, and 3.3 Bias Benchmarks 4 Instruction Fine-tuning 4 Instruction Fine-tuning 5 Routing analysis 5 Routing analysis 6 Conclusion, Acknowledgements, and References 6 Conclusion, Acknowledgements, and References 3 Results We compare Mixtral to Llama, and re-run all benchmarks with our own evaluation pipeline for fair comparison. We measure performance on a wide variety of tasks categorized as follow: • Commonsense Reasoning (0-shot): Hellaswag [32], Winogrande [26], PIQA [3], SIQA [27], OpenbookQA [22], ARC-Easy, ARC-Challenge [8], CommonsenseQA [30] • Commonsense Reasoning (0-shot): • World Knowledge (5-shot): NaturalQuestions [20], TriviaQA [19] • World Knowledge (5-shot): • Reading Comprehension (0-shot): BoolQ [7], QuAC [5] • Reading Comprehension (0-shot): • Math: GSM8K [9] (8-shot) with maj@8 and MATH [17] (4-shot) with maj@4 • Math: • Code: Humaneval [4] (0-shot) and MBPP [1] (3-shot) • Code: • Popular aggregated results: MMLU [16] (5-shot), BBH [29] (3-shot), and AGI Eval [34] (3-5-shot, English multiple-choice questions only) • Popular aggregated results: Detailed results for Mixtral, Mistral 7B and Llama 2 7B/13B/70B and Llama 1 34B[2] are reported in Table 2. Figure 2 compares the performance of Mixtral with the Llama models in different categories. Mixtral surpasses Llama 2 70B across most metrics. In particular, Mixtral displays a superior performance in code and mathematics benchmarks. Size and Efficiency. We compare our performance to the Llama 2 family, aiming to understand Mixtral models’ efficiency in the cost-performance spectrum (see Figure 3). As a sparse Mixtureof-Experts model, Mixtral only uses 13B active parameters for each token. With 5x lower active parameters, Mixtral is able to outperform Llama 2 70B across most categories. Size and Efficiency. Note that this analysis focuses on the active parameter count (see Section 2.1), which is directly proportional to the inference compute cost, but does not consider the memory costs and hardware utilization. The memory costs for serving Mixtral are proportional to its sparse parameter count, 47B, which is still smaller than Llama 2 70B. As for device utilization, we note that the SMoEs layer introduces additional overhead due to the routing mechanism and due to the increased memory loads when running more than one expert per device. They are more suitable for batched workloads where one can reach a good degree of arithmetic intensity. Comparison with Llama 2 70B and GPT-3.5. In Table 3, we report the performance of Mixtral 8x7B compared to Llama 2 70B and GPT-3.5. We observe that Mixtral performs similarly or above the two other models. On MMLU, Mixtral obtains a better performance, despite its significantly smaller capacity (47B tokens compared to 70B). For MT Bench, we report the performance of the latest GPT-3.5-Turbo model available, gpt-3.5-turbo-1106. Evaluation Differences. On some benchmarks, there are some differences between our evaluation protocol and the one reported in the Llama 2 paper: 1) on MBPP, we use the hand-verified subset 2) on TriviaQA, we do not provide Wikipedia contexts. Evaluation Differences. This paper is available on arxiv under CC 4.0 license. This paper is available on arxiv under CC 4.0 license. available on arxiv [2] Since Llama 2 34B was not open-sourced, we report results for Llama 1 34B. Authors: (1) Albert Q. Jiang; (2) Alexandre Sablayrolles; (3) Antoine Roux; (4) Arthur Mensch; (5) Blanche Savary; (6) Chris Bamford; (7) Devendra Singh Chaplot; (8) Diego de las Casas; (9) Emma Bou Hanna; (10) Florian Bressand; (11) Gianna Lengyel; (12) Guillaume Bour; (13) Guillaume Lample; (14) Lélio Renard Lavaud; (15) Lucile Saulnier; (16) Marie-Anne Lachaux; (17) Pierre Stock; (18) Sandeep Subramanian; (19) Sophia Yang; (20) Szymon Antoniak; (21) Teven Le Scao; (22) Théophile Gervet; (23) Thibaut Lavril; (24) Thomas Wang; (25) Timothée Lacroix; (26) William El Sayed. Authors: Authors: (1) Albert Q. Jiang; (2) Alexandre Sablayrolles; (3) Antoine Roux; (4) Arthur Mensch; (5) Blanche Savary; (6) Chris Bamford; (7) Devendra Singh Chaplot; (8) Diego de las Casas; (9) Emma Bou Hanna; (10) Florian Bressand; (11) Gianna Lengyel; (12) Guillaume Bour; (13) Guillaume Lample; (14) Lélio Renard Lavaud; (15) Lucile Saulnier; (16) Marie-Anne Lachaux; (17) Pierre Stock; (18) Sandeep Subramanian; (19) Sophia Yang; (20) Szymon Antoniak; (21) Teven Le Scao; (22) Théophile Gervet; (23) Thibaut Lavril; (24) Thomas Wang; (25) Timothée Lacroix; (26) William El Sayed.

Mixtral Outperforms Llama and GPT-3.5 Across Multiple Benchmarks

মিক্সট্রাল একাধিক বেঞ্চমার্ক জুড়ে লামা এবং GPT-3.5কে ছাড়িয়ে যায়

Mixtral překonává Llama a GPT-3.5 ve více benchmarkech

Mixtral overgår Llama og GPT-3.5 på tværs af flere benchmarks

Mixtral supera a Llama y GPT-3.5 en múltiples indicadores

Mixtral は複数のベンチマークで Llama と GPT-3.5 を上回る

Mixtral бир нече көрсөткүчтөр боюнча Llama жана GPT-3.5тен ашып кетет

Mixtral ги надминува Llama и GPT-3.5 низ повеќе одредници

Mixtral supera Llama e GPT-3.5 em vários benchmarks

Mixtral tejkalon Llama dhe GPT-3.5 në shumë standarde

Миктрал надмашује Лламу и ГПТ-3.5 у више мерила

Mixtral เหนือกว่า Llama และ GPT-3.5 ในเกณฑ์มาตรฐานต่างๆ

Mixtral ایک سے زیادہ بینچ مارکس میں Llama اور GPT-3.5 سے بہتر کارکردگی کا مظاہرہ کرتا ہے۔

Stories, News & Mentions about 5X

5x’s Stories on HackerNoon