GraceRevive

@airesearchtools Kindly suggest prompt to use instead of uploading the file

I suppose there is a future, not too far away, where we don't read papers, but interact with them. If this is important for researchers, I can't imagine what it could be for society at large. If it is done well, we will have citizens who are informed and who will make more correct decisions. I am not able to quantify it, but for sure it can be important: better health, better planet, better technology, ...

@airesearchtools I think this spoils market for prompt engineering gurus

I am already seeing more than one paper that seems to suggest that it doesn't matter much what kind of prompts we write. The use of CoT prompt engineering does not significantly alter GPT-3.5's effectiveness in handling medical calculations or clinical scenario questions styled like those in USMLE exams. Performance of ChatGPT remains consistent regardless of whether a CoT technique is used instead of direct prompts. This consistency could be instrumental in simplifying the integration of AI tools like ChatGPT into medical education

@airesearchtools Mistral or Llama? Who wins

Mistral 2 does know which number is higher, 9.9 or 9.11. And it is free

@danielarpm @kimmonismus Smart questions. The ai king is coming from somewhere else

@kimmonismus Meh why achieve gold and not say it ? Why play the game of releasing only when others release ? Hubris ? Are they naive to the point that they believe they can forever keep the higher ground ?

A few weeks ago, I discussed with some of you whether AI development is slowing down or whether we have just gotten used to it. I am now convinced that the development of AI has not slowed down, but on the contrary is continuing to accelerate. - OpenAI has probably already achieved gold in the Math Olympiad. Something that even the most benevolent AI researchers would not have expected before 2025. - Essentially, it will be the Q* / Strawberry method that I have repeatedly mentioned - Being able to really do math as Gen-AI instead of reproducing the results by means of probability is an essential prerequisite for AGI. I maintain that AGI has already been achieved internally, just not scaled, and we will probably see it officially unveiled in 2025. That would still be 4 years earlier than Ray Kurzweil would have expected.

@rowancheung It's not free

🚨 Emergency free workshop series: To celebrate the launch of Llama 3.1, my team and I are hosting a free 3-day remote workshop series. Join us starting tomorrow, July 26th, to learn: - What is Llama 3.1, and why is it important - Real-world use cases of using open models - Beginner to advanced integration tips - How to build your own private, local model from your own data All sessions will be live for hands-on learning, but classes will be recorded and posted in the community too. Link below to RSVP.

@emollick AI is becoming everything

Many of the computer scientists who comment a lot on here about the limitations of AI are actually skeptical of the long-term value of LLMs and the companies building & promoting them They do not doubt that "AI" can get better, they just think the current approach is a dead end

@airesearchtools How do teachers use this newly released LLMs without having to install locally on PC? The sizes seem too large and require certain pc specifications to work

With the new releases, we have affordable intelligence available for teachers and students in all universities. Probably, at this moment, all universities should be creating an artificial intelligence laboratory, if they don't have one already.

@JackFawcett1704 Oh 😂😂

@airesearchtools Super fast

New Llama 3 + Groq

@TenHagBall_ United tax is back

Youssouf Fofana to #MUFC for €35M. YES/NO?🔴🇫🇷

@TenHagBall_ @TheAthleticFC United have a better keeper now. He needs to find himself a club

🚨 David De Gea was upset with those he held responsible for his departure. Erik ten Hag and John Murtough. De Gea even told his inner circle that he would be happy to return to United if those two figures were no longer there. [@TheAthleticFC]

@JackFawcett1704 This guy is less progressive with his passes. Chooses safer options all the time

Martin Zubimendi - Last 25 Actions under Pressure twitter.com/JackFawcett123…

@MushtaqBilalPhD Should we now trust AI checker? Are they better than the best models? Kindly list the open source models

Won't recommend doing it...but if someone wants to beat an AI checker, they can easily do it by not using ChatGPT, Claude, or Gemini. Use an open source model and the AI score will drop by 30%-40% even if the text is 100% AI-generated.

@airesearchtools 820GB?🤔

It looks like we already have the new Llama model. I would download it, but I think those 820 GB wouldn't fit on my laptop or my desktop computer. I'll have to wait a bit.

@DeadlineDayLive @sport Lol, this can't be true

🚨 Manchester United could make an offer of around €100M to PSG for Xavi Simons. 💰🤯 (Source: @sport )

@tactixology I didn't see this before. This is well detailed review of how Ten Hag wants to play next season. All transfer targets are tailored to fit this model too. Will it be sustainable? Will United have the lion share of possession?

ETH’S THIRD SEASON: A tactical preview Based on first two preseason games, public quotes and transfer intent, it’s safe to assume Erik Ten Hag’s game model won’t be changing too much this season. His basic United vision is still that of a transition-oriented team, with backline possession, direct progression, strong wing play and pressing as a chance-creation tool. What to expect, then? Well, if things go well, I think we can hope for a repeat of 2022/2023: • top 4 finish with some impressive results but never too dominant in most games • strong cup runs, with a chance of winning another trophy • a very good season by an outlet like Rashford, helped by Zirkzee and Amad facilitation • insane season by Martinez in terms of progression and Case / Ugarte in terms of covering ground & stopping transitions • a lof structural gaps and vulnerabilities masked by individual quality In terms of tactics, this would mean: A) in possession • keeping the ball in the back, often progressing directly from 1st to 3rd phase • bypassing central midfield, with our backline in charge of progressive passing • build up resulting in switches to wing, direct passes to CF who drops deep or direct passes to players in half-space areas • two 8s moving up the field, to provide additional avenues for progression • alternatively, one CM staying deeper, with FB inverting and moving to half-space • when the ball gets up front, relying on isolation play & third man runs to penetrate further and get into shooting positions • no intent / patterns for sustained final 3rd pressure: key is attacking unsettled defenses whenever possible The ETH attacking blueprint for United is simple: a technical backline to keep the ball and finding openings + having an extra man up front, since central midfielders are not as active in build up nor 2nd stage progressiom. B) out of possession: • since losing the ball happens often in direct attacking, a lot of counter-press by forwards and players positioned up front • narrow front pressing, not aggressively backed up by FBs, CMs or backline moving up • stopping central progression and often leaving an opposition FB open • empahsis on box defending, by dropping players deep as soon as front press is bypassed • rest defense structure problematic due to huge taps between players (as a result of our possession methods) • reliance on very mobile CMs to cover ground and make up for rest defense gaps The main OOP difference between ETH’s 1st and 2nd season was in using a mid block much more often in the former and much less in the latter. Since I don’t think it’s likely we’ll drop back to a mid block, it’s safe to assume we’ll still be seeing a high front press comined with safety of box defending. There might be an attempt to develop a more compact pressing structure with a higher line - I really hope there is - but there’s been zero evidence of that in our first two preseason games.

@tactixology All I see Ten Hag do is get defenders to progress the all from deep, midfield supports attackers in the final third. He doesn't need a CM. Is this a good thing? I don't know. And I might be wrong with this assumption, I don't know any other team in Europe that plays this way

Of course, the game has changed drastically, and a Carrick of 2024/2025 would have to be more mobile, better at pressing opposition and covering ground, more skilled at receiving and carrying under intense pressure. But for reasons completely unclear, we've been avoiding central midfielders and pivots with playmaking skill for years now. Under ETH, it fits the box-to-box vertical model, where the emphasis is on quick box attacking and deep box defending. Bypassing the midfield for him is a feature, not a bug, of our system. Hence it's quite logical for us to be looking at Ugarte as the most dynamic and mobile DM option out there: though severely lacking in progressive passing ability, he can cover ground and protect the defense. But even ETH's model would benefit greatly from having an elite playmaker at the center of the park; as United fans know best, most great teams do.

MANCHESTER UNITED'S CARRICK CURSE Ever since United had one of the best deep-playing pivots in the game - and one of the best in the history of the Premier League - in the squad, it seems like we've been hit by a curse. Or some kind of a irrational fear of deep-playing pivots. As if having Michael Carrick somehow rendered us incapable of having anyone but Michael Carrick. We've been investing heavily into our midfield for a decade, and yet, it seems like there's been a conscious effort to avoid pivots who can scan, receive, pause and play a line-breaking pass, or switch wide, or calm the tempo. And avoid them like the plague, at that. Consider United's main central midfield investments of the past 10 years: - Pogba: €105m - Casemiro: €70.6m - Mount: €64m - Fred: €59m - Matic: €44m - Donny: €39m - Herrera: €36m - Schneiderlin: €35m - Fellaini: €32.4m A total of almost € 500 million in 10 years. And you will notice that, besides Matic, not a single player on the list comes even close to Carrick's playmaking profile and progressive passing.

@TheEuropeanLad I know you have your reasons, and I also hope it is so for Manchester United. Fit squad and the new set of coaches? I think 3rd or second is doable

@MushtaqBilalPhD Team work is you citing other people's work, going through series of research papers and peer review process. So you believed in team work

@getpaidwrite We survive by multitasking these days

If you want to create. If you want to expand your mind. If you want to make something that lasts. Do less. Your mind needs space to create.