JP Raymond

5.6K posts

JP Raymond banner
JP Raymond

JP Raymond

@J_P_Raymond

Data scientist interested in ML, DL, NLP and CV. I race karts and play cards whenever I can afford it ;)

Katılım Nisan 2012
1.9K Takip Edilen215 Takipçiler
JP Raymond retweetledi
Maxim Lapierre
Maxim Lapierre@Lappy14·
Je me rappel d'un certain printemps où un gardien inattendu et un défenseur de la ligue américaine nous ont amené en finale de l'Est...😉
Français
41
11
656
53.8K
JP Raymond retweetledi
GTOWizard
GTOWizard@GTOWizard·
We benchmarked every major AI model at poker. GPT-5.4, Claude Opus 4.6, Gemini 3.1 Pro, Grok 4 and more. All played 5,000 hands of heads-up no-limit against our state-of-the-art poker agent. Every single one lost. Here's the full breakdown 🧵
GTOWizard tweet media
English
16
37
201
211.3K
JP Raymond retweetledi
RDS
RDS@RDSca·
Juraj Slafkovský devient le meneur au chapitre des points inscrits par un joueur de 21 ans et moins dans l'histoire du CH! 👀👏 #GoHabsGo
RDS tweet media
Français
4
15
278
10.5K
JP Raymond retweetledi
RDS
RDS@RDSca·
Félicitations à nos collègues Pierre Houde et Bertrand Houle pour ce bel honneur! 👏
RDS tweet media
Français
14
22
631
15.4K
JP Raymond retweetledi
Klara
Klara@klara_sjo·
In China, driverless delivery vans have become a total meme, they plow through crumbling roads, fresh concrete, motorcycles, anything. Nothing stops them.
English
2.4K
12.7K
132.2K
6.1M
JP Raymond retweetledi
doomer
doomer@uncledoomer·
german pastimes include a competition to see who can drive the local light rail the most precisely
English
193
1.3K
56.6K
2.6M
JP Raymond retweetledi
MMitchell
MMitchell@mmitchell_ai·
Missed the memo on what TL;DR means.
Sam Altman@sama

Yesterday we did a livestream. TL;DR: We have set internal goals of having an automated AI research intern by September of 2026 running on hundreds of thousands of GPUs, and a true automated AI researcher by March of 2028. We may totally fail at this goal, but given the extraordinary potential impacts we think it is in the public interest to be transparent about this. We have a safety strategy that relies on 5 layers: Value alignment, Goal alignment, Reliability, Adversarial robustness, and System safety. Chain-of-thought faithfulness is a tool we are particularly excited about, but it somewhat fragile and requires drawing a boundary and a clear abstraction. On the product side, we are trying to move towards a true platform, where people and companies building on top of our offerings will capture most of the value. Today people can build on our API and apps in ChatGPT; eventually, we want to offer an AI cloud that enables huge businesses. We have currently committed to about 30 gigawatts of compute, with a total cost of ownership over the years of about $1.4 trillion. We are comfortable with this given what we see on the horizon for model capability growth and revenue growth. We would like to do more—we would like to build an AI factory that can make 1 gigawatt per week of new capacity, at a greatly reduced cost relative to today—but that will require more confidence in future models, revenue, and technological/financial innovation. Our new structure is much simpler than our old one. We have a non-profit called OpenAI Foundation that governs a Public Benefit Corporation called OpenAI Group. The foundation initially owns 26% of the PBC, but it can increase with warrants over time if the PBC does super well. The PBC can attract the resources needed to achieve the mission. Our mission, for both our non-profit and PBC, remains the same: to ensure that artificial general intelligence benefits all of humanity. The nonprofit is initially committing $25 billion to health and curing disease, and AI resilience (all of the things that could help society have a successful transition to a post-AGI world, including technical safety but also things like economic impact, cyber security, and much more). The nonprofit now has the ability to actually deploy capital relatively quickly, unlike before. In 2026 we expect that our AI systems may be able to make small new discoveries; in 2028 we could be looking at big ones. This is a really big deal; we think that science, and the institutions that let us widely distribute the fruits of science, are the most important ways that quality of life improves over time.

English
5
1
77
18.1K
JP Raymond retweetledi
internet hall of fame
internet hall of fame@InternetH0F·
Nothing will ever beat the 1992 Olympic Torch lighting
English
942
8.6K
144.6K
7.6M
JP Raymond retweetledi
Justine Moore
Justine Moore@venturetwins·
ChatGPT when another Studio Ghibli request comes in
Justine Moore tweet media
English
545
5.4K
92.3K
4.6M
JP Raymond retweetledi
Paras Chopra
Paras Chopra@paraschopra·
How to achieve success without trying to shoot for it! My notes from the #book "Why Greatness Cannot be Planned" by @kenneth0stanley and @joelbot3000 🧵
Paras Chopra tweet media
English
49
248
2.5K
457.9K
JP Raymond retweetledi
bencb
bencb@bencb789·
I've been waiting to share my take on "Laptopgate." I wanted to consider all the facts and hear from the players and people involved to keep this objective. I'm late, but I don't care about the buzz. The goal is to make the best possible decisions to ensure a safe environment for pros and recreational players. I'll address a couple of bizarre takes. I have been addressing this issue for years, warning that it will worsen if we don't have clear rules to ban any RTA (Real-Time Assistance) or help at the tables. People will always take advantage of unclear rules and grey areas to gain the most from them. And now, here we are ... A clear ruling and enforcement during the final table would have protected the players from this clearly unfair way of coaching. I want to focus on some opinions and takes on this topic and explain why they are nonsense. ❌"This can't be entirely enforced." (I have often heard this excuse in the past years). And? Is speeding entirely enforced? No. Does it stop us from having speeding limits, rules, and hard punishments in some cases? However, the risk of punishment creates a somewhat safe environment. It's not perfect, of course. No law or rule is 100% enforceable. Should this stop us from introducing new regulations to provide a safer and fairer environment for everyone? Using RTA or any assistance is an instant DQ (disqualification) and a one-year ban. Do it a second time, and it's a lifetime ban. It will be hard to detect and require surveillance footage or multiple witnesses. But it makes it much harder for people to cheat. They have to be more careful, hide it more, and use it less obviously, which will benefit them less. It becomes less worthwhile because developing effective cheating solutions is a hassle. We, therefore, get fewer cheaters in the first place since the risk isn't worth it, and the punishments are too harsh. ❌"In baseball, the coaching staff also uses tools." Comparing a solo sport with a team sport is quite a stretch. You should compare Poker with chess. ❌"Checking spots between hands is fine." Would you be okay with asking ChatGPT during an exam? It's the same. Poker is about preparing and studying in advance, like an exam. You don't know which spots you will face. That's the challenge. You have to navigate through unknown situations, adapting and extrapolating from the spots you studied. The more you prepare, the more likely you will win. Allowing solving spots between hands lets you always know what to do pre-flop and, to some extent, post-flop. You gain such a big edge that punting it off is almost impossible. Sometimes, we are clueless, and our opponent benefits from this. Reducing the amount of time we are clueless through real-time assistance is cheating. In Poker, you win if you make fewer mistakes than your opponent (and, of course, luck plays a role, too), so by using real-time assistance, you will remove not only the frequency of making mistakes but also the magnitude of every single mistake. In Poker, we would say, "You would punt less," which gives you an unfair advantage. Imagine you play against someone who can review hands in between and prepare for the stack sizes in the following hands. Especially on final tables with the impact of the ICM (Independent Chip Model), it becomes even more unfair since you are playing against an opponent who is unlikely to make any big mistakes. ❌"They (coaches) didn't know about the announcements at the start of the tournament." Well, this doesn't protect you from punishment. Ridiculous. At least Tamajo should have been aware of this and told his staff to remove the laptop from the rail. ❌"It's a grey area." Let me get this straight. We all know that running solves and using RTA online is considered 100% cheating, and now running sims on final tables is a grey area? Especially Dom, who has been calling out other training sites and solvers for ridiculous things, has been the ultimate white knight when it comes to poker integrity (which I kind of respected him for, even though he was shitting on me/us for years), now bending the rules in his favor and responding to this matter in the most childish way shows his level of morale. ▶️What needs to happen? 1⃣ WSOP and other TOs implement a rule that is 100% clear on forbidding ANY poker assistance at the poker table and in the tournament area. (If you go to the toilet to check your solver, whatever, there is nothing we can do, but again, it makes it harder. Instead of running SIMs secretly on your phone after every hand, you can only check 1-2 spots every 1-2 hours, which is much less harmful. Again, this won't provide 100% protection, but it is a start; it will keep it fairer). 2⃣ Implement HARD punishment: Instant DQ + minimum one-year ban (I'm also okay with a lifetime casino ban). 3⃣ Allow coaches on the rail but define and limit coaching communication (only during breaks). Set up a coaching area where 1, 2, or 3 coaches can sit, but the player cannot talk to the coaches between hands. I don't want to take away from the fun and emotions of the rail. Go celebrate with your friends and family. (This might be a loophole for coaches sneaking among friends and family, so we must see how this goes). Celebrating with the rail is also something that needs to be monitored. "But it's about showing emotions, celebrating, etc." Well, not if it comes at the cost of game integrity. The rules are much stricter in esports (with much smaller prize pools). Communication with the coaches is monitored, and during tech breaks, players are not allowed to talk to each other about strategy, etc. It is observed and monitored. If you go on stage, you cannot take your phone or talk and chat with the crowd. I don't mind moving us in a similar direction if players are placed on a similar stage and cannot talk to their friends and family if we have no other choice. (Again, I love seeing players chatting and celebrating with friends and family, but not if this gives too much room to cheat). Game integrity and fair play come first. A lot of things need to be considered. But we need a start, we monitor, and potentially adapt. The earplug theory is complete nonsense, by the way.
English
33
18
200
55.3K
JP Raymond retweetledi
James Lucas
James Lucas@JamesLucasIT·
Places on Earth that don't look real - thread 🧵 1. Namibia, where the desert meets the ocean
James Lucas tweet media
English
2.2K
46.7K
389.6K
55.2M
JP Raymond retweetledi
Agreeable Greg
Agreeable Greg@AgreeableGreg·
Diet Coke: Making people feel better about ordering two Big Macs and a large fry since 1982...
English
39
267
1.3K
2.5M
JP Raymond
JP Raymond@J_P_Raymond·
@gdb Could not agree more. Unfortunately, it seems most ML teams in the corporate world are incentivized to settle for good enough.
English
0
0
0
26
JP Raymond retweetledi
Greg Brockman
Greg Brockman@gdb·
in ML engineering, seek perfection rather than good enough
English
83
98
1.2K
183.2K
JP Raymond retweetledi
Massimo
Massimo@Rainmaker1973·
The ocean sunfish is the world's heaviest known bony fish: it reaches up to 2,300 kg of weight & 4.2 meters of size across the fins [read more: buff.ly/2p3FDYk] [video: buff.ly/2PTq3fA]
English
256
2.1K
13.3K
1.9M
The New York Times
The New York Times@nytimes·
A new game called Digits is available for a limited-time beta release by our Games team. The beta test is available to everyone and is free to play. nyti.ms/3othoT2
English
9
6
37
204.8K