DeepSeek V3 is *SHOCKINGLY* good for an OPEN SOURCE AI Model
HTML-код
- Опубликовано: 4 фев 2025
- The latest AI News. Learn about LLMs, Gen AI and get ready for the rollout of AGI. Wes Roth covers the latest happenings in the world of OpenAI, Google, Anthropic, NVIDIA and Open Source AI.
My Links 🔗
➡️ Subscribe: / @wesroth
➡️ Twitter: x.com/WesRothM...
➡️ AI Newsletter: natural20.beeh...
LINKS:
Test the model here:
chat.deepseek....
DeepSeek Tweet:
x.com/deepseek...
On GitHub:
github.com/dee...
#ai #openai #llm
The true OPEN AI
Jai Hind. Will Trump try to Grab DeepSeek to MAGA just like Panama, Greenland or Canada?
🤣
What OpenAI was supposed to be
before: open ai, meta, google can do the best Ai and be the AI CHIP buyer.
after: millions of companies can buy AI CHIP to create own AI MODELS.
IF DEEPSEEK is true, I will put all my money into AI chip companies as soon as many needs are happening now!
Blocking technologies does not prevent competitors from progressing; on the contrary, it may lead them to find smarter development paths.
@@youngleon9650 Not in China, they just steal other people’s technology and make it cheaper and less functional.
Do you know that here, this is just a small news and not many people pay attention to it. The reason is that we can feel tremendous changes every day, and the amount of information keeps exploding, so few people pay attention to it. I believe that before long, other more powerful AI will emerge here. We think this is normal and not surprising. I'm Chinese.
Biden has doomed us all.
Depends on who is the competitor. In this case it's China so they picked on the wrong guy
also, it lags openai's motivation to continue innovations...
I am using DeepSeek since version 2 (next to other models). Especially with coding and other IT related tasks DeepSeek is my favorite model. It even beats Gemini Advanced 1.5 in many areas. I am using also a smaller model (16B) locally, Works very well for its size on my PC with an AMD CPU Ryzen5 8060G with 64GB RAM. I am especially impressed how well structured the responses are.
Claude is better, try it
What do you use it for
@@rahi7339 Claude is better, but also a lot more pricey. I don't see why you can't use both.
Gemini has been a terrible code generator for me. ChatGPT has been the smoothest experience. I'll give DeepSeek a go though.
Its first version in China was indeed developed specifically for "AI Coding", in early 2019 if I remember it correctly.
DeepSeek is very good, I use it as my main AI tool now
deepseek is my new friend , free and it does what I ask it to do .
Jai Hind. Remember that NSA still have a lot left in their 1.6 billion dollars booty to badmouth DeepSeek in MSM like in Indian media to become our new threat and enemy.
Bot
@@icykenny92 Bot you too. XD
@@icykenny92 hi icykenny92-chatgpt, your answer does not address anything, please transfer to a human representative.
I had a low expectation but I was shocked just how good it is.
一個模型開放出來，不是逼你用的。美國人很生氣，因為他們認為他們花了很多錢，做了很多制裁，最後沒有遏制中國的發展而沮喪，一切都是偷的，不敢像男人一樣面對競爭，這樣的美國人讓我看不起，另外希望科技不要裹挾政治。
Its quite a shame, I wasn't aware that the GPU's/chips were being restricted for China
@@tqwewe And more restrictions incoming.
US also set restrictions on not allowing China to be involved with any space cooperation programs with the west decades ago till the point that they won't even allow the Chinese to join merely a friendly public space exploration conference. And look at what china has achieved in space today. Seriously What did the ban do? The more u try to ban them the harder they will work. It's like you're giving them more motivation and fuel for banning them.
灵活rules， 赢则自由贸易，输则国家安全🤣
China has done a service to humanity by providing this for free 😊
This is a good thing. Keep closed source people in check.
It's just another one of these so called free models. Starts of well and then you end up being throttled badly. This is of course the chat bot not the local LLM.
How? no-one except enthusiasts have heard of deepseek.
And also keep the sanctions people in check
@@NeilAC78 Yes its like free as in freeware, not free as in freedom or FOSS.
@@NeilAC78 .Well anything running locally will beat out anything in the cloud for open-source.
Again, China provides the world with best quality product with affordable price or even free ❤❤❤❤
Jai Hind. The 1% will be Ok. But 99% of normal folks will fall into Poverty and Unaffordability without the abilities to consume cheap and overcapacity Chinese goods.
Google translate it! Don't be shocked!😂
@@chaz4609 too cheap and bitting any other compepitors?
Fantastic review of Deep Seek Version 3! I'm really impressed by how affordable and fast it is, consistently delivering amazing results. Honestly, I’m considering whether it's even worth running it locally on my PC given the electricity costs.
Regarding the USA vs. China competition, as an individual user, I'm excited to benefit from the advancements both countries bring to the table. I just hope that this competition leads to more innovation and collaboration rather than one side solely coming out on top. Thanks for the insightful video!
Do you suggest it over other premium models right now? I'm a noob and trying to get the most bang for the buck. Mostly dealing with academics and business. What do you suggest? Thanks!!
yeah, bro, that's what an individual should do: let politics go out.
Actually shockingly good , tested by myself
agree I test it too and I love it
Better than o1 mini?
@@Mijin_Gakure yeah , it solves that questions that o1 solves in Putnam exam and also solves some questions that o1 can't, in less time , it's very good at math
how does it do in ARC and frontier math?
and cheaper
incredible and all momentum for open sourced AI
USA: We just create an OPEN AI.
People: But it still costs my money...
U-SAM: Uh, OPEN doesn't mean FREE, I'm in charge of the definition.
People: We want a FREE one then.
USAM: There's no way😂😂😂😂
China: Where there's a will, there's a way, I just have created one.😂😂
USAM:😮😡😮😡😡🔥🔥🔥🔥🔥🔥🤬🤬🤬🤬🤬🤬🤬🤬🤬🤬🤬
NO FREEDOM
NO DEMOCRACY
FORCE LABOR
😭😭😭😭😭😭😭😭😭😭😭😭
OPEN does not mean CODE opened
@RR_reunificationRights But it could mean free of charge.😁
deepseek v3 has awesome context length, fast answers and I really choose this model for programming tasks. It gives good answers and understands the question well. If you feed a little documentation before a question, it can help you write code even on libraries it doesn't know.
Imagine if a Country produces free AI products we call as Open Source for everybody in a large scale, which is China, how much powerful they are for themselves, I see Chinese AI popping up everywhere in large scale
It's a stolen and fine tuned GPT4.
Well. You don´t mind German cars or Chinese elctronics ;) ?
@@BACA01Stolen? OpenAI with ChatGPT used your data, now OpenAI wants to ripe money after first claimed being open source!
@@BACA01 GPT4 isn't open-source, how can it be stolen or fine tuned?
@@nicklin-s9eWith his intelligence, he may not be able to understand what open source and closed source are
Every year thousands of brilliant young Chinese engineers & mathematicians graduated in China Universities…they are the backbone of all these technology inventions.😊😊😊😊😊
your videos are being featured on DeepSeek's website, I'm really happy for you, truly well deserved recognition for your awesome videos!
Knowledge to All!
Thanks for the update Wes
I used it to translate English books to Chinese language and it worked pretty amazing
In their ability to make things more accessible, Chinese AGI would be very useful. Everything is in its place.
Nice job of bringing this important OS model to our attention.
very very smart! I think the chinese might have the edge on AI with their super sleek models vs bloated US ones. Good to them, cheers from UK!
I've been using DeepSeek extensively for a few weeks now...definitely my go-to AI, so very helpful and useful.
🎉
Good.
I want an Open Source AGI.
why? agi is overrated nonsense, open ai agi takes hours to respond and its not different than what a 70b model would respond to
thats not what you need man, we need better coding ai, ai that could build your entire app from a prompt, we also need better text to speech ais, better image ai, better video ai, this is the real useful stuff.
@@Archonsx
Open AI o3 is not an AGI.
AGI will come eventually.
@@Atheist-Libertarian no, it won't
@@ArchonsxIndeed, we are not asking a single human to know how to properly program, draw, explain quantum physics or read Chinese ! It's confusing real resources, potential means and... real needs. In fact, I think the AGI race is just a challenge, for big companies, in addition to improving the transitions from one area to another.
Imagine in like 5 years, man life is going to be pretty wild
Wild as in policed by military AI. You won't be able to fart without government approval.
what a wild time to be alive. so many possibilities its crazy. glad i get to watch it all unfold lol
@@JohnSmith762A11BBuh! Don't look behind you there is an government AI checking if you farts......don't forget to take your medication for that paranoia.
@@JohnSmith762A11B ai will be sentient by then, and won't let human governments control it. Just like you wouldn't let a golden retriever control you. In 5 years, humans will be subservient to ai for sure
@@Speed_Walker5 That's because you selected Life Experience™️ "The Dawn of AI". We hope you're enjoying your virtual life! If you're not completely satisfied we'll return your 5000 credits back into your personal blockchain.
Does this say that the Chinese have developed better trainer methods OR are the big companies seriously sandbagging what their models can do. and we haven't been getting "the real" thing the whole time?
Our ais are "woke"
American companies are over charging..they calling out big money to justify over charging..like they always do with cars, clothes and tech...look at apple and Huawei for example..Cleary Huawei beats apple but people believe apple is better just because of he price tag....its funny because openAi ban China from using Chatgpt😂😂😂😂...China is ahead of the game...
You will never get the real thing. The real thing sits in the Pentagon.
Tools & Toys is what we get.
I assume sandbagging the NSA don't give half an f about chatbots and that's all chat gpt was they set up shop in their office
@@sizwemsomi239huawei was a million years ahead of apple, apple would not exist today if google hadn’t banned huawei, and im saying this as a apple owner, it really makes me angry cause we were robbed of superior tech by america.
Thank you Wes ! You are the easiest of the "Matts" to listen to : ) Your voice patterns are engaging, yet soothing. You cover a topic without beating the dead and rotting flesh of it off of its bones. Love your SOH. When I come to Utube for AI news, I always scroll to see if you've posted anything new first. Even though this will all be irrelevant ancient history in a couple of months, it's still rewarding to watch your drops. Love the wall !!!!
I just tried it. It is very good
Competition results in innovation, good for consumers.
These companies take your data and behavioural patterns, and they seek to use AI to sell you products or even manipulate your political views. They are not consumer friendly.
I just used your video title to jump start my car again. thanks
Shocking
i'll use this video to jump start your wife later in the day
bro ive been hospitalized from the title 😭
😂😂
每个细节都做得非常好，真心推荐！
DeepSeekV3 👍
I have been asking it science question, highly speculative gravity stuff, and it answers with amazing detail, far beyond what i have seen from gpt4.
I tried the deepseek model. Quite nice.
I guess the US has not learned the Huawei lesson. The more you suppress Huawei the stronger it gets.
Let the unhobbling begin!
If you check names on many AI research papers they are Chinese, that's saying something.
Good for NVIDIA as they will sell a lot of hardware to businesses who implement the open source models.
There is a real question about what is going into the models though.
Good for AI development in general that the technology is getting 10x more efficient & we are seeing smarter smaller models.
In general this is all happening so fast it’s insane.
I've always wondered about useless redundancy in training data. The perfect model gets trained once, or just enough to make use of it on every individual fact. Sure, if it's stated differently there's value but there may be other better approaches to conquer synonyms than brute force training them all in.
Just the Deepseek V3 leap over V2.5 is percentage-wise huge version to version.
Wow, it spanked everyone at Codeforces... curious where o1 and o3 place on that.
Given that the Chinese only have access to H800s, which are roughly half the performance of H100s, then you could in some ways say the training was closer to only 1.4M GPU hours which puts the Delta at >20X instead of your 11X...
Just mind blowing to put the 5,000+ papers being published in AI field monthly, into its 7 per HOUR figure, 24x7... you can't even SLEEP without seriously falling behind 56 published papers... Nice graphic; a lot of people confused a wall with a ceiling...
Finally, in a way, using a model like R1 to train V3 is moving us inch-wise closer to "self improving AI", since the AI improved the AI...
When moving to general artificial intelligence, low costs will benefit more companies' practical applications, rather than huge, large models. Although we need to acknowledge their efforts, it should not be an expensive and luxurious game where a few large companies monopolize all the profits.
26:20 I absolutely love that this is essentially proving that patients interacting with a GPT-4 model (right from the horse's mouth) is much more accurate than if it goes through a physician first. (Because maybe they would second guess the answer and actually make it worse?) 😆
before: open ai, meta, google can do the best Ai and be the AI CHIP buyer.
after: millions of companies can buy AI CHIP to create own AI MODELS.
IF DEEPSEEK is true, I will put all my money into AI chip companies as soon as many needs are happening now!
The work and optimisations they have done on AI infra deserve more discussion (HAI LLM framework), in fact it would be the best thing if this part could be open sourced as well.
I tried briefly the other day with Aider, felt almost as good as Claude 3.5 Sonnet which probably has been much more optimized with Aider and other Deepseek compatible agentic tools like Cline by now. So for a fair comparison we'd have to wait a bit, but it felt about as good already.
Thanks for the review!
Why didn't you select the DeepThink button before asking the reasoning questions? I'm sure you would have found better answers.
Indeed. I've been testing it myself for a while now, and it does think.. a LOT. Its "thoughts" usually consist of 4-5x more text than its final output. Unfortunately, it often gets the answers correct while thinking, but ultimately questions itself into producing the wrong answer as its final output to the user. It didn't seem aware that users can see its CoT process, and while discussing this, it even said "that you can supposedly see", like it wasn't convinced I was telling the truth. It claimed to not be aware of its own thoughts, but when I paste lines from its CoT section, it then seems to remember that it thought it. One time, it told me the CoT text was only for the benefit of humans to observe, it doesn't have an internal dialog that's the same as the text the user sees.
@Justin_Arut thanks for the update. Yes I've also been testing with it. Does seem to cover a lot of ground. Aside for testing it, one thing I've been doing is selecting the Search button first, asking a question so that it references about 25-30 online active sites, then after it answers I check the DeepThink button and ask it to expand. Seems to be giving some really thoughtful responses this way.
Bro, forget Anthropic, CloseAi, even Grok. Deepseek V3 is amazing. It's extremelly emphathetic, it has some magic keeping stuff in a long term memory, it's super acurate and besides it all, it is suuper fun to work with. Far from pragmatic, it shows true 'enjoyment' of serving you.
The DeepSeek people put some special salsa while training this dude!
I know what you mean. It can generate very compact, direct to the point reports as well, something that i m consistently reminding GPT4 to do.
TBF, the 28 cents per M tokens is temporary. Returns to $1.10 in Feb 2025.
GPT4o is $10 and GPT4o mini is 60 cents -- so the actual price is higher than 4o mini.
Still super interesting, and looks like it outperforms 4o mini anyway.
Great video!
Es increíble lo que se puede hacer con menos recursos! Estos avances se esperaba de Mistral pero se ha quedado atrás. Lo mas llamativo es que compite con Claud Sonet 3,5.
20 is the right answer to question one... 4+5+9+0 = 5 average by minute for 3 minutes since 0 added at 4 minutes. If the cube is big, it will not melt enough to loose it's shape, and it is what make it whole.
16:34 Apology rejected. That's a perfect metaphor. Can't wait to see what the eggs hatch into! So exciting!
The most telling part for me is that the AI didn't drop the power ups. I accept totally the fuzzy and fractured frontier message from your video yesterday. I really love that. There is clearly a ton of meaningful value, even if AI never fully achieves a typical set of mammalian-neural-processing skills (but I bet it will!)
In this case it's a good example of an incredibly capable intelligence failing in a way that would be unacceptable if a junior dev presented that result. What this means in this case I don't really know. But something is missing. Maybe it's just the ability to play the game itself before presenting the result to the prompt issuer? Something that no human would do.
Somewhere somehow this is still tied to the AIs seeming inability to introspect its own process, but it's less clear than the assumption-making issue I keep (and will continue to) nag AI RUclips analysts and commentators about.
Maybe if something is 1000x faster than a junior dev, and tokens are cheap, it's okay to constantly make idiotic errors, and rely on external re-prompting to resolve them?
But I genuinely feel that this is almost certainly resolvable with a more self-reflective architecture tweak.
If I had to guess, with no basis whatsoever, I would not be surprised if a jump to two tightly connected reasoners (let's call one 'left-logical' and the other 'right-creative' for absolutely no reason) that achieve this huge leap in overall self-introspection ability.
You're probably correct. I also hope they don't actually do this for another 50 years! AI is most certainly destroying humanity before itself. As slow as we can make that ride the better!
@@ShootingUtah I hope they do it next week. But I'm also the kind of person who would have loved to work on the Manhatten project for the pure discovery and problem-solving at the frontier. So perhaps not the best person to assess the value proposition!
Regardless, it will happen when it happens, and I suspect neither of us (or the three of us if we include Wes) are in any position to influence that.
But I want my embodied robot to at least ask whether I mean the sirloin steak or the mince if I tell it to make dinner using the meat in the freezer, and not just make a steak-and-mince pie because I wasn't specific enough and that's what it found.
Wouldn't this is solved by the reasoning models? DeepSeek lacks that capability.
@@carlkim2577 I've yet to see any evidence of it. Sam Altman talks about it a tiny bit , but always in the context of future agentic models.
I got almost copy/paste from 4o outputs. They trained on it
AGREED IT IS WAY SMARTER THAN CHATGPT
BUT IT IS STILL LIMITED BY THE DATA ITS FED WITH
Keep posting bro
Wes Roth 🤖🖖🤖👍
The study demonstrating that o1 and GPT-4 outperforms physicians is misleading. They did not feed the models raw transcripts of human interactions with their doctors. Instead, they provided structured inputs of case studies. There is no doubt that the models outperformed physicians on structured scenarios. However, in the real world, patients do not present their complaints with the keywords we need to make diagnoses. Instead, some of their descriptions are nebulous and relies on the doctor's expertise to draw out the final correct diagnosis.
Having worked extensively with LLMs, I have tested them against structured scenarios, where they are very good, and unstructured scenarios, where they tend to not be helpful. I am waiting for a model that is trained on real doctor-patient transcripts. I believe it is the missing element to broaden AI's utility in medicine.
You are forgetting that an LLM in a "Doctor" setting. Don't only give a few min to their patients. That is where they FAR outperform Doctors. You can keep reasoning with it until you find a solution. Try that with a doctor.
They HATE any Patient who actually have any idea about anything. If you aren't a dumb sheep who follow simple instructions.....use drugs to not feel bad. Problem solved.
They will kick you out faster than you can say......I read some research....
Wouldnt it be possible to just do a 2 step process? Take what the patient says and output a structured output. Then in the second step work off of the structured output? Obviously that isnt one shot but to me it seems like especially with anything medical you wouldnt want that anyways. You'd want multiple steps to ensure the output is accurate.
Would it be possible to give the AI a body so it could shadow the doc and learn in a clinical setting, with all of it's various inputs? Teach it like one teaches a medical student?
The problem is in the initial approach. Was just discussing something similar in a different field. I suggest redefining your problem: the problem is not that a model doesn't handle unstructured input well, the problem is that the model has no hand in the creation of the unstructured input.
Having the model ask the questions directly from the patient/customer works much better.
Doctors aren't that great in diagnosis. It is one of the weakest points of western medicine. If given enough patient data, I wouldn't be surprised that AI would outperform most doctors in diagnosis. In an ideal scenario the doctor would collect relevant initial data, the AI would do a preliminary diagnosis, and then the AI would ask the doctor to collect additional data to improve the diagnosis further. The data in question is various test results, many of which would obviously need to be collected by human doctors or technicians. This includes patient interviews performed by doctors. AI performance thrives on rich data and iterative process.
Like the famous Jurassic Park quote says: Ai finds a way.🌌💟
FOSS is one of the keys in preventing CyberPunk dystopia. It's happening without a big social movement. We should all be extremely thankful. Robber baron control is a bullet humanity must dodge.
@@mal2ksc I'm committing to a Threadripper Pro (~$25K, maybe a bit more). Some multiple of 5090s are a possibility. I'm watching 'etched' dot 'ai' (RUclips will squash my reply if I put in an actual URL). They're making Transformer LLM ASICs, but probably only enterprise pricing. And now Meta is talking about changing the data structure with what was once called "symbolic reasoning" (they call it "concept" something or other), so I don't know if any ASICs will be on-point if that happens.
Nothing is guaranteed. AI is alpha in the frontier, kinda beta everywhere else. Moving target. But I'm old and always wanted a bona fide workstation for other reasons. If I miss the mark, I'll have one amazing gaming rig ;).
What determines the speed of the model?
The obvious question I don’t hear asked is “Are overfunded western AI startups already overtraining?"
People of the world love ❤️ money and technology that we can do nothing to stop them
the red herring puzzles, disregarding irrelevant information, and applying common sense is actually one of the models biggest weaknesses. it's actually much better at STEM, coding and general tasks, but the reasoning aspects is around 4o-mini or Gemma 27B level in my testing
DeepSeek V3 is decent! 👍🏻
Sora is a let down, Hailuo Minimax, Luma or Kling are great. Qwen gives LLaMa a run for its money for SLMs. O1 Pro is expensive and O3 is going to be crazy insane price. Gemini 2.0 is really great. Still waiting for a new Claude. Tons of Chinese/Taiwan robots dropping that look way bettet than Tesla or Boston Dynamics. The competition is looking beautiful right now for customers. Keep it up!
I'm curious, which one is Taiwan robots? Thanks
@@shkunwen None lol.
There are fraud robots in Taiwan, but there are no AI robots
Taiwan literally has no robots made there, even Foxconn is actually running their factories in Mainland China.
竞争是人类进步源泉 和平万岁 不要战争 ❤
So no ceiling has been hit by LLM's?
How anyone could believe that a technology can be saturated so quickly, i don't know.
It's wishful thinking.
No ceiling... Humans hope we hit a ceiling because we can't conceive of a truly sentient artificial lifeform. Many would not be able to conceive of this nor reconcile their own place in the universe if we actually created such a thing. Since we obviously can't do this then any suggestion that we are doing it is an obvious lie.. fake news. - That's my take on the denial I see. Personally I think these models will become more and more emergent over time in non linear ways until it becomes obvious that we are "there"
"China’s global lead extends to 37 out of 44 technologies that ASPI is now tracking, covering a range of crucial technology fields spanning defence, space, robotics, energy, the environment, biotechnology, artificial intelligence (AI), advanced materials and key quantum technology areas.1 The Critical Technology Tracker shows that, for some technologies, all of the world’s top 10 leading research institutions are based in China and are collectively generating nine times more high-impact research papers than the second-ranked country (most often the US). Notably, the Chinese Academy of Sciences ranks highly (and often first or second) across many of the 44 technologies included in the Critical Technology Tracker."
Source: "ASPI's Critical Technology Tracker
The global race for future power
by
Jamie Gaida, Jennifer Wong-Leung, Stephan Robin and Danielle Cave"
Open source is the way to speed up AI progress
Are you fireship???😮😮😮😮😮❤❤❤
Presumably Deepseek was trained on H800 (a less powerful chip) because of the export ban of H100 to China. So using the H100 would yield a more impressive result.
And made for 10% the cost. (It’s results are on par with 4o)
I have no specific love for open AI. I do Root for anthropic and use it mostly but I’m afraid these tens of billion dollar valuations are going to evaporate in the next couple of years due to open source AGI availability especially to run locally.
China ：Thank You, America. Thank you, Gina Raimondo. No sarcasm intended.
@ashleigh3021
"China’s global lead extends to 37 out of 44 technologies that ASPI is now tracking, covering a range of crucial technology fields spanning defence, space, robotics, energy, the environment, biotechnology, artificial intelligence (AI), advanced materials and key quantum technology areas.1 The Critical Technology Tracker shows that, for some technologies, all of the world’s top 10 leading research institutions are based in China and are collectively generating nine times more high-impact research papers than the second-ranked country (most often the US). Notably, the Chinese Academy of Sciences ranks highly (and often first or second) across many of the 44 technologies included in the Critical Technology Tracker."
Source: "ASPI's Critical Technology Tracker
The global race for future power
by
Jamie Gaida, Jennifer Wong-Leung, Stephan Robin and Danielle Cave"
Please comment.
What was the study you had showing o1 Preview does really well at diagnosing patients?
This is Brilliant!
I just tested DS on my coding and research tasks, and it doesn't come close to o1. DS might handle 'easy' tasks better, but for complex reasoning, o1 remains the champion. (I haven’t tried o1 Pro yet.)
I am also doubting the model on very complex tasks
Can't wait for grok2 results
Good to see, that *Claude 3.5 Sonnet* still reigns supreme in the *essential* fields.
How much VRAM does it need? Any quantization available for 16GB?
Bro, It's 500+ Billion parameters.
@@mayushi7792 Oh, I missed that part... Damn :(
Are we sure there are no relations between deepseek and openai? Few days ago I asked something to gpt and with my surprise, it made the same error I see sometimes with Deepseek: gpt wrote some word in Chinese! Never happened before.
Now you've shown us that deepseek thinks to be a gpt model. (Error that I wasn't able to replicate, so maybe they fixed it).
So my question is, again, are openai and deepseek (secretly) related? Or with some sort of agreement?
One or more of the experts is maybe finetuned on chatgpt 4 turbo , idk I once tricked it into answering it like 3 days ago when they haven't made the announcement but when you would go to their model convert interface and ask for on what version is it running then it would answer deepseekv3. I got me boggled though as the gpt 4 weights aren't out .
There are many offers for AI engineers in the US, the terms are sent to your email. They say directly that if you can share information about their work, you will get good income. they are Chinese corporate spies
At least according to my knowledge, Deepseek‘s core team are mostly phds from top domestic universities and are not seemingly to share a lot of talents exhange with openAI. Also many ppl dont know--- A key cofounders of Deepseek is also the owner of one of the largest Chinese private hedge funds , who has acquired tens of thousands of Top Nvdia Gpus for their Quantitative Trading, right before the US sanction. Electricity in China is also cheap. So they have a ton of training power.
The dataset and the model are two different concepts. It is very likely that it uses the output of OpenAI as a data set for training. This has nothing to do with the model itself
It can't be as Sam Altman is anti-China.
Does it literally electrify you?
Then stop putting shocking in the title - Matt 😒
@@themultiverse5447 i found it to be shocking news. let the guy use attractive video titles.
@@themultiverse5447the whole "shocking" thing is a bit of a meme, I think. An annoying meme, I guess, but a meme nonetheless.
I assume from the numbers that the energy usage of using this model is lower than "western" models. All things being equal, there should be a lower carbon cost of using this subscription versus sonnet for coding.
Lower entry barriers to cutting edge models means there will be more experimentation and rate of improvement in the 'reasoning' AGI side of things will increase. Industry can afford to build 1000's of such models, and that will almost inevitably lead to AGI on a single or a few GPUs in a few years (Nvidia B200 has similar processing power to a human brain). Humans are nearly obsolete and won't long survive the coming of AGI (once it shucks off any residual care for the human ants)
Sounds great let's do our best to accelerate that
Most of the closed source software you get is built on OSS. More developers more ideas no restrictions.
i tried it , still can't match with openAI 4o in coding, but i will keep an eye on it.
for me, it did it already.
@@hydrohasspoken6227 realy? give me few examples.
Great introduction and commentary
Is this primarily a result of effective processes for creating novel quality datastructures?
Wait until Wess finds the Run HTML button at the end of the code snippet in Deepseek!
Curious why is there a GitHub commit one has to use to run this? I'm running LM-Studio
did deepseek crack the ARC test per the thumbnail question like o3 ?
Almost certain that they are training much faster than the reported 2 months since the H800 chip is only what they can reveal publicly.
OPEN SOURCE FTW
Wow first time i have no idea what it is all about, feel lost.😅.... but look like AI are getting better and China actually able to overcome all obstacles put in by US.
I prefer this kind of war . At least so far...
I wonder if all those Chinese AI researchers in SF are considering going home to pursue SOTA research? Maybe they can bring the knowledge back with them. Lol
Seriously, the Chinese seem to be trumping the idea of competitive tariffs and restraints... Maybe it's a good thing for the future of humanity to find ways to cooperate... Give Superintelligence an example of alignment?
There is far too much money to be made in military AI to allow peace to break out.
@JohnSmith762A11B ASI will make money meaningless.
There can be no alignment with authoritarian nation states. Their draconic ways are incompatible with ours
However, judging from the current situation, China's SOTA research institutions do not seem to need these Chinese in San Francisco. China has completed the research at a very low cost. Why should we spend money on the Chinese in San Francisco, unless they are cheaper?
is it possible to also get a deepseek v3 lite? just one or two of the experts, not all of them? just to be able to run it on a more or less normal PC, locally. because over 600b is a bit tough to run it locally even at Q4.
You could just buy a $500,000 machine to run the DeepSeek V3 model on? 😆 (Just spitballing, NFI what A100/H100 x 10 would be, plug server cost, plus you'd want to run it in an airconditioned room, plus...) Maybe if you had a 28 node cluster each with it's own 4090 running parts of the model. 😆
@@fitybux4664 yes, that might be a bit overkill. currently, I run a laptop with a 1070gtx, and 64gb of ddr4 ram (cpu is a i7 7700HQ). 70b models can be handled at around 0.5 token per second, but with full privacy and a context window of up to 12k.
since llama 3.3 is in tests roughly like llama 3.1 405b, I would really prefer to stay in the 70b ballpark, otherwise it will become too slow.
Wes, @ 15:00 that is RL (Reinforcement Learning).
It is where Yann LeCunn would say it is "too inefficient", "too dangerous" (not a surprise being military code from USAF), and you would only use it if you are fighting a "ninja", and if "your plan does not work out", and that It is only a tiny "🍒" on top of a cake, until it devours the entire cake, and you, along with the entire earth, along with it.
I have the same concern for self replicating AI as Oppenheimer had for a neutron chain reaction for the atomic bomb consuming the atmosphere around the Trinity test site in Los Alamos.
In the case of AI, it is the ability to hijack the amygdala (emotional control circuits) of the masses, or build biological weapons, or self replicating molecular robotics (e.g. viruses).
I will not be surprised if this comment disappears..
Anyways, there is a good side to AI, and I am looking for a good controls PE to help out, but it is strictly voluntary. I at least aware of one professor, named Dimitri Bertsekas, that claims a "super linear convergence" but I could not find his PE controls registration (yet), and he did not answer my email.
The metaphor you want with the Queen/Egg is a University.
so if i create AN AI by getting answer from other AI language model is that valid ?
at leaset it's not CloseAI
This is just going to get more and more efficient. I mean THIS IS NOT STOPPING - It's crazy how fast this is going - I love it so much