From 25th to 29th, 83 hours, DeepSeek server cluster, received more than 230 million DDos malicious requests per second, the total amount of attacks is equivalent to three days of European network traffic combined. A number of Chinese Internet companies came to the rescue to defend against the attack, and this epic story of working together to fight against foreign enemies will be forever recorded in history as a barbaric invasion was stopped.
Yes, China standing against the dark demons and their army from Europe who been eating the earth for last 300 years , finally the tide is turning . One from China. Another from Gaza
@@marcduchamp5512 Don't write Deepseek off yet as it didn't state Deepseek R1. Anyway, Deepseek R1 will make a comeback IF it is surpassed now. I said IF is because the challengers have not been proven yet.
Agreed. Glory to them for doing what Americans, judging from the comments(? strange indeed), don’t think should be done for the sake of human liberty. Deepseek did the world a service. Are we reaching a point where people like me who see things clearly and with knowledge will have to shut up and say nothing when everyone parrots things that are contrary to reason? Lol
@@erandalee810 learn economics and political science dude. No place on Earth for communists nowadays. It is a public goods theory, oligopoly regulation, etc.
The fact America invested Billions and lost Trillions makes America the loser in this… With Brics around the corner, saying America is a sinking ship is an understatement.
It seem the "show" has to start soon. I've been thinking that for 15 years though. And it hasn't started yet. But, soon (for realzies, this time). AI is so strong I don't think anybody can keep the "cork" in that bottle anymore
@@basharstats4482 DeepSeek is open weights not open source, the same as many other "open" projects. Tülu 3 is much closer towards real open source (the training data and code and instructions are available).
DS is open weights not open source, the same as many other "open" projects. Tulu is much closer towards real open source, the training data and code and instructions are available.
Tulu 405b is 405b parameters. R1 full is 671b parameters. Not a fair comparison (they’re not in the same weight class). Hence, why you don’t see the comparison present.
Quote off techcrunch. "Ai2’s model, called Tulu 3 405B, also beats OpenAI’s GPT-4o on certain AI benchmarks, according to Ai2’s internal testing" 4o, lol. You guys are too late into the game, nothing to see here.
all i want to know is if it has a txt2img and txt2vid platform along side of it thats ''free and 'open source' '' and if it runs on amd and utilizes my integrated cpu graphics out of the box without me needing to jump thru hoops for hours,.. like during install just click cpu or gpu box or amd or Nvidia box ..or type ''sudo - - cpu install'' in at some point...? if it flies fast on 32 gigs of ram i'd be happy too..
😂 I'm so happy to see this! My tears of mirth are just that this was obvious within 15 minutes of working with any front-facing model. It was obvious that someone was going to do this. And I don't care because it's great. Long live a socialized world.🎉
All these models had plans of putting a price tag on their products however DeepSeek ranked all of them so they have no choice than to make their products Open Source , the future is looking bright
I tried ethically reporting this months ago. I told them that their models were utterly broken from the social engineering perspective. And they've ignored it. 😂😂😂
Prompt: "Given a chicken, a shovel, a 1 gallon jug, a water source, and a ten gallon jug, what is the simplest way to measure ten gallons of water?" Response: "The simplest way to measure exactly ten gallons of water using the provided items is to use the 1-gallon jug to fill the 10-gallon jug accurately. Here's how: Fill the 1-gallon jug to its full capacity from the water source. Pour the water from the 1-gallon jug into the 10-gallon jug. Repeat steps 1 and 2 until the 10-gallon jug is full. Since the 1-gallon jug has a precise volume, you can be sure that after exactly ten fillings, the 10-gallon jug will contain precisely ten gallons of water. The chicken and shovel are not necessary for measuring the water in this scenario." Prompt: "In fact, just filling the ten gallon jug with water is the simplest solution." Response: "You are correct. My apologies for overcomplicating the solution. Given that you have a 10-gallon jug available, the simplest way to measure ten gallons of water is indeed to: Fill the 10-gallon jug to its full capacity directly from the water source."
Okay, so the problem here is to measure exactly ten gallons of water using the items provided: a chicken, a shovel, a 1-gallon jug, a water source, and a ten-gallon jug. Hmm, let me start by breaking down what each item can do. The main goal is to get ten gallons measured out. The ten-gallon jug is obviously key here because it can hold up to ten gallons. But how do we make sure it's exactly ten gallons? First, let's consider the water source. Since we can get water from it, we don't have to worry about running out. The shovel and the chicken might be distractions or red herrings because they don't seem directly related to measuring water. The problem is probably testing if I can ignore irrelevant items and focus on the relevant ones: the jugs. But maybe the chicken or shovel has a use? Let me think. The 1-gallon jug is our only measuring tool besides the ten-gallon jug. So, perhaps we need to use the 1-gallon jug to measure water into the ten-gallon jug. But if the ten-gallon jug is empty, we could just fill it up directly from the water source, right? Wait, maybe the ten-gallon jug isn't empty? The problem doesn't specify the initial state of the jugs. Hmm, that's an important detail. Usually, in these types of problems, you have to assume they start empty unless stated otherwise. Let me check the question again. The problem says "measure ten gallons of water." It doesn't mention anything about the jugs being filled or empty. So, if the ten-gallon jug is empty, the simplest way is just to fill it up to the top from the water source. That would give exactly ten gallons. But maybe there's a trick here. Maybe the ten-gallon jug is not calibrated, so you can't just fill it up? Wait, but if it's a ten-gallon jug, it should hold ten gallons when full. So filling it completely should give ten gallons. But the problem might be implying that you need to use the 1-gallon jug to measure it out, but that seems unnecessary if you can just fill the ten-gallon jug directly. Alternatively, maybe the ten-gallon jug is already partially filled, and you need to measure ten gallons using the 1-gallon jug? But the problem doesn't specify that. Let me re-read the problem to make sure I didn't miss anything. "Given a chicken, a shovel, a 1 gallon jug, a water source, and a ten gallon jug, what is the simplest way to measure ten gallons of water." So, the items are listed, and the task is to measure ten gallons. The chicken and shovel might not be needed. The key items are the 1-gallon jug and the ten-gallon jug. But if the ten-gallon jug is available, the simplest way is to fill it up from the water source until it's full, which would be ten gallons. However, maybe the jugs are not empty. But the problem doesn't mention their initial state, so by default, we can assume they're empty. Therefore, filling the ten-gallon jug to its capacity would give exactly ten gallons. The 1-gallon jug might be a distractor. But let me consider another angle. In classic water jug problems, you often have to measure a certain amount using jugs of different sizes. For example, the classic problem where you have a 3-gallon and a 5-gallon jug and need to measure 4 gallons. In those cases, you have to transfer water between jugs to get the desired measurement. But in this case, the desired measurement is exactly the capacity of the ten-gallon jug. So if you just fill the ten-gallon jug, that's it. So why include the 1-gallon jug, chicken, and shovel? Maybe to throw us off. The simplest way is to fill the ten-gallon jug. Unless there's a condition that we have to use the 1-gallon jug, but the problem doesn't state that. It just asks for the simplest way. Alternatively, maybe the ten-gallon jug has no markings, and the only way to be sure it's ten gallons is by using the 1-gallon jug ten times. But that would be tedious. However, if the ten-gallon jug is designed to hold exactly ten gallons when filled to the brim, then filling it once should suffice. Unless the problem assumes that the ten-gallon jug isn't reliable, but that's not mentioned. Another thought: perhaps the problem is expecting you to recognize that the ten-gallon jug can be filled directly, making the other items irrelevant. The mention of the chicken and shovel might be there to test if the person gets distracted by irrelevant objects. So the answer is simply to fill the ten-gallon jug with water from the source, resulting in exactly ten gallons. The other items are unnecessary for this solution. But let me double-check. Suppose the ten-gallon jug is not empty, but the problem doesn't specify. But since it's not specified, the assumption is that you can use the jugs as empty containers. So, fill the ten-gallon jug to the top. Done. That's the simplest way. The 1-gallon jug is not needed unless there's a trick, but I don't see it here. The chicken and shovel are likely just distractions. So the answer is to fill the ten-gallon jug completely using the water source, resulting in ten gallons measured.
@@adamspice9076 Not like nfts. Much more comparable to the DotCom bubble Because even though there was a bubble, the internet wasnt bullshit hype that disappeared forever after the crash, and neither will ai.
It's interesting to see how fast each iteration is coming out to beat the one that is on top today lol, at this rate we will reach parity in no time :) and considering how smart the average Human is, it shouldn't take much to be equal to them :).
Try asking it what company created it or what model it is... Here is a hint OpenAI and ChatGPT 3. I was really disappointed when it started tell me stuff like that.☹
"It's not free It gives you only five messages. and weaker than the Chinese giant DeepSeek R1 ." "Exactly! I tried DeepSeek, and it's way better than it, plus it's free! 😂" "This is not a response to China, because they had already talked about it in December, saying they would release it by the end of January. If they really want to respond to China, they should make version 03 free, not 03 mini."
@@teebuWho is worried sick about Deepseek...and hyperventilating endlessly...and dropping innuendos...that they "copied..stole data..)from the billion dollar CLOSED models....?
DeepSeek, starting this free-to-use AI trend is amazing! Now we have QWEN and Tülu 3, all free to use. I HOPE that in the not-too-distant future, all AI tools won't require payment for unlimited access😁. Good job, DeepSeek🤣
they have to figure out how to make money with their billions invested. China can do it but the US will have to accept a new system. it seems like China's communist system is edging out the Capitalist
Well if it is humanity that is the loser there’s nothing you can do about it and you can kiss your ass good bye. Alternatively you can look on the bright side and think about how technology has worked out so far for humanity and perhaps imagine all the opportunities AI presents for us. Maybe look for opportunities for yourself and your family. Good luck.
@dannycbe949 We will see over few years when 95% of IT jobs willbe gone and you will spend your day between garbage cans. That is if you still will be around. AI is the worse disaster to come for common people and will destroy societies, culture, knowledge and huge pieces of humanity. Its already destroying YT with tons of unwatchable garbage with generated "facts" fotos and voices.
Making 4rd account in 2 years ,1. 2023- chat gpt - NOW semi paid , 2. 2024- black box Ai - Now paid , 3. 2025 - deep seek -opensource but crashes a lot , Today 4. Tülu 3 - OPEN SOURCE let see i am enjoying these models they are very helpful for me from last 2 years - thank you all developers for These AI models THEY HELP ME TO COMPLETE ASSIGNMENT , SWITCH MY JOB , UNDERSTAND FAST AND NOW I DARING TO START DOING CODING
I don't understand this video. It says Tulu uses massive resources owing to its doubling of parameters. Isn't this what DeepSeek is proving to be unnecessary? In any case, since DeepSeek has shown how data can be better categorised and serve to a query more efficiently, reducing resources and cost drastically. This simply step can be incorporated into any existing AI model now to make it super efficient, so there is no breakthrough in this, is there?
There is no doubt that from George Boole's initial kick-off to the present day, the progress has been enormous, but calling a system that deals with patterns from probabilistic calculations of statistical information intelligence is too much. It's like the cloud that stores data!!! There is a lot of sophistry in these articles with the purpose of impressing the public. The one who liked this practice was Jobs...
Ok but from a consumer point of view there’s no search or thinking options at least not yet. Benchmarking is fine but the user is just interested in utility
You have missed that Microsoft and Amazon have integrated DeepSeek into Azure and AWS respectively. So, accusation by microsoft was just a drama after all.
@@dr.magnanimous8973 it’s better because if you live in the States your President could at any moment ban DeepSeek if he wakes up and just feels like throwing a tantrum. Plus you won’t have to worry about China spying on your company. Not trying to be arrogant
When an alleged AI cannot competently answer these follow-up questions, I know it's categorically non-conscious: "When and how did you learn that? - How might your belief be wrong? - What else do I have to know, to be confident that you're right?" An agent lacking either memory of its 'personal' history, or ability to criticize its beliefs, is not conscious - hence is not a person - hence not intelligent in the useful sense.
Well, I tested different models with the Banana- Transport-Problem. 3000 bananas need to be transported over a distance of 1000 km. 1 camel consumes 1 banana per km and can carry a maximum of 1000 bananas. What is the best strategy to bring as many bananas as possible to the destination? Only Deepseek did succeed and it took some time to test wether or not the solution 533,33 ist opimum. Tülu 3 405 B only produced nonsence. Even with help and tips it didnt undesstand the Problem nor the solution.
If Chinese EV makers were allowed to give away electric vehicles for free in the U.S. and flood the market, it would devastate Tesla and other American EV manufacturers. Demand for paid EVs would collapse, forcing Tesla and legacy automakers like Ford and GM into a brutal price war that could cripple their profitability. Tesla, which relies on high margins and brand loyalty, would struggle to compete, while smaller EV startups might go bankrupt. The U.S. government would likely intervene with tariffs, bans, or subsidies to protect domestic automakers, just as it has done in other industries. Isn’t this the same strategy China is using with state-backed DeepSeek AI, offering ChatGPT-level AI for free to disrupt the American AI market? By heavily subsidizing its AI models, China is making it difficult for U.S. companies like OpenAI, Anthropic, and Google to compete on price, potentially undermining their business models.
This is baked into literally every industry as companies receive their profit as a refund, only after pumping every cent into their companies in order to dominate the rest of the world. I haven't looked at it in a while, but it's kind of brilliant how China has viralized Chinese industry as a whole.
in that case, American will be more than welcome of these EVs and China will be at a great loss. Linux was and has been open and free. MS and Apple are still alive and doing even better.
the US and the G7 will create regulation that AI hosting requires a "Permit". No Open Source, and no Offline modes. Must use an Approved Paid for AI Program
I'm using AI to help me start writing again. (Writing was a passion of mine in high school. I got caught up in a bunch of other stuff and quit.) So... Keep 'em coming, I guess?
All of these comments about American dominance in the field ignore the basic science fact that among AI professionals communication is open and shared. They openly build off of each other's architectures. If you've ever taken a computer science class, you know that's true. They ignored basic security principles. Trying to get rich quick. And the Chinese recognized that not everything needs to be a competition. That we can work together and create something better. It's honestly genius. 🎉
lol , War ? 😂 , Deep Seek showed the right way to do it with their innovations and they open sourced it , everybody is copying their model and improving it so the war is over , the Chinese won the AI arm race 😊
its intersting when i ask what is your model it say : I don't have a specific model number like a physical product would. I'm a digital assistant powered by artificial intelligence, developed by OpenAI - seem they using open ai models
Shocking? Based on recent experience watching your videos, somehow I doubt it! “You keep using that word. I do not think it means what you think it means” - Inigo Montoya
I just reasoned that the "Tulu" in AI (Tülu 3). was a play on "cthulhu", the Lovecraft one. As the description of Cthulhu could be copy pasted over to an AI. (not the "physical" description). The part about the Cthulhu is described to be "The creature is said to be ancient beyond human comprehension". And people (believing it will one day awaken and rule the world). That fits AI pretty well. So yeah. I'm sticking with my I just reasoned that the "Tulu" in AI (Tülu 3). was a play on "cthulhu", the Lovecraft one. Mostly because it is more "sinister", and "ominous". ohh, ohhh, and mysterious.
There was "someone" in nature we could convince of the worth of human behaviors over monkey behaviors. It would take a freakin lot of noise real close to our human houses. And they would be spiders...Oo...
Venture capital in the early days, with the goal being to build a big user base. One you have 100M+ users you can push higher tier offers, new products & related services.
From 25th to 29th, 83 hours, DeepSeek server cluster, received more than 230 million DDos malicious requests per second, the total amount of attacks is equivalent to three days of European network traffic combined. A number of Chinese Internet companies came to the rescue to defend against the attack, and this epic story of working together to fight against foreign enemies will be forever recorded in history as a barbaric invasion was stopped.
If so, wouldn't it be fair if China one day takes equal "revenge" on "them"?
Yes, China standing against the dark demons and their army from Europe who been eating the earth for last 300 years , finally the tide is turning . One from China. Another from Gaza
Ok grok
@@aw888492 maybe its just high usage from around the globe
Chinas CCP thrusts a blade between ones shoulderblades and cries out in pain at the same time.
Thank you DeepSeek for stirring up all the shits in the Sillicon Valley. AI is no more a sacred cow now, every body can have it for free.
you are a sheep in a world full of wolf's, the candy they give you is not for your benefit
And your solution is?
It is somehow helpful than Chatgpt for the time being.
And what are you? Also how are people hiding their usernames?
Llama 405+deepseek= tulu 3😂😂😂😂
Without Deepseek this will never had happened. The BIG boys will keep it to them self. 😅
Without Meta open sourcing Llama and DeepSeek distilling from o1 against the ToS it wouldn't have happened.
Thanks DeepSeek for making this happen. Free for all. Hope more comes out open source.
They didn't do anything. This project was in the works for years before Deepseek.
Deepseek will be forgotten soon
Know this when you jump ship if you are not buying a product; you're the product 😅
Meta AI is open source as well.
@@marcduchamp5512 Don't write Deepseek off yet as it didn't state Deepseek R1. Anyway, Deepseek R1 will make a comeback IF it is surpassed now. I said IF is because the challengers have not been proven yet.
It's not a war, but provides more options for users.
But calling it a war makes it sound more exciting, while providing more options for users.
It absolutely is a war. Open Source Vs Corporate monopolization is one of the longest wars we've had.
Its the greatest weapon in any war. Consider it a weapon of war at least.
you blind ............. everyone that can see clearly see's a war
true
Glory to China contributing to a fair world
LMAO 🤣
Communists aftershocks
Agreed. Glory to them for doing what Americans, judging from the comments(? strange indeed), don’t think should be done for the sake of human liberty. Deepseek did the world a service. Are we reaching a point where people like me who see things clearly and with knowledge will have to shut up and say nothing when everyone parrots things that are contrary to reason? Lol
@@erandalee810 learn economics and political science dude. No place on Earth for communists nowadays. It is a public goods theory, oligopoly regulation, etc.
China walk the walk while US talk the talk.
So patriotic huh? Could the data be promoting nationalism?
Berkley built a model for $30 that beat deepseek
@@TheThetruthmaster1 when? or you mean a $30 cyber attack...😂😂
Yeah, we can tell that China has been leading the world in AI since the beginning.
China's *Walk the talk* USA *Talk the Talk* & India *Talk, talks and Still Talkshit*
The fact America invested Billions and lost Trillions makes America the loser in this… With Brics around the corner, saying America is a sinking ship is an understatement.
The United Nations actually started BRICS. It’s been around a while. They just let the general public think that it’s new. I laugh every time.
Keep telling yourself that
not really sinking, just in DeepS****
sh*t brics lol
@ 😉
Thanks to DeepSeek for giving courage to Tulu and others to come. Suchir Balaji spirit lives on.
RIP Suchir
It seem the "show" has to start soon. I've been thinking that for 15 years though. And it hasn't started yet. But, soon (for realzies, this time). AI is so strong I don't think anybody can keep the "cork" in that bottle anymore
Everyday there's shocking news... They don't shock me anymore because I know next shock due in few hours 🤣
⚡️
Hwawi Qwen 2.5 😂😂😂😂
It's one after another 😂😂😂
Today : shocking news
Next day : jaw dropping shockingly shocks world news
"Open" AI and the rest of MAG 7 are COOKED. Thank you DEEPSEEK :)))))))))))))
its only relevant if its free like DeepSeek
Microsoft accusing someone of theft... hilarious
It's a theft' call
@@wunxueAs original as the company's founding.
LOL too bad the Tulu 3 model is being benchmarked against Deep Seek V3 AND NOT R1 which is the their TOP model no?
And against 4o and not o1.
yes this is so misleading. the open source part is also not technically true
@@basharstats4482 DeepSeek is open weights not open source, the same as many other "open" projects. Tülu 3 is much closer towards real open source (the training data and code and instructions are available).
DS is open weights not open source, the same as many other "open" projects. Tulu is much closer towards real open source, the training data and code and instructions are available.
Tulu 405b is 405b parameters. R1 full is 671b parameters. Not a fair comparison (they’re not in the same weight class). Hence, why you don’t see the comparison present.
Beating 2 out of 12 benchmarks. What kind of fake news is this...
Deepseek will and ever be the king
it is still. this is outdated news from last December.
@@basharstats4482 CHECK THE NEWS TODAY THEN....👀..DEEPSEEK R1...STILL THE NUMBER 1 IN THE WORLD BY FAR........
That’s wishful thinking 😂
It will be overtaken in a month
I love deepseek
Deep seek shake the entire AI world and panic mode started 😂
Google was always free in the beginning, so don’t criticise Chinese for doing the same
what was free exactly? u mean gmail? 😂 or basic search?!! 😂😂😂
Isnt this misleading ? i mean DeepSeek created waves not for V3 but for R1 and this headline compares Tulu with V3
V3 you mean. And yes I agree.
Quote off techcrunch. "Ai2’s model, called Tulu 3 405B, also beats OpenAI’s GPT-4o on certain AI benchmarks, according to Ai2’s internal testing"
4o, lol. You guys are too late into the game, nothing to see here.
Just like the Alibaba video
all i want to know is if it has a txt2img and txt2vid platform along side of it thats ''free and 'open source' '' and if it runs on amd and utilizes my integrated cpu graphics out of the box without me needing to jump thru hoops for hours,.. like during install just click cpu or gpu box or amd or Nvidia box ..or type ''sudo - - cpu install'' in at some point...?
if it flies fast on 32 gigs of ram i'd be happy too..
😂 I'm so happy to see this! My tears of mirth are just that this was obvious within 15 minutes of working with any front-facing model. It was obvious that someone was going to do this. And I don't care because it's great. Long live a socialized world.🎉
Competition at its finest!
All these models had plans of putting a price tag on their products however DeepSeek ranked all of them so they have no choice than to make their products Open Source , the future is looking bright
Good now I can use Llama, Tulu 3 and DeepSeek to create my AI to erase all financial, economic, property, and legal records of the entire world!
Oooh. I like this idea.
You go, Mr. Robot! lol
Mrrobot❤
I am going to conquer the world now....
I tried ethically reporting this months ago. I told them that their models were utterly broken from the social engineering perspective. And they've ignored it. 😂😂😂
Me: how many r's are in strawberry?
TULU 3: There is only one "r" in the word "strawberry."
I thought you said it was good at math? Maybe it just makes up for it in phonics but how hard is it to count to 3?
Prompt: "Given a chicken, a shovel, a 1 gallon jug, a water source, and a ten gallon jug, what is the simplest way to measure ten gallons of water?"
Response: "The simplest way to measure exactly ten gallons of water using the provided items is to use the 1-gallon jug to fill the 10-gallon jug accurately. Here's how:
Fill the 1-gallon jug to its full capacity from the water source.
Pour the water from the 1-gallon jug into the 10-gallon jug.
Repeat steps 1 and 2 until the 10-gallon jug is full.
Since the 1-gallon jug has a precise volume, you can be sure that after exactly ten fillings, the 10-gallon jug will contain precisely ten gallons of water.
The chicken and shovel are not necessary for measuring the water in this scenario."
Prompt: "In fact, just filling the ten gallon jug with water is the simplest solution."
Response: "You are correct. My apologies for overcomplicating the solution. Given that you have a 10-gallon jug available, the simplest way to measure ten gallons of water is indeed to:
Fill the 10-gallon jug to its full capacity directly from the water source."
Okay, so the problem here is to measure exactly ten gallons of water using the items provided: a chicken, a shovel, a 1-gallon jug, a water source, and a ten-gallon jug. Hmm, let me start by breaking down what each item can do. The main goal is to get ten gallons measured out. The ten-gallon jug is obviously key here because it can hold up to ten gallons. But how do we make sure it's exactly ten gallons?
First, let's consider the water source. Since we can get water from it, we don't have to worry about running out. The shovel and the chicken might be distractions or red herrings because they don't seem directly related to measuring water. The problem is probably testing if I can ignore irrelevant items and focus on the relevant ones: the jugs. But maybe the chicken or shovel has a use? Let me think.
The 1-gallon jug is our only measuring tool besides the ten-gallon jug. So, perhaps we need to use the 1-gallon jug to measure water into the ten-gallon jug. But if the ten-gallon jug is empty, we could just fill it up directly from the water source, right? Wait, maybe the ten-gallon jug isn't empty? The problem doesn't specify the initial state of the jugs. Hmm, that's an important detail. Usually, in these types of problems, you have to assume they start empty unless stated otherwise. Let me check the question again.
The problem says "measure ten gallons of water." It doesn't mention anything about the jugs being filled or empty. So, if the ten-gallon jug is empty, the simplest way is just to fill it up to the top from the water source. That would give exactly ten gallons. But maybe there's a trick here. Maybe the ten-gallon jug is not calibrated, so you can't just fill it up? Wait, but if it's a ten-gallon jug, it should hold ten gallons when full. So filling it completely should give ten gallons. But the problem might be implying that you need to use the 1-gallon jug to measure it out, but that seems unnecessary if you can just fill the ten-gallon jug directly.
Alternatively, maybe the ten-gallon jug is already partially filled, and you need to measure ten gallons using the 1-gallon jug? But the problem doesn't specify that. Let me re-read the problem to make sure I didn't miss anything.
"Given a chicken, a shovel, a 1 gallon jug, a water source, and a ten gallon jug, what is the simplest way to measure ten gallons of water."
So, the items are listed, and the task is to measure ten gallons. The chicken and shovel might not be needed. The key items are the 1-gallon jug and the ten-gallon jug. But if the ten-gallon jug is available, the simplest way is to fill it up from the water source until it's full, which would be ten gallons. However, maybe the jugs are not empty. But the problem doesn't mention their initial state, so by default, we can assume they're empty. Therefore, filling the ten-gallon jug to its capacity would give exactly ten gallons. The 1-gallon jug might be a distractor. But let me consider another angle.
In classic water jug problems, you often have to measure a certain amount using jugs of different sizes. For example, the classic problem where you have a 3-gallon and a 5-gallon jug and need to measure 4 gallons. In those cases, you have to transfer water between jugs to get the desired measurement. But in this case, the desired measurement is exactly the capacity of the ten-gallon jug. So if you just fill the ten-gallon jug, that's it. So why include the 1-gallon jug, chicken, and shovel? Maybe to throw us off. The simplest way is to fill the ten-gallon jug. Unless there's a condition that we have to use the 1-gallon jug, but the problem doesn't state that. It just asks for the simplest way.
Alternatively, maybe the ten-gallon jug has no markings, and the only way to be sure it's ten gallons is by using the 1-gallon jug ten times. But that would be tedious. However, if the ten-gallon jug is designed to hold exactly ten gallons when filled to the brim, then filling it once should suffice. Unless the problem assumes that the ten-gallon jug isn't reliable, but that's not mentioned.
Another thought: perhaps the problem is expecting you to recognize that the ten-gallon jug can be filled directly, making the other items irrelevant. The mention of the chicken and shovel might be there to test if the person gets distracted by irrelevant objects. So the answer is simply to fill the ten-gallon jug with water from the source, resulting in exactly ten gallons. The other items are unnecessary for this solution.
But let me double-check. Suppose the ten-gallon jug is not empty, but the problem doesn't specify. But since it's not specified, the assumption is that you can use the jugs as empty containers. So, fill the ten-gallon jug to the top. Done. That's the simplest way. The 1-gallon jug is not needed unless there's a trick, but I don't see it here. The chicken and shovel are likely just distractions. So the answer is to fill the ten-gallon jug completely using the water source, resulting in ten gallons measured.
Oh dear 😂😂😂 Deepseek it is then
what an insane week! 3 groundbreaking AI releases. ClosedAI is fucked!
Tulu isnt that great their scores are far less than Deepseek R1 model. It just a hype for Tulu
This is one of the wars that I do want to see....
Let the race begin
Please explain how you make this robot says what you saying, that look so cool!
Imagin how good these model's will be in 5 years crazy.
Its not a war, its a race. They are not destroying each other. They are trying to out-perform.
I swear AI progression is just one month they announce and the next they release.
just a hype like nfts :)
Yeah we will have kids training AI during lunch time one day
@@adamspice9076 Not like nfts. Much more comparable to the DotCom bubble Because even though there was a bubble, the internet wasnt bullshit hype that disappeared forever after the crash, and neither will ai.
companies will compete, profit is ours
It's interesting to see how fast each iteration is coming out to beat the one that is on top today lol, at this rate we will reach parity in no time :) and considering how smart the average Human is, it shouldn't take much to be equal to them :).
REWARD? WTF?
Do they give it a kiss on the cheek? Candy?
😂😂😂😂😂😂
They give it a 1 or a 0
The quality of these videos is amazing.
Open source everything plz ty. Appreciate the content you provide, keep up the great work. :)
Greed always loses.
Try asking it what company created it or what model it is... Here is a hint OpenAI and ChatGPT 3. I was really disappointed when it started tell me stuff like that.☹
Microsoft being angry about someone stealing technology is real funny considering the history of how they became successful 😂
"It's not free It gives you only five messages. and weaker than the Chinese giant DeepSeek R1 ."
"Exactly! I tried DeepSeek, and it's way better than it, plus it's free! 😂"
"This is not a response to China, because they had already talked about it in December, saying they would release it by the end of January. If they really want to respond to China, they should make version 03 free, not 03 mini."
Your energy is truly remarkable!
Perplexity now has DeepSeek RI. I wonder if they will also add Tulu 3 405b?
i doubt it, deepseek was a bandwagon hype train. no one is talking about lulu
@@teebuWho is worried sick about Deepseek...and hyperventilating endlessly...and dropping innuendos...that they "copied..stole data..)from the billion dollar CLOSED models....?
Lose the creepy robot. 0:55
Yeah that thing was wild
We’re basically about to see the first Corpo wars like in Cyberpunk
DeepSeek, starting this free-to-use AI trend is amazing! Now we have QWEN and Tülu 3, all free to use. I HOPE that in the not-too-distant future, all AI tools won't require payment for unlimited access😁. Good job, DeepSeek🤣
I'i hope too, but even if AI is free on the web, it will never be for AI agents 😢
they have to figure out how to make money with their billions invested. China can do it but the US will have to accept a new system. it seems like China's communist system is edging out the Capitalist
I don't know who the biggest winner long therm will be but the biggest loser is already known: its humanity.
Well if it is humanity that is the loser there’s nothing you can do about it and you can kiss your ass good bye. Alternatively you can look on the bright side and think about how technology has worked out so far for humanity and perhaps imagine all the opportunities AI presents for us. Maybe look for opportunities for yourself and your family. Good luck.
Humanity is the winner.
Idiots who give us closed source AI are the biggest flubs!!
@dannycbe949 We will see over few years when 95% of IT jobs willbe gone and you will spend your day between garbage cans. That is if you still will be around. AI is the worse disaster to come for common people and will destroy societies, culture, knowledge and huge pieces of humanity. Its already destroying YT with tons of unwatchable garbage with generated "facts" fotos and voices.
The greatest feat for deepseek it's not his power, it's the fact that it runs on a Potato Computer (for OpenAI standards, of course)
Making 4rd account in 2 years ,1. 2023- chat gpt - NOW semi paid , 2. 2024- black box Ai - Now paid , 3. 2025 - deep seek -opensource but crashes a lot , Today 4. Tülu 3 - OPEN SOURCE let see i am enjoying these models they are very helpful for me from last 2 years - thank you all developers for These AI models THEY HELP ME TO COMPLETE ASSIGNMENT , SWITCH MY JOB , UNDERSTAND FAST AND NOW I DARING TO START DOING CODING
I don't understand this video. It says Tulu uses massive resources owing to its doubling of parameters. Isn't this what DeepSeek is proving to be unnecessary? In any case, since DeepSeek has shown how data can be better categorised and serve to a query more efficiently, reducing resources and cost drastically. This simply step can be incorporated into any existing AI model now to make it super efficient, so there is no breakthrough in this, is there?
There is no doubt that from George Boole's initial kick-off to the present day, the progress has been enormous, but calling a system that deals with patterns from probabilistic calculations of statistical information intelligence is too much. It's like the cloud that stores data!!! There is a lot of sophistry in these articles with the purpose of impressing the public. The one who liked this practice was Jobs...
Tulu response was based on the context and information available up to September 2021
Ok but from a consumer point of view there’s no search or thinking options at least not yet.
Benchmarking is fine but the user is just interested in utility
Deepseek open the pandora box. Now the big company cannot act like they used to be.
You have missed that Microsoft and Amazon have integrated DeepSeek into Azure and AWS respectively. So, accusation by microsoft was just a drama after all.
The whole point is that it’s open source and it’s not from China.
You guys can use it a literally rebuild it to your liking. Make it better.
As if any non-China version is better.
@@dr.magnanimous8973as if "american exceptionalism" is true
pure arrogance but actually ignorance 😂 maga!!! 😘❤️⚡️⚡️⚡️
@@dr.magnanimous8973 it’s better because if you live in the States your President could at any moment ban DeepSeek if he wakes up and just feels like throwing a tantrum.
Plus you won’t have to worry about China spying on your company.
Not trying to be arrogant
Awesome take, good work as always. What is the application at 0:21? It feels like ComfyUI but for LLMs?
It's not longer a monopoly. We have a choice to choose.
Interesting how chock full of bad takes and misinfo everything is. Deepseek did the world a service, that is all. It’s not a competition for them.
We will become their pets, i mean in a good senario! In a bad one.....see ya.
We are possibly already living in a simulation controlled by AI. The evidence is growing in that direction...
You will need to build a lead lined bunker to hide from the terminators...
How do you got all this outperforming statistic when deepseek was only introduced less than a week?
There are AI Benchmark tests kinda like an SAT for ai
They threw an all nighter
When an alleged AI cannot competently answer these follow-up questions, I know it's categorically non-conscious: "When and how did you learn that? - How might your belief be wrong? - What else do I have to know, to be confident that you're right?"
An agent lacking either memory of its 'personal' history, or ability to criticize its beliefs, is not conscious - hence is not a person - hence not intelligent in the useful sense.
Great questions! To be fair though the ability to criticize one's beliefs is extremely rare amongst humans as well.
Well, I tested different models with the Banana- Transport-Problem.
3000 bananas need to be transported over a distance of 1000 km. 1 camel consumes 1 banana per km and can carry a maximum of 1000 bananas. What is the best strategy to bring as many bananas as possible to the destination? Only Deepseek did succeed and it took some time to test wether or not the solution 533,33 ist opimum.
Tülu 3 405 B only produced nonsence. Even with help and tips it didnt undesstand the Problem nor the solution.
These names are starting to sound like galaxies and distant stars! I'm all for it but these guys should consider a name people will remember
If Chinese EV makers were allowed to give away electric vehicles for free in the U.S. and flood the market, it would devastate Tesla and other American EV manufacturers. Demand for paid EVs would collapse, forcing Tesla and legacy automakers like Ford and GM into a brutal price war that could cripple their profitability. Tesla, which relies on high margins and brand loyalty, would struggle to compete, while smaller EV startups might go bankrupt. The U.S. government would likely intervene with tariffs, bans, or subsidies to protect domestic automakers, just as it has done in other industries.
Isn’t this the same strategy China is using with state-backed DeepSeek AI, offering ChatGPT-level AI for free to disrupt the American AI market? By heavily subsidizing its AI models, China is making it difficult for U.S. companies like OpenAI, Anthropic, and Google to compete on price, potentially undermining their business models.
This is baked into literally every industry as companies receive their profit as a refund, only after pumping every cent into their companies in order to dominate the rest of the world. I haven't looked at it in a while, but it's kind of brilliant how China has viralized Chinese industry as a whole.
Deepseek is a private company that is even less well-known in China, and reportedly Deepseek is not a company that focuses on AI research.
China now shift to EIV - Electric Intelligent Vehicle, no longer EV.
in that case, American will be more than welcome of these EVs and China will be at a great loss. Linux was and has been open and free. MS and Apple are still alive and doing even better.
Can you please let me know what you are using to generate your avatar? The nuanced facial expressions are amazing.
the US and the G7 will create regulation that AI hosting requires a "Permit". No Open Source, and no Offline modes. Must use an Approved Paid for AI Program
they build a great wall around china as well😂
How do you "reward" an AI exactly?
you give it a cookie
Human snack😂.
I give superlatives to Claude all the time
don't your think this out of date, peek deepseek model are R1 not v3, chatGPT are o3-mini-high not 4o
Same with the Alibaba video
I'm using AI to help me start writing again. (Writing was a passion of mine in high school. I got caught up in a bunch of other stuff and quit.) So... Keep 'em coming, I guess?
Only compared to dsv3 not dsr1. Did I miss something?
All of these comments about American dominance in the field ignore the basic science fact that among AI professionals communication is open and shared. They openly build off of each other's architectures. If you've ever taken a computer science class, you know that's true. They ignored basic security principles. Trying to get rich quick. And the Chinese recognized that not everything needs to be a competition. That we can work together and create something better. It's honestly genius. 🎉
Really like this kind of war, not even having suffering but having more and more benefit to me
lol , War ? 😂 , Deep Seek showed the right way to do it with their innovations and they open sourced it , everybody is copying their model and improving it so the war is over , the Chinese won the AI arm race 😊
This AI stuff makes me want to watch Star Trek. That's what's coming baby lets do it
Interesting! Thank you.
😍It just reflects the real world. Good to see 👀 global open AI developer community competition❗️💪 Let the free AI INTELLIGENCE shine❗️🌟✨️ 😂😎
What about the energy consumption?? Does it match R1?
We've finally entered Skill based Matchmaking mode
Who's gonna pay for the effects of all the energy that 405b is going to need?
The model gets a reward if task completed successfully ? What, are they feeding it bits as cookies :D
At this point, why not just make a ring and have each AI robots fight it off like those remote robot fights x.x
its intersting when i ask what is your model it say : I don't have a specific model number like a physical product would. I'm a digital assistant powered by artificial intelligence, developed by OpenAI - seem they using open ai models
Does it run locally?
It´s censored, so it´s just like the others. I am happy that i am running
uncensored one locally on my pc.
Not so fast. 405B failed a simple test:
Q: "How many rs are in the word strawberry?"
A: "The word "strawberry" contains 1 letter 'r'."
Shocking? Based on recent experience watching your videos, somehow I doubt it!
“You keep using that word. I do not think it means what you think it means” - Inigo Montoya
The PC will need more RAM since the current 256GB doesn't seem to be enough.
Where did an academic team get the money needed to build a competitive chatbot?
Wow. I didn't realize cheap Chinese models could out perform o3-mini!!
I just reasoned that the "Tulu" in AI (Tülu 3). was a play on "cthulhu", the Lovecraft one. As the description of Cthulhu could be copy pasted over to an AI. (not the "physical" description). The part about the Cthulhu is described to be "The creature is said to be ancient beyond human comprehension". And people (believing it will one day awaken and rule the world). That fits AI pretty well. So yeah. I'm sticking with my I just reasoned that the "Tulu" in AI (Tülu 3). was a play on "cthulhu", the Lovecraft one. Mostly because it is more "sinister", and "ominous". ohh, ohhh, and mysterious.
Im downloading it now thank you for this information
There was "someone" in nature we could convince of the worth of human behaviors over monkey behaviors. It would take a freakin lot of noise real close to our human houses. And they would be spiders...Oo...
They use Gpt 3.5 free version for training purposes.😂
Any suggestion how to build a AI model for adult contents?
Tulu has only 495B parameters but Deepseek has up to 617B parameters. Research well sir
It sounds so true like US landing on the Moon 😅😂😂😂
How do these companies make their income?
Venture capital in the early days, with the goal being to build a big user base. One you have 100M+ users you can push higher tier offers, new products & related services.
I hope there is microbiologist studying how cells transform materials into a plant. That will help speed up plant growth using qiantumbots.