DeepSeek R1: This Free AI Model is Mind-Blowing.

Hey friends hope you're well in case you missed  it DeepSeek R1 has been making waves in the AI and Tech scene. It's an open-source AI model that  was apparently developed for less than $6 million a fraction of the billions of dollars spent by  OpenAI and Google for example to create their AI models the good news for all of us is that  DeepSeek's free to use and it's shot up to the most downloaded app on App Store surpassing  ChatGPT within days it's now one of the most advanced free and open source AI models we can  use I've been playing around with deep seek R1 for a couple of days and I have to say it's  gamechanging but it's not without its flaws so let's run through what deep seek R1 is capable  of and see what the fuss is all about so let's get up to speed one of the main reasons why deep  seek R1 is so hyped is because it doesn't rely on expensive human label data sets or supervised fine  tuning which is how most AI models are trained and it costs Millions if not billions instead deep  seek R1 uses a self-reinforced learning method without the need for human supervision and  effort you can think of supervisor fine tuning like teaching a child to cook by writing up a long  and precise recipe and then showing them step by step while reinforcement learning is allowing the  child to sort of like experiment in the kitchen and gently guiding them when dishes don't turn  out well so they're learning through trial and error and that's exactly how deep seek was trained  and The Benchmark results are incredible on the aim 2024 mathematics Benchmark deep seek achieves  71% accuracy while gb1 mini achieves 63. 6 accuracy and on the math 500 Benchmark it beat both 01  mini and 01 0912 but it performs worse on coding tasks in code force and Live code benchmarks but  of course say much more to benchmarks so let's jump onto the laptop and I'll show you what  I found while playing with deep seek over the past couple of days so jumping onto deep seek.

com  here's where you can create an account or you can go ahead and download the app on your phone but  currently their servers are super slow because of the crazy demand so I recommend avoiding  signing up with an email you'll be probably waiting forever for an email verification code  so I suggest logging directly through a Google account so once you're in here toggle on the Deep  think R1 model here it's an advanced reasoning model similar to gpt's 01 model but without gpt's  01s 50 message per week restriction and also R1 is able to work alongside internet search toggle  this toggle right here simultaneously something I believe 01 still can't do yet okay so R1 model  uses the Chain of Thought prompting approach which basically encourages es the A1 model to break  down the reasoning into simple to understand steps this isn't new but deep seek R1 does  this really well so let's use this simple math problem as an example the first part here is the  problem to solve uh and the second is the prompt that I've added to show its Chain of Thought So  that's specifically let's solve this step by step for each step explain your thinking and show  your calculations so hitting enter you can see deep seek thinking and reasoning with itself and  this is what makes R1 different it transparently reasons through each step individually and figures  it out in the same response in real time whereas GPT can often be sort of clinical and political  I found deep seek R1 to be direct but also great at showing you the reasoning and you can also  extract the reasoning and send it to other AI models too something that's unique to deep seek  R1 the other cool thing is how deep seek R1 solves hallucinations so hallucinations is a term to  describe when AI gives you an incorrect answer and it's a big challenge with current AI models  but I've noticed that R1 is particularly good at understanding why it hallucinates almost as if  it's truly self-aware and then it also corrects itself so I started recording this specific clip  here when I noticed that it gave me an incorrect answer to the vague question of what happened  to Hershey's in 1998 it says Hershey's launched Arman kisses in 1988 when in reality they were  actually launched in 1990 so I pointed out the mistake and asked why it made the mistake because  of its Chain of Thought approach it's fascinating to see it run a search on this mistake confirming  why it made a mistake and then it corrects itself here compared to other AI models deep seek R1  thinks way more naturally almost humanlike and elaborates on its mistake clearly so I highly  recommend challenging R1 when it hallucinates and give this a go yourself it does seem to  be slower though than chat jbt 40 especially when it comes to coding tasks I've been playing  around with creating games on deep seek uh like if we ask it to create a Tetris game and and then  take the python code and run it in HTML it takes longer than it would in 40 before you can preview  the game right from the chat so if you have coding tasks 01 and particularly Claude 3. 5 son a still  does a better job overall and will help remove the need to debug as a coder but if you're looking  for a free option or an open source option R1 here is definitely the way to go currently and worth  checking out so based on my short time with R1 I feel like deep seek was probably trained on GPT 4  o generated data the responses on both models are eily similar and if you're concerned about privacy  but still want to leverage deep seek R1 you can actually run it locally because it's open source  you can download and use the AMA app to run this R1 model on a local server so all your questions  and interactions remain completely private rather than on the cloud but it is a very large model  so you'll need a beast of a setup to run its full R1 model locally it's roughly like 1,300 GB  of vram that you'll need to run it fully but there are distilled llm versions of R1 that run on a  single GPU version 1.