💬┊general

[foggy flicker]

but main catch is need whole of data

on memory

[agile cave]

the best i have is train it offline and then import the weights for it in submitted code. As far as I can tell, allowed by competition:

(from competition page)

there is not enough RAM allowed for each participant, to train with this approach, in cloud. But you can on your own. And save it on kaggle as a set of data available to any program (kernal) you have

This has to be the worst LLM out there, btw:

[foggy flicker]

don't got tht much compute locally as well☠️

[agile cave]

What I'm saying is you compute this offline, without limitations. You uplaod weight generated to Kaggle account. You read and use weights in submitted file to solve problems. Get it?

[foggy flicker]

I don't have the compute for it locally aka offline

[agile cave]

How old is your system?

?

[foggy flicker]

only got 16gigs of RAM on my laptop but need at least 32 for this coz the data itself 12+ GB

[agile cave]

And your training logic can't fit under 4 gigs?

remember, once training is done, weights will not be 12 gigs

[foggy flicker]

OS itself consumes RAM & so does VS code, Kaggle already provides 16gigs & it wasn't possiblle

[agile cave]

you want to do it all directly with a Kaggle notebook. Might not be possible. You may have to do some of it offline. Do you get it?

offline, you are figuring out weights for XBG to be used

you know, it would be simpler if you sated what you want to do and how you think you could acheive it, technacalities aside. DM me, if you prefer. I get it. It's an ongoing competition. Not mine personally (I'm all about the Chess one).

[foggy flicker]

I just been playing with sample notebooks so far

[rocky spoke]

i have some doubts in DL can you please help me

[agile cave]

feel free to

[brisk eagle]

can somebody tell me, how can i disable this feature, its so irritating man

@serene imp

@fair cliff

[analog seal]

Hey, any one else getting problems with TPUs? while I'm training my model, it freezes after a while....

[vestal ruin]

Anyone can give me a suggestion for good datasets related to healthcare.

[fathom rover]

Which ide this is?

[brisk eagle]

Not IDE, its kaggle notebook

[severe storm]

what you want to build and which type of data you need?

[fresh basin]

Hello guys....completely new to kaggle and this area

Can anyone guide me how to get started with machine learning and start participating

From scratch

[vestal ruin]

Any type of data would work.

[deft lynx]

@fresh basin if you're interested in generative AI, try this for an introduction: https://www.kaggle.com/learn-guide/5-day-genai

[fresh basin]

@deft lynx thx..like I want to compete on kaggle so what's needed before I jump into the competitions

[deft lynx]

Hi @fresh basin to be honest I have yet to have time to join any, but do learn more. Kaggle, huggingface and many sites have plenty of tutorials. Have you signed up to kaggle? You can explore available competitions here:
https://www.kaggle.com/competitions

@fresh basin sometimes competitions are announced in the general announcement channel: https://discord.com/channels/1101210829807956100/1128429049736335371

[severe storm]

I think go through WHO database

[feral badger]

hy

[flint aspen]

guys anyone here i need help urgent , im working on classification ml model and i have this restaurant rating , thre is categorical , numerican , binary and one cuisine column that contains various cuisines seperated by comma , so task is to classify restaurant based on cuisines , how u do it exactly , i did on model where i splitted all the cusiines and got very low sxore

[feral badger]

it's due to overfitting ig

[manic lava]

Hello

[true otter]

Hello

Cool name where are you from

[vestal ruin]

Okay, I'll have a look at it. Thanks.

[agile cave]

numerican? 🤣

Seriously though. What are you trying to do? unsupervised classification or regression/ prediction?

[agile cave]

I feel you

[reef hollow]

Hi all, I'm Solomon Christ (Also known as Mr Metaverse) and I'm diving deep into AI and ML and excited to be here. I'm totally new to it all 🙂 Thanks for having me!

[cunning dove]

hi

[worn jungle]

Sweet. I’m in.

[uncut rose]

Hi, everybody.
Is there anyone who knows HMM(hidden markov model) tokenization?

[deep nebula]

hi

[agile cave]

Work through the Kaggle "Learning" series. It will give you a grasp of basic solid tools you'll use. Once you are done, you can expand on your own (and will know how you want to expand).

[agile cave]

Not quite sure what you mean by tokenization. Good to think about HMMs: Hidden Markov model (HMM) can be considered as a collection of probability tables, specifically representing the probabilities of transitioning between hidden states ("transition probabilities") and the probabilities of observing certain outputs given a hidden state ("emission probabilities") - making it a probabilistic model for sequences of data where the underlying states are not directly observable. The states aren't vectorized (turned into numerical values) with HMMs.

Take for instance an HMM coming up with music. It will calculate the probabilities of another note being played (or chord) based of the most recently played note/chord (or the last n). The inputs, last note/chord, or sequence of notes/chords will be fed into model as straightforward categorical data. for instance (C#, D, A) or whatever.

[uncut rose]

Thanks for response.
I mean tokenization for llms.
I know BPE, wordpiece and unigram widely used in llms.
I think we can combine hmm and subword tokenization.

[nova merlin]

Do anyone knows any free model to use in scikit-llm ?

[woeful fiber]

Hey everyone! I'm Ranjith, a Software Engineer and a newbie in the world of AI and Machine Learning. I'm deeply passionate about these fields and excited to explore and learn more in this space.

[vale pine]

How did I receive kaggle 1 year badge when I my account is only 6 months old

[steep forum]

We Wish you all a Very Happy and Prosperous New Year 2025

[nova merlin]

do anyone use uv package manager ?

[cyan berry]

Hi!
Just dive into Kaggle!

[wary oyster]

While using sweetviz for my EDA there is a numpy error...what do i do!?

I have updated all my packages... But it didn't work!

[nova merlin]

do anyone using uv package manager ?

I cant find a way to uninstall all uv venvs

[fathom rover]

So guys For those of us on windows doing titanic competition you may have noticed that tfdf is not there on windows rather we are being provided with ydf. Now the thing here is that ydf cannot convert the dataset from keras to tensor is there any work around for it in local environment except for wsl?

Nvm it works with python 3.11 looks like I will have to downgrade

[river pagoda]

hi everybody, i would like to kwon some kind of place where i can find information about data analyst, i want to get information about tobacco and implements new solutions in this place, if you have some information and also, if you want to guide me for the best road, let me a message .

best regards

im a little beginer trying to learn data analyst for to take my profesional degree to another place

[agile cave]

Everyone did

[fathom rover]

Yea I know that it's just that I was following a decision forest tutorial notebook in competition that's where the syntax was stated

[agile cave]

coding AI ...

[tough pollen]

Hi all

[earnest flame]

https://www.kaggle.com/akandeayobami

[visual summit]

how do i clear the cells of the practice questions in the programming courses?

[uncut rose]

Hello, everybody.
I saw the llm models supported by vllm.
But, can we use quantized variants of those models using vllm?

[uncut rose]

Hi, everyone. Please tell me what is the difference between using API and open source models.
Well, when we use open source models on serverless gpu, how much money can we save?
Thanks.

[agile cave]

An API is a set of prewritten Classes and functions you may use.

Open source models tend to be models and/or their weights they use to perform some type of calculation.

[agile cave]

Why don't you check out and familiarize yourself with BERT, a text tokenizer

[cyan prawn]

What should my college major be

[agile cave]

You can always pick up new skills after college. Something you are interested in. Something you commit yourself to excelling as much as you can in. This will transform you. It may not be what you do next. Changes often, in y XP, over a lifetime.

[uncut rose]

you mean my try for hmm tokernizer?

[agile cave]

I wouldn't use HMMs. They are really about prediction, espcially when dealing with NP Hard problems (wtf problems). What BERT tries to do is spatialize token vectorization and map meaning to spatial (numerical) coordinates in a manifold. I think original Bert had 768 dimmensions, per token.

WIth that said, clear you are thinking about creative solutions. Continue that.

[uncut rose]

Yes, I know BPE, wordpiece and unigram.
But, I think, those are based on vocabulary and loss few information due to unseen tokens or words.
In some cases, that can be sensitive and important. So, we're adding hmm for dynamic tokenization for that.

Thanks, Dahoud.

[agile ferry]

Anyone want to collab on any interesting project!

if u have any interested project idea we can collab and learn!

[nova merlin]

dl nlp or cv ?

[agile ferry]

nlp

dl

do u have any project idea?

[nova merlin]

arc-agi competition (what do u say using GNNs ?)

[agile ferry]

projects instead of competition!

[nova merlin]

Ok but on GNNs (since it's interesting n very parameter efficient)

[agile ferry]

ok I don't have any idea about Gnn but we can try it!

[nova merlin]

https://distill.pub/2021/gnn-intro/#graph-to-tensor

[agile ferry]

wow It's interesting!

Just learnt something new today!

[nova merlin]

there's alot

[agile ferry]

so what do u like in gnn! and difference b/w ann and gnn

[nova merlin]

GNNs as u saw re not limited to sequences or grid like structures graphs can be alot general n maybe could be a key to agi(artificial "general" intelligence)

[agile ferry]

have u trained a simple problems using gnn and ann any metrics difference?

[nova merlin]

no not yet was just lost in theory will try going through docs from tomorrow

[agile ferry]

u can answer!

i think gnn has very useful in supply chain and Transportation problems

[nova merlin]

well u can't compare simply GNNs usually work on graphs like structures but when combined with other ANNs they can be quite powerful

[nova merlin]

I was already struggling with understanding Shannon 's entropy bro 🤧 (anyways the paper was too deep n better than few thesis-es) anyways I will consider it properly tomorrow

[nova merlin]

bro i can't understand no. of bits needed per outcome on avg even in uniform distribution where S(E) not power of 2 😔

[nova merlin]

I mean to generalise Shannon entropy for non uniform distribution from uniform distribution

[nova merlin]

Compare ur kaggle notebook with this

[nova merlin]

Bro pls explain how for each outcome in uniform distribution in the video, entropy was log2(1/p) bits

Still it just talked too abstractly n without even algos just shown data mostly

even grok can't give explanatory explanation about this generalization

damn true thnx

I m trying

[wicked berry]

Hi I need some advice as I have never done research before it's my first time I need to do some research on how ai can be integrated in aviation like both helicopter and large ones so what resources over the internet should I look for or how to start it I am confused...

[median jackal]

Hey Everyone ! Myself Abhishek new to discord community , nice to meet you all

[burnt quiver]

Hi, I am new to discord here to learn data science and skills to build up my cv for my upcoming campus placements!

[worn flare]

What's up, everyone! I'm here to continue growing in data analysis!

[uncut rose]

Hi, everybody.
Who knows deepseek well?
Which version of deepseek is for deepseek chatbot?

[formal tendon]

Hello World 🌎

[viscid epoch]

Hi 5-Day Gen AI Intensive Course with Google brought me here.
Happy New Year and ready for more "learning" 😉

[sullen gale]

Can anyone please tell me why I keep gettimg 404 when trying to load Kaggle?

[devout kettle]

What do you mean by that can you share the link?

What is the meaning of this 399 % cpu usage?

[noble pumice]

hello!

[glacial rock]

hi!

[glacial rock]

Multiple cores in the cpu, each core can go up to 100%. If 4 cores and they are 100%, 99%, 99%, and 99% usage, adds up to 397.0%

[upper quartz]

https://www.kaggle.com/datasets/steve1215rogg/student-lifestyle-dataset my faculty has asked me to prove the validity of this dataset. how can i go about doing that? please help 🥹

[sweet merlin]

hey guys i am a student looking to team up for a question. dm if interested.

[uncut rose]

Hi, everybody.
I'm looking for ar related to project where that can detect palne and argument the objects on that.
If there is anybody who knows, please tell me.
Thanks.

[fierce sundial]

Hi All, I am cybersecurity professional interested in AI. Thanks

[rocky spoke]

Hey guys I recently joined the Playground Series - Season 5, Episode 1 and have noticed that Iam getting a low train and test mape score(on splitting on the training data somewhere around 0.06-0.07) but while I do my predictions on the test data I am getting a very high mape score(19) could anyone please help me

[rapid yoke]

Hi everybody, I'm a medical physicist and interested in AI in medicine.

[mild vault]

slenderman

[mystic monolith]

Hi, math

[uncut rose]

Hi, everybody.
We're building news analysis models and need to collect news data of 20 years.
Is there anybody who knows news data service well?
Please tell me.

[mint yoke]

Scrap bro

Hey any grandmaster, how good are time series forecasting deep learning models to get edge in trading ?

[bright notch]

Hello guys
Looking to learn more about Data Science

[bright notch]

How do I get the PDF?

[agile cave]

possibly: https://archive.org/details/deep-learning-a-visual-approach

[main blade]

Hello everyone, I Trained my model on GradientBoosting Regressor. Now, I wanna trying to build page. that take input from user and predict price. But when i try to build on Pycharm, there is an error. maybe bcz i use Gradient Boost. so, how can i solve it ???

[bright notch]

Thanks

[agile cave]

First off, I can't even see what error message is (pic is too small). But no, GradientBoost model is not goig to be your problem, if you have a succesfully created one. The issue lies in incorparating it into your age. Assuming you are running algo server side, you just need to build (code) a better bridge between collectings user input and processing it with your GBR model. Also, I prefer VSC to PyCharm. More powerful, especially with CoPilot builtin.

[worldly hemlock]

Anyone have any idea how to handle the phone verification issue? No code is being sent and it keeps saying too many requests (even on first try) when I press send code

[wheat tulip]

hi

[main blade]

Thanks, Let me try..

[formal spruce]

Hi everyone, I am data center service provider professional interested in AI. Thanks!

[hearty axle]

hi

[vale pine]

What does fork mean?

[short wedge]

Fork means to make a copy of a piece of code. So in this case, other users made copies of your notebook (yay!).

https://en.wikipedia.org/wiki/Fork_(software_development)

[vale pine]

Thank you 🙏

[agile cave]

It would be nice to have a VC channel or two, so we could come and discuss all things ML. Just a thought...

I'm sure people have ideas and discoveries they would enjoy dropping and brainstorming with people live

[nova merlin]

it would be more nice if there were vc to brainstorm with ai

[digital lodge]

https://tenor.com/view/duh-gif-11084198450447296338

Yayy! I finally landed on Kaggle's Discord. Nice to see y'all 👋 ...

[timber gorge]

ive been getting this error on kaggle for a few mins on both pc and phone

does anyone know what the cause of this is

[agile cave]

ill formatted json file maybe?

[timber gorge]

i mean this is kaggle.com, i cant access the site

says too many requests, i think i tried to do too much stuff or had multiple tabs open

[hallow thistle]

You mean like speaking/ voice chat to Alexa ?

[agile cave]

I meant brainstorm ABOUT ai, via VC channel

[nova merlin]

Yes but more like brainstorming also with llms not voice assistants (somewhat like in notebooklm)

[hallow thistle]

Speech to text and text to speech? Just use Apple Shortcuts on your phone to talk to a ChatGPT as that will keep the context in that conversation for multiple iterations.

[nova merlin]

How to use apple shortcuts in non-apple phone

[hallow thistle]

Don’t know much about that. 😩

[eager brook]

Hi everyone please I need list type of flat file. What the all types of flat file

[jovial imp]

I am not to verify my kaggle ID plz help

[zenith sky]

https://www.kaggle.com/code/omidsakaki1370/sticker-sales-eda-forecasting
If my notebook was useful for you, please give me a upvote. Thank you for your support

[forest epoch]

When creating sentiment analysis, do you also use external sentiment lexicons? If so, could you suggest a good one? Or do you rely on libraries for better precision?

[jovial imp]

@Kestin Merritt Moon Please help me here. I am getting rejected and contact support has been useless since

@ke\

@feral kestrel Please help me here. I am getting rejected and contact support has been useless since

[mint rapids]

https://www.kaggle.com/kurtgilby new to kaggle just getting started

[kind field]

Hey fellow Kagglers,

I have shared my journey of becoming notebook master with 5+ years of experience as a Data Scientist. I have worked for Big4 and Fortune 500 company. Follow me if you would like to see more of this content. Also, feel free to ping me if you are starting your journey. Happy to help 😊

https://www.kaggle.com/arshmankhalid

[wild iris]

Hello

[naive mauve]

i know how to use pandas matplotlib and numpy what now? a senior from 3rd year suggested to learn beautiful soup

[quick brook]

Apologies for the length in advance, there's been a lot on my mind.

Hey there, I'm fairly new to data science and machine learning, started about 4 months ago in which I spent the 1st month learning Calculus 1 & 2 and revisiting trigonometry (takes me longer because I'm one of those people who need to know and understand the proof to use anything), I was initially interested in transformers and learned their architecture, just need to at some point understand the self attention layers. The other 3 months were spent just learning linear algebra and the theory and inner workings behind linear regression models, cnns and rnns.

I have an immediate unease due to there being a lot of uncertainty which it seems can only be partly rectified from experience. I can kind of get ahead of myself and think of so many possibilities to extremes (like what if you used found a way to observe matrices visually in models and observed how each layer "agrees" or "disagrees", would be difficult in high dimensional space though).

When it comes to feature selection, reduction and engineering, I'm stuck on deciding where to start because from my perspective it's like having 10 knobs and when you change 1 knob, all other knobs change and your goal is to get the minimum loss 😂 I like to understand the inner workings and why things work the way they do, for example:

With heatmaps, I don't see the point in relying on them if you're using a model and expecting it to make non-linear patterns/connections. It's effectively having linear relationships contribute to your feature selection decision for a model that identifies non-linear patterns via activation functions =.=

I came up with alternative idea though which derives two values: A "Sensitivity" rating for all features, and a "Highest total significance" rating by looking for not only direct significance ratings, but codependent as well up to n elements then aggregating and normalizing the results. Haven't tried it yet though

My end goal is to contribute to research but there's many pathways from here to there, such as quant-related roles, health care, biotech or specializing in one particular area spanning multiple domains like reinforcement learning 😵‍💫

The only way I was able to try and reconcile the feature engineering part (of the 10 dial analogy) was to attempt to automate it, by way of using some kind of pairing of models like classification and random forest models or even diffusion models to automate the creative part and maybe even find some more patterns we're unaware of, while either including backstepping (I view backstepping more or less as brute force to a degree) or a kind of "Feed forward" loop where the initial features are always accessible

[manic lava]

Hello.

Which channel do we need to watch to share the kaggle profile link?

[agile cave]

So by lexicons, you mean textual data. So, you need to find a way to turn words and phrases into numericla vectors. Maybe start with something like the BERT framework

[forest epoch]

Thank you!

[inner snow]

Hey everyone! 👋 I wanted to share that I'm organizing the Data & AI Blogathon. What’s it all about? You’ll be writing blog posts on topics like Data Science, AI, Machine Learning, and more. We’ll have a variety of categories, including topics like Data Science and Data Engineering, as well as different formats like case studies, tutorials, how-to guides, and more. With over 7 categories to choose from, you’ll have plenty of chances to get noticed and win!

What’s in it for you?

• Get featured in big newsletters
• Mentorship from experts in the field
• Connect with top mentors, ambassadors and other AI professionals.
• Get your work shared with over 500,000 followers

If you’re looking to grow your network, get advice, and get your work noticed, this is for you! 👉 Register here: https://forms.gle/FD9FfKJMYp6QCYEE7
Feel free to connect with me on Linkedin as well! https://www.linkedin.com/in/ginacostag/

[kind field]

Apologies for the lengthy answer in advance 😂

Well, I have good and bad news. The good news is I really admire the time you've spent understanding the statistical part of machine learning, which will certainly help you in the long term. It’s clear you have a strong inclination towards thoroughly understanding the mechanics before applying them, which is a rare and valuable trait.

The bad news, or perhaps it's more of a challenge, is that data science and machine learning inherently involve a lot of experimentation and tolerance for ambiguity. It’s great that you’re thinking creatively about feature selection and engineering your idea to derive a "Sensitivity" and "Highest total significance" rating is intriguing and could lead to innovative approaches in model interpretation and refinement.

In terms of where to start with feature engineering, consider simplifying your approach initially. Begin with basic techniques and gradually introduce complexity as you understand the impact of each 'knob'. Automating feature engineering can be powerful, but it also requires a good understanding of what each model and technique brings to the table, which you’ll develop over time with hands on practice.

For your career path, it sounds like you have a broad range of interests. My advice would be to continue exploring different domains and projects to see what truly captures your interest. Whether it’s healthcare, biotech, finance, or another field, each will benefit from your granular level of approach to understanding and applying machine learning techniques.

Lastly, don’t hesitate to engage in collaborative projects or seek mentorship. Networking with peers and more experienced professionals can provide insights and guidance, helping to clarify your path and refine your techniques. Keep pushing the boundaries of what you know and you’ll find your niche!

[kind field]

Hello mate!

[kind field]

Hey Yash, my suggestion would be to stick to one path at a time and master it before moving on to another. Since you already know fundamental packages you are well equipped to start working on real data analysis projects.

[quick brook]

Thank you. I've been looking into decision trees, as well as how to measure significance between variables to attempt my idea, I've noticed some parallels between the cosine similarity method and some of the techniques used, the main difference being that the cosine similarity method has the magnitude normalised to 1 so it doesn't matter if feature/vectors have different sample sizes/elements which could probably be good or bad

[kind field]

This can be beneficial when the size of the data isn’t as relevant as the direction in which the data trends. However, there are scenarios where the magnitude of the variables is important, and in such cases, other methods like Euclidean distance might be more appropriate. For instance, in decision trees, feature selection is inherently performed based on the information gain or Gini impurity decrease, which might align well with your interest in variable significance.

[quick brook]

I've also considered making a map of sorts for both the sensitivity rating and the highest total significance per feature in the method I noted where it is attempted multiple times based on both empty cells being filled with medians, replacing rows or columns with synthetic data within the SD and to have the empty cells simply labelled as "Empty" to be treated as though they're a value in themselves in case it's a deliberate empty cell (forgot the abbreviation for this), it could tell me if for every feature I should remove it, synthesize the feature or fill in the empty parts of the feature with medians.

I'd only go up to n = 3 for large datasets though, would be too expensive otherwise

Or limiting the sample size but keeping the ratio of outliers and SD values in terms of sample selection, unless I somehow find out I want to dilute the effects of outliers through adding synthetic data

[worldly hemlock]

Kaggle keeps telling me that "Notebook successfully pushed to Github" but its not showing on the repo

[agile cave]

Might not be to the repo you are assuming it's pushing to. Just a thought

[modest cypress]

Beautiful! Gina, thanks for sharing!! :-]

[wild iris]

Hey hey!

[uncut geyser]

Our next pitch will envision Explainable Agentic Insurance, which is a novel approach that “programmatically” ensures transparency by providing clear, human-understandable insights into how AI systems make decisions.

Please JOIN the "debate" and the Camelot Network will benefit:

https://www.camelotmarketplace.com/event-details-registration/camelot-insurtech-clinic-13

Best,
Aleš

[lime isle]

https://medium.com/@spamwilliamz/the-pattern-economy-why-ancient-greece-will-shape-modern-day-artificial-intelligence-411b2eb4fddb

[vale pine]

What could be causing my Kaggle progress to remain stagnant?

[static bobcat]

AI Agents: The Game-Changer of 2025

Introducing our latest workshop: Build Your First AI Agent of 2025

📅 Thursday, Jan 16, 9 PM IST

From developers to complete beginners - learn to create AI Agents that work autonomously 24/7—featuring both code and no-code approaches!

Register Now: https://lu.ma/y8pp2fxc

Note: It’s a free but invite only event

[hallow thistle]

We are a very small startup. Our Firefox add-on just got approved. https://addons.mozilla.org/en-US/firefox/addon/second-opinion-with-naval/ ❤️❤️😇 earlier it was just links: Link 1: Any PDF parser https://8ad5-49-36-107-213.ngrok-free.app/ and Link 2: https://7472d5d47ef51a9c81.gradio.live/ CV Deep Dive. https://youtu.be/khGulN2vAyY?feature=shared

[vale pine]

Why is my kaggle progress stuck at 40% ?

I have
17 discussion bronze medals
2 notebook bronze medals

[jovial imp]

Very bad platform kaggle really is. Cannot verify my account and have recieved no help from their contact support

[nova merlin]

No matter how small (u can't call a simple transformer model a startup)

If the extension/app is also just like gradio then it's no more than joke calling it startup actually

There should be a list of options for jobs too

(besides i would say it might be a good small project to learn/sharpen ur skills but u cant call it a startup)

[short wedge]

The ring around each user profile isn't a progress ring. It goes up when you reach the expert tier.

[hallow thistle]

Like recommender engine?

[nova merlin]

In general , if u think more, u will get at the time when jobs re lessening layoffs re increasing, there's no sense of honour/pride any one will full getting the recommended jobs

U need to think from the perspective of any newly graduate person who ve no job exp n willing to get a good job

Or anyone who wanna transit to another job

[hallow thistle]

Very interesting. I have sent you a friend request. Lots of great ideas. https://youtu.be/Fdghi-tdRfs?feature=shared

[echo holly]

hey guys! I just had time now to actually learn the crush course so I wonder if anyone knows if all the course materials are still accessible? Thanks a lot!

[vapid fractal]

Hello! if anyone can help me with geospatial data just DM me. it would be highly appreciated.

[vale pine]

But earlier,
For 1 bronze in discussion it went +2% and +5 or +10% for each notebook bronze medal ,but now it's static.
I can't understand why did is stop progressing all of a sudden

[cedar garden]

hey guys can you help me build a model for my eeg analysis you can find the notebook here - https://www.kaggle.com/code/pramitroy/data-processing

[vale pine]

ayo

[wild iris]

ayo

[quick brook]

awoo 🐺

[kind field]

Nicely done mate!

[wind meteor]

One message removed from a suspended account.

[olive ember]

Hey everyone! I’m currently learning and improving my skills in the field of data science. I’m open to connecting with like-minded people—growing and getting better together would be so valuable. Anyone interested in connecting and growing together? Ping me if interested. Thankyou ^_^

[kind field]

Happy to get connected and collaborate, mate!

[harsh remnant]

Sure bro

[lone wyvern]

Let's start a group

[cerulean crow]

Hi everyone! I’m currently learning and improving my skills in data science. I’m open to connecting with like-minded people.

[onyx birch]

I hope this is the right channel to post about the LLM Research Cohort we are going through in **Cohere For AI **Discord. Join in if y'll find this interesting -
https://x.com/cataluna84/status/1877689686639992872

Mayank Bhaskar (@cataluna84) on X

From the BIRDS(Beginners in Research Driven Studies) organized by @akankshanc of @cohere Open Science Community, we're thrilled to announce our new LLM Cohort! 🎉 🚀

This isn't just another learning program; it's a hands-on, collaborative research initiative designed to push the

X · Jan 10, 2025, 12:11 PM

[mystic sinew]

@fair cliff (i can't ping the moderator role so pinging the one online moderator, hope that's ok) please ban this spammer, rule 7 breach

[short wedge]

Banned the spammer, thanks

[fair cliff]

Thanks @short wedge you beat me to it!

[fair cliff]

Thanks for reporting to us!

[true otter]

can someone help reset my kaggle verification? i was wearing my glasses and it denied me

[kind field]

My Experience: Difference between an average and a good Data Scientist?: https://www.kaggle.com/discussions/general/557761

[soft schooner]

do you guys think within the next few years it will be more beneficial for someone to get a graduate (possibly just a masters) degree over having experience if jobs will slowly stop hiring as many juniors?

[soft schooner]

and as a side note to that, do you think LLMs (or just NLP) should be studied more than other machine learning topics (like shallow learning) within the next few years?

[nova merlin]

GNNs should be studied more enthusiasticly

But not a single paradigm is not less important than the other

Prev concepts forms basis for upcoming ones

[nova merlin]

experience >>>

[agile cave]

Instead of focusing on "A" perfect plan, focus on "Your" perfect plan. Gonna sweat, either way.

[junior scaffold]

hello, is anyone familiar with GANs ? ive got some question for my final year project if anyone could help ✌️

[rare lantern]

So am I. Nice to meet you Eric!

[soft schooner]

oh okay thank you!!

[jovial birch]

HELLO GUYS, IS THERE ANY PLACE WHERE WE CAN PRACTICE SQL (FOR FREE)

[worn crescent]

hi guys

[soft schooner]

@jovial birch try data lemur! it's like leetcode for sql (and some other data science related questions)

[nova merlin]

Codechef too

[jovial birch]

ok thanks, i will try some questions

[main stratus]

Hi everyone!
I hope everyone is practicing ok kaggle regularly!

[digital finch]

I used to practice code-chef a few years ago, now it seems that they made most of their service paid

[neat token]

I am here from now and i am begineer

[zenith sky]

Handwritten Persian numerals - Generative Adversarial Networks: DCGAN, CycleGAN

DCGAN : https://www.kaggle.com/code/omidsakaki1370/handwritten-persian-numerals-dcgans-pytorch
CycleGAN : https://www.kaggle.com/code/omidsakaki1370/handwritten-persian-numerals-cyclegans-pytorch

Website: https://omidsakaki.ir/Projects

[kind field]

Hey David, I have worked on GANs and GMMs extensively throughout my career. Im happy to help if you have questions feel free to ask.

[brisk locust]

Hello guys, is there any platform where we can get Power Bi free Certificate

[flint light]

Hello

[wary oyster]

I am trying to build a ML model for my upcoming hackathon... Which can predict future insights on real time market data and give prsonalised recommendation... How should I do it..!?

[soft schooner]

@wary oyster you don't!!! i tried making a ml model for a hackathon and found out fast (as in last few hours) that if you are making a model where it takes hours to train, you might run out of time just waiting for the model to run. i'd suggest seeing if the hackathon let's you use code from before the hackaton, but just judge on new code. if that's the case then you can pre-train the model and then do everything else during the hackathon

[normal walrus]

Hello everyone

[wary oyster]

Ohhk

Thanks..for the concern

[bitter perch]

Hello Everyone , I am new to kaggle competitions, i was looking at sample notebooks for https://www.kaggle.com/competitions/ai-mathematical-olympiad-progress-prize-2/overview this competition , i see people loading models from input directory , but i dont see any language models on my directory , is there anything i am missing or do i have to train my own model and upload for inference or can load from huggingface?

[surreal plaza]

Love to connect and collaborate

[mild vault]

Let's connect 😄

[serene sage]

i would be much happy to connect

[zenith grotto]

Hello @everyone, I'm new here.

I am a blockchain developer with a passion for innovation and growth. This year, I joined Kaggle to improve my Python skills and further expand my expertise as a full-stack developer.

I sincerely seek your support in this journey.

And I can help you with your job applications by sharing my blockchain knowledge and experience.

Let's work together with an open mind to create a better future for all.

[hexed veldt]

Hello, @zenith grotto
I am a AI developer with 5-year experience.
I have bunch knowledge and experience on developing AI projects.
We can share our knowledge each other.
Contact me if you are interesting

[vague dragon]

Online compiler

[wicked berry]

leetcode and interviewbit are the best
and if u wanna learn in a sequencial manner try sqlbolt.com

[short tide]

Hi I am new here and also kaggle

can someone help me to learn Gen AI and provide me some resources about it

[zenith grotto]

nice to meet you liam1. I would like to discuss more seriously via dm.

[uneven sapphire]

Hey everyone! I'm currently on a path to becoming a data analyst.
I'm hoping to connect and grow with like-minded people.
I would really to like learn from each other, share insights, and maybe even collaborate on projects! If you're interested in growing together, feel free to reach out. Looking forward to connecting!

[muted trail]

General is the new intro channel

[quick brook]

Better than a dead general channel lol

[cold spruce]

[nova merlin]

Hot potato type sayed

[zenith grotto]

Hello. RE you moderator?

[wet tapir]

have anyone tried stupid human-like ensembling of LLM through debates, IDK.. "democracy"? I wonder if this approach can outperf single LLMs on various benches

given input message I

send I to K LLMs

now run a consensus mechanism (e.g. voting), so all LLMs end up with one response

[quick brook]

I've thought this as well, if we can get creative models (such as diffusers) to work on features and prompts or enquiries instead then get a secondary validation model, cool things can happen.

If the above occurred and it did well, and if activation functions were either replaced or fully understood in terms of the degree of non linear patterns identified, I don't think the data scientist role would be much use anymore

[wet tapir]

validation of what?

[quick brook]

Metaphorically, model 1 would keep telling model 2 "I have this idea...", model 2 would quantify how useful it thinks that idea might be

[muted trail]

Why

[quick brook]

Why is "unban" visible? 😆

[hushed socket]

does anyone know about any good AI youtuber?

[wet tapir]

theprimetime if counts

[worldly hemlock]

Is the Phone Verification issue solved yet?

[short wedge]

There are no ongoing phone verification issues (we have thousands verifying every day). If you are having issues with your number please contact support by email so they can help you out.

[worldly hemlock]

I have contacted support many times, no response and also I saw similar complaints dating back months on here

[short wedge]

Some countries have higher failure rates than others and we certainly get people asking about it frequently (we have to manually help dozens of people every day through support too), unfortunately phone verification is less consistent then we'd wish it would be. But such is the cost of fighting endless swarms of bitcoin miners.

Sad to hear you didn't hear back from support, you should definitely check your spam inbox. That is the correct path to go down if your phone number doesn't work.

[worldly hemlock]

It says this the immediate first time I tried to send verification code and has continued saying while not sending any code

[worldly hemlock]

Nothing in spam sadly

[short wedge]

The too many requests error message is misleading (we have an open bug to improve that), but it basically means we can't verify it normally right now. Nigeria is one of the hardest countries sadly, the system rejects a lot. Send another email via this form: https://www.kaggle.com/contact#/account/activate/phone

If you don't hear back after 24 hours dm me with your email address and I'll check with them.

[worldly hemlock]

Will do, Thanks

[quaint creek]

Did anyone use deepseek! I am struggling to load the model !

[glad echo]

Hey everyone! 😊

I'm looking for 2-3 learning partners who are passionate about Machine Learning and AI. Learning alone feels boring sometimes, and I believe that if we learn together, we can improve faster, stay motivated, and reach our goals more effectively.

A little about me:
🔹 I have learned the basics of ML, including supervised & unsupervised learning, feature engineering, and model evaluation.
🔹 I have participated in a few Kaggle competitions but want to improve my skills and rankings.
🔹 Currently, I’m learning Deep Learning and want to explore advanced AI topics together.
🔹 I'm committed to learning and working on projects consistently.

What I’m looking for in a partner:
✅ Someone who is eager to learn and practice ML/Deep Learning
✅ Open to working on Kaggle competitions, projects, and research together
✅ Ready for weekly video/audio meetings to discuss progress and challenges

We'll help each other stay accountable, track our improvements, and push towards our goals. If you're interested, feel free to DM me or reply here! Let’s grow together! 🚀🔥

[sharp sable]

hello everyone!

[crystal scroll]

whats up

[orchid nebula]

Hello guys , Can anyone tell me how I can import data using Spotify API ? I am trying to import data using Clientcredentials call but it's telling me that there's an error in my request.

[rich ridge]

spotify has api? thats pretty cool i didnt know that

[rich ridge]

@orchid nebula u waan post ur header and response? w/o api key of course, maybe just syntax error

[muted trail]

AI search
Andrej Karpathy

[orchid nebula]

This is the error

[rich halo]

___zeitgeist___ has been warned

Reason: Duplicated text

[rich ridge]

@orchid nebula this user seemed to have similar issue and had solve, maybe it will work for u - https://stackoverflow.com/questions/67264163/spotipy-exception-raised-status-403-reason-none

[fringe pollen]

hi

[gloomy meadow]

https://x.com/lichess/status/1886022562959839271

Lichess joined Kaggle! ♟️

lichess.org (@lichess) on X

Lichess is now on @kaggle!

Use our puzzles, openings, and engine evaluation datasets directly in your kaggle notebooks: https://t.co/2Q8qrrprcb ♟️

X · Feb 2, 2025, 12:03 PM

[nova merlin]

When chess.com joining 🤔

[crystal shale]

i created notebook and before even starting it, its saying "Failed to save draft."

is there server problem?

[fleet wigeon]

hellow everyone! I'm new here.

[uncut tulip]

welcome

new here or new to data science as a whole ?

[fleet wigeon]

new here. I have experience with comperter vision.

[uncut tulip]

yeah I am new to this whole thing

crazy math stuff

but I love it

[iron wigeon]

me too

[hot aurora]

Hola buenas tardes me llamo ezequeil

[weary dew]

Hi Guys, I'm new here, I'm math, coding geek

[supple spade]

u eat...kids..

[lilac vault]

New here, too! I am just learning - really no experience - mind is already blown.

[sage jacinth]

Joined just now, super excited

[limber ledge]

hola

[mild jackal]

New way to compress LLMs is out - https://arxiv.org/abs/2501.18596. Code implementation on GPT 2 small https://www.kaggle.com/code/shreeshabhat1004/delta-llm-new-efficient-llm-compression-idea/notebook?scriptVersionId=220459799

[long trench]

3Blue1Brown's Neural Networks playlist is VERY good.

[wet tapir]

SimPO preference optimization method requires a reward model or not? I am kinda stuck at this question

[peak grotto]

hi! konnichiha--

[hybrid topaz]

hi guys,i currently learning python,and sometimes kinda hard for me,and sometimes i dont know,what should I type when I'm in vscode,did u have anything in common when u first started learning ?

[manic lava]

Hi

[crystal scroll]

It can be a little confusing sometimes

Since it's really high-level there's multiple ways you can do the same thing

Once you've got the basics, if you wanna write some new code but you forget the syntax/approach then search it up or use AI-assisted coding

Eventually you'll be able to write everything on your own with no issues

[wicked berry]

Hello, I experienced same when I started learning and still experience it as I try to learn advance topics.
I'd say if you're learning from a book or YT playlist, after each chapter or video, try following it up with a set of practice questions to reinforce what you learned.
Thanks.

[hybrid topaz]

i totally self taught,and i just learn via udemy,and i’m doing this learning for switch career,have u guys advice like,to make my learning process more accurate and efficient,thanks

[slim summit]

hello, anyone know what does the number of leverage statistic represents?

[late saffron]

Hello, just stepping into ML

[formal tendon]

How can I win that 1 million cash prize if your phone verification is preventing me from submitting 🫠 , I already emailed support and it keeps spamming me with automatic reply.

[nova merlin]

figure out other ways to make millions

[short wedge]

Support typically takes 24-48 hours to respond to requests. Make sure you filled out the correct phone verification form.

[pale iris]

hi

Can anyone guide me how to use it/how it works?

[formal tendon]

They already fixed the issue, thanks for the response.

[smoky jasper]

are u talking about kaggle?

[lone wyvern]

Is there a way to do ML without good specs

[formal tendon]

What do you mean by 'without good specs' all modern PC or laptop model are good spec, Im confused. Are you still using a PC from way back early 2000s?

[lone wyvern]

Yeah

I live in a third world country

[formal tendon]

Ahh well that's sounds quite a bit of a problem

basic Machine learning is doable, but your problem would be in like 1 or 2 months you might find playing with those a bit boring

and you'll get an itch to do more complex, that is where you will hit your limitation

[digital finch]

yes :), kaggle offers free 30 hours of GPU every week to its users.

[wary oyster]

@everyone may u suggest me some good Ml course...I am quite friendly with python and various libraries...I want an end to end course...include maths...coding ....projects.... certification...plzz let me know

[zenith sky]

Building Recommender systems with Gaussian Mixture Model (GMM) and KMeans

1- Data preparation
2- Standard Scaling
3- PCA
4- KMeans: Uses hard clustering (each point belongs to exactly one cluster). Computationally faster than GMM.
5- GMM: Uses soft clustering (each point has a probability of belonging to each cluster). More flexible but computationally heavier.

kaggle: https://www.kaggle.com/code/omidsakaki1370/gaussian-mixture-model-gmm-and-kmeans

[ancient shell]

hey, was wondering. Is it possible to use tensorboard in kaggle notebooks?

since i've read this functionality has been added

but has been disabled after some time

due to issues with slowing down some notebooks

[lone wyvern]

Thank you

[lone wyvern]

Oh, well I'll see about that one but thanks

[hallow dirge]

@formal tendon what's your PR bro

I got 160 Kg squat squat in arsenal at 66 kg bw

[tardy osprey]

Anyone know how to submit work on kaggle competition?

[formal tendon]

Mine is not much, 120kg squats, 160 kg dead lift, 120 kg bench and 100kg overhead press at 70 kg

I do calesthenics too

[hallow dirge]

Yeah you beat me at Bench lol I'm at 105 kg and overhead too cause I never did it

[formal tendon]

but you beat me at squats by a huge margin, keep the grind goin brother! fitting data here and getting fit irl lol

[modern radish]

thats crazy

the only workout I do on a regular basis is sled leg press 💀

[red atlas]

Guys Which is/are the best Machine Learning resource(s) for a strong academic and practical foundation? ISLP or Andrew Ng (2018 - YouTube Version) or some other resource? Which one to pick up first?

I am looking forward to build a strong academic/theoretical and practical foundation in ML. So that it would help me out during advanced courses in my masters and also in building projects

Please suggest

[agile ferry]

I bet on Islp!

[red atlas]

Ok thank you so much 🙌

[trail vortex]

Hey, I have data and results that I gathered from 2 datasets but my validation and training graphs don’t look good

I asked ChatGPT and it’s saying my sample size is less

This is wrong because I’ve seen other people use the same 2 datasets and get good accuracy for their model

What can I do

[echo oyster]

https://iap.sundai.club/

[main blade]

I installed tensorflow in vscode(in virtual environment). After this i typed from tensorflow.keras import models (in this code im getting yellow line means it cant find keras) and import import tensorflow as tf (in this line it works perfectly ). I use python 3.10 version. I tried all ways.... If someone has a solution so tell me..

[agile ferry]

r u solving islp? dm me we will solve together!

[agile cave]

WOW! ChatGPT's Superbowl Sunday commercial, from last night. Brilliant. Absolutely brilliant. A creative Masterpiece. https://www.youtube.com/watch?v=kIhb5pEo_j0

[exotic crown]

Hello, can someone explain the scoring system for "Housing Prices Competition" more thoroughly? I don't really understand how I can get such a big score if the formula is sqrt( log(predicted/actual) / n )? Or is the score scaled by some constant factor

[red atlas]

I havent started it as of yet. I will start it but not sure when. Please proceed with it, I will get in touch when I start.

[zenith crater]

Hey there,

is anybody here familiar with object-detection? - In specific: Grid-based Systems like YOLO.
I'm currently building one. So if you'r interested, DM me at any time.

And to be honest I also would appreciate someone to talk about AI stuff and build something together. My whole environment is everything but interested in AI topics...😦

My current projects are the object-detection model and a GAN.

Thank you.

[waxen helm]

Check out this repository for an open-sourced version of the Deep Research App: https://github.com/GitsSaikat/Open-Deep-Research-App

[modern radish]

im actually working on something very similar right now

you can dm me if you want

[wicked berry]

@digital finch

[jovial pelican]

Hey folks, check out this Course from HuggingFace! 🫡
https://www.youtube.com/live/iLVyYDbdSmM

[nova merlin]

@digital finch

[waxen helm]

Would you like to do a literature review in a Fast, Simple, and Reliable way?

Check out this Repo at: https://github.com/GitsSaikat/Deep-Research-Arxiv

Would you like to do a literature review in a Fast, Simple, and Reliable way?

The app is available at: https://huggingface.co/spaces/AlignAI/Deep-Research-Arxiv

[unique ocean]

why i can not choose TPUs in settings?

[opal shard]

You've crossed the usage limit. It's 30h per month as I recall.

[digital finch]

for TPU it is 20 iirc

[tough bison]

Hii

how should we know our kaggle I'd?

[jolly inlet]

step 1: visit kaggle.com and click on your profile icon

step 2: click on "your profile"

step 3: there is your kaggle id

[tough bison]

tqsm

[jolly inlet]

you're welcome

[tardy osprey]

The null hypothesis should be:

𝐻0: 𝜇 ≤ 82

[delicate kraken]

my today's work
share ur thoughts on it!
https://www.kaggle.com/code/huzaifawatto/text-classification-spacy

[wet tapir]

Where to train LLMs?

[young leaf]

yes, and the alternative hypothesis : H1 : 𝜇 > 82

[tardy osprey]

Thanks mate

[dull fractal]

@tardy osprey , Hi Sujal, I have a question. shouldn't null hypothesis H0 : μ = 82 instead of μ ≤ 82?

[tardy osprey]

✅ Correct way to define hypotheses for a one-tailed test:

H₀: 𝜇 ≤ 82 (Null Hypothesis: the mean is 82 or less)

H₁: 𝜇 > 82 (Alternative Hypothesis: the mean is greater than 82)

Why not H₀: μ = 82?

If we set H₀: μ = 82, we would be testing only against that single point, which is not the correct way for a one-tailed test.

Instead, we assume the mean is not greater (i.e., less than or equal to 82) and only reject H₀ if we find strong enough evidence.

[dull fractal]

Thank you

[tardy osprey]

Feel free to ask, it's encouraged me

[dull fractal]

Sure, I am new to Kaggle and right now doing AI/ML course online. Just started up with titanic project.

[tardy osprey]

Cool

[dull fractal]

R u working on any ML project

[tribal vector]

Hello all! I am trying to find team( or people)...but is hard...I have no one to share.

[forest whale]

I am new to this. Can someone help me get started?

hello room

[forest whale]

I am trying to start one as well. Have you figured it out yet?

[gusty cloud]

hii, i am an ML enthusiast and I really want to know how actually do we participate in Kaggle comps, like I have made various time series forecasting models using SARIMA and ARIMA, etc, so how does we actually get to win kaggle comps like Backpack price prediction, pls help

[daring fern]

hello @everyone i hope you're doing well 😊
i'm starting my learning path in agent creation with langgraph
is there any recommended courses? thanks !

[wicked berry]

Hi, I have completed a BSc in Data Science and wish to pursue a master's in the same field. However, finding good colleges has been challenging since my current college does not have a strong curriculum. Can anyone suggest some good colleges in India for a master's in Data Science?

[nova merlin]

@gray bough

[agile cave]

challenge accepted

[long shore]

Hi This is Harshit a 2nd year students from India. Learning ML , Can anyone help what is titanic machine learning completion is.

[mint stump]

Hi. Im new here so sorry if this isn't the right place to ask but...
How are datasets curated for Causal or Seq2Seq LLMs to be fine-tuned using SFT to perfrom with ReAct CoT?
Is the dataset simply mapping initial input with the ideal chain output? e.g

input: tell me the number of sunny days in the past year output: Thought: ... \n Action: ... \n Observation: ... \n Output: ...
Or do i have to map step-to-step mapping? e.g
input: tell me the number of sunny days in the past year, system: Thought ....\n output: Action: ...
input: tell me the number of sunny days in the past year, system: Thought ....\n Action: ... \n output: Observation: ...

I hope I am understandable.

[sterile granite]

Curious, creative, and always evolving—I turn challenges into opportunities and value authentic connections.

[drifting trout]

Hello I'm thamidu from Sri Lanka. I'm a student majoring in AI&ML and I'm in my 1st year of university.

[rugged stream]

Hi I'm Hezron a 2nd year student from Kenya. Taking Computer Science

[nova merlin]

Waiting for it 😔🔥

[shut panther]

happy for feedback enjoy https://systemsandlife.medium.com/smashing-the-tariffs-for-fun-and-profit-how-deepseek-v3-outsmarted-the-ai-ban-ace2a8314ad9

[tender patio]

Hey anyone here published any papers in ML/AI field can you guys just Dm me?......i just need some tips to start with

[frank obsidian]

Hi, I am from Australia last year student studying Business and Data Analytics

[tender patio]

Hello

[silver parcel]

Hello!

[tardy osprey]

Nope right now but in future will definitely do it before working on ml i should learn stats

[pliant hearth]

hi

[brazen smelt]

Hello

[sturdy canopy]

Hello!

[lilac abyss]

Hi
I am ganesh from india
I am 14 years old.
I am here to increase my chances to get into a good university.
although i am particularly passionate about my career in ai nice to meet you all

I am a begginer with zero knowledge can you guys give me a basic view of it

[indigo dock]

What language are you knowledgeable about?

[lilac abyss]

My native language is not English
That is why there may be some mistakes in my English

[indigo dock]

I was talking about computer languages btw

[lilac abyss]

Oh I answered the same ques
English is the new Computer language
_Jensen Huang

[lilac abyss]

Was just joking I learnt c++ in childhood which I don't even remember

[indigo dock]

Revise it already. Although C++ isn't that important, I would recommend you learn Python.

[bold phoenix]

🚀 Looking for a Team for Upcoming Hackathons in Delhi! 🚀

Hey everyone! I'm looking to build a team of passionate students for 2-3 upcoming hackathons. If you're a Machine Learning enthusiast or a Full Stack Web Developer in Delhi, I’d love to have you on board!

✨ What we need:

• ML Enthusiasts who are comfortable with tools like TensorFlow, PyTorch, or any other ML frameworks.
• Full Stack Web Developers with experience in React, Node.js, MongoDB, or similar tech stacks.
• A great attitude and willingness to collaborate and brainstorm with a team!

If you're interested and ready to take on some exciting challenges, please DM me or drop a message here! Let’s create something amazing together.

Looking forward to hearing from you! ✨

[main python]

💼 Exciting Business Opportunity – Seeking a U.S. Partner! 🇺🇸

I’m looking for a reliable U.S. citizen to collaborate on a new business project.
No tech skills required—just a willingness to work together.
This is a long-term opportunity with monthly payments and a chance to grow.
D-M me for details if you’re serious about exploring this opportunity! 🤝

[hallow pewter]

Am I the only one facing this issue? or is it that pressing TAB to auto complete code lines doesn't work anymore? for example I type torch.nn. and press TAB, it doesn't show options for auto completing? did something change?

[lilac abyss]

Is it necessary when artificial intelligence can write our code

[indigo dock]

🤦🏻‍♂️

[smoky jasper]

what

dude what r u saying xd

[smoky jasper]

it's working fine...
What r u using?
I am using jupyter notebooks inside vscode

[lilac abyss]

By the time we get to work it will

[lilac abyss]

My dumb brain couldn't understand the reason for your frustration can you explain

[indigo dock]

Bud, why are you here? Tbh, AI can do pretty much everything, but to create an AI yourself, you need to learn coding. Jeez, it's simple, bruh. I'm not here to teach ya, so you are free to learn ML using English. 🤷🏻‍♂️

[smoky jasper]

AI will automate things
But then there will be new technologies that AI can't code and learn as soon as a human can.

AI is more like an assistant. And people learn to code from scratch just because they are tested in the interviews and sometimes the basic knowledge of human can create some specific solutions where AI models might lag.

And I am not criticizing you. But it won't workout arguing with people like this, because people here know ML and English, both, properly.

If u are that confident about it, you can try some kaggle competitions, with 'English'. 😂

[smoky jasper]

I'm with u my friend

[indigo dock]

Well said

[lilac abyss]

Hmm if I feel like Rude or arguing I am sorry .
Anyhow thank you for clarification

[smoky jasper]

Yeah it's good that u realise
You can search some yt videos to get the roadmap, and then u can start ur learning
And if u have any doubt u can ask here

[lilac abyss]

Oh definitely

[signal remnant]

hello anyone know how to finetune a quantized model? like deepseek r1 distill qwen 2.5 awq ??? Would you guys advice of it? or would it be better to just finetune the original model then distill it???

[arctic burrow]

Hi, anyone having issues accessing the Kaggle website now. I keep getting 404 not found. Just wanted to check whether it's a problem only for me or for everyone.

[lilac abyss]

I am getting the access

[indigo dock]

It's working fine for me. Try clearing the cache of your browser.

[arctic burrow]

Gotcha, problem with my wifi. Thanks for the response

[indigo dock]

Gud for ya, mate.

[main python]

💼 Exciting Business Opportunity – Seeking a U.S. Partner! 🇺🇸

I’m looking for a reliable U.S. citizen to collaborate on a new business project.
No tech skills required—just a willingness to work together.
This is a long-term opportunity with monthly payments and a chance to grow.
D-M me for details if you’re serious about exploring this opportunity! 🤝

[main blade]

(tnv) PS C:\Users\DELL\Desktop\ht_pd> python app.py
Traceback (most recent call last):
File "C:\Users\DELL\Desktop\ht_pd\tnv\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 70, in <module>
from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal: A dynamic link library (DLL) initialization routine faile
d.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
File "C:\Users\DELL\Desktop\ht_pd\app.py", line 5, in <module>
import tensorflow as tf
File "C:\Users\DELL\Desktop\ht_pd\tnv\lib\site-packages\tensorflow_init_.py", line 40, in <module>
from tensorflow.python import pywrap_tensorflow as _pywrap_tensorflow # pylint: disable=unused-import
File "C:\Users\DELL\Desktop\ht_pd\tnv\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 85, in <module>
raise ImportError(
ImportError: Traceback (most recent call last):
File "C:\Users\DELL\Desktop\ht_pd\tnv\lib\site-packages\tensorflow\python\pywrap_tensorflow.py", line 70, in <module>
from tensorflow.python._pywrap_tensorflow_internal import *
ImportError: DLL load failed while importing _pywrap_tensorflow_internal: A dynamic link library (DLL) initialization routine faile
d.

Failed to load the native TensorFlow runtime.
See https://www.tensorflow.org/install/errors for some common causes and solutions.
If you need help, create an issue at https://github.com/tensorflow/tensorflow/issues and include the entire stack trace above this
error message.

what this erorr massage try to say? i already installd Microsoft Visuals c++ 2015-2022 Redistributable (x64)

this error occurs when i try to import only tensorflow.

[lilac abyss]

What do u guys think about optical gpus

[hidden cairn]

HI guys

[opal shard]

The GPU cannot be optical at this stage, it's the path where the data travels. It only increases the bandwidth not the performance.

[opal shard]

There is no fully optical GPU yet. Those are only hybrid systems. Such as the Nvidia NVlink, which still uses electronics to operate and perform the main task.

[grizzled jasper]

Hi Guys!

[misty crystal]

hey guys i got an issue when writting my notebook which is each time a try to run something i got spammed by <class 'list'> and i dont know why
this is the code im running if any one know i fix i will be very happy if u help me

[hallow pewter]

I'm using kaggle's inbuilt notebook, am not talking about jupyter or vscode

[odd breach]

It worked for me too in the first few coding parts in Python course, but since then it haven't worked at all. I've been doing python notebooks in google colab since, because the autocompletion doesn't work in kaggle notebooks

[tardy osprey]

*I’m working on a data project analyzing NYC taxi rides to see how payment methods affect revenue. One surprise? I found 8-passenger rides making up about 7% of my data!

Since I’m not sure what type of vehicles are in service, I decided to combine 5+ passengers into a single category. This keeps the data (instead of dropping 7%!) while simplifying analysis.

My question to the community: Is this the right approach? Or would you handle these outliers differently? Let me know your thoughts and any best practices you follow in such scenarios!

#DataScience #Analytics #EDA #DataCleaning #NYCTaxiData #MachineLearning*

[barren hinge]

I would make sure to explain the thought process clearly because it makes sense to me. If the average vehicle has five seats, with one occupied by the driver, then labeling anything larger as "5+" would be reasonable—assuming "5+" isn't significantly larger than the others in the group. Then you might want to break it up into 5-7 and 8+.

[tardy osprey]

Ok so will make groups / bins

[meager thistle]

Hi guys, I am applying for internship and need to write SOP. If you have written then please help me with the format and what i need to write. Its my first intenship and i don't know what to write.

[compact hamlet]

try chatgpt

[trail vortex]

If I don't use an accelerator to run my Kaggle notebook, will the runtime be unlimited or will it stop at a certain time

[rose dawn]

did kaggle just launch GPU L4?

[agile cave]

So you are trying to categorize. Perhaps you use a genetic algorithm to come up with bins, in which to place data. Shouldn't be to hard.It's a start. This is an unsupervised learning task.

[agile cave]

First, make it Truthful. Then Relevant. Take what you have and in the final pass, write that up but make it Compelling Sounding. Get creative here but keep first two steps in mind.

good luck on the internship

[manic lava]

Hello

[frail wind]

Do y'll have any idea how to get started making money quickly, which specialization to choose, get a job or something else and once I've acquired the skill/specialization what should be the next step.

[tardy osprey]

Naah they are in scientific form doesn't contribute so much so will remove it

[agile cave]

You are familiar with Unsupervised Learning, right? You spoke of finding a Category. So This method would provide you a way to create Categories of Data, for processing, further down the line, say as inputs to a ANN. By what I can gleam of the problem you are tackling, by what you said. Not quite sure what is the detailed problem statement

Anyway, good luck

[tardy osprey]

It's contribution value is something like that 0.000000067

[agile cave]

So why did you bring it up?

Besides, categorization would create bins to store a range of values and orgaize meaniful organizations of data. Anyway, you've made up your mind. Experiment and see.

[tardy osprey]

Didn't know about it previously now i understand

[rose dawn]

Just a generic questions, If I am writing my dependencies here and let's say this is my submission notebook. So do I need to turn on the internet? If I am turning on Internet, then I can't submit right? Because competition rules states that

[calm zephyr]

hello

[true epoch]

In every ML project or problem tree model always outperform basic ML algorithm so what is use cases of linear regression and logistic regression, svm, naive bayes

[opal shard]

You must not know there use cases properly 😂.

[opal shard]

Simply save your submission as a CSV and submit in the competition. Why to think in the first place 🤔

[rose dawn]

no no, submission notebooks are basically like scaffolds, you write your code. Then during actual submission, kaggle backend populate the directories or CSV with their actual test cases.

[stoic barn]

Object { message: "The service is currently unavailable.", stack: "xV@https://www.gstatic.com/_/mss/boq-rsvp/_/js/k=boq-rsvp.Rsvpguest.en_US._Dexu3WWss4.es5.O/am=gBE/d=1/excm=base/ed=1/dg=0/wt=2/ujg=1/rs=ANfW6yoIyJMa29mjGYDpYxNwTfPDfsqLYQ/m=base:2628:292\nPk.prototype.onError@https://www.gstatic.com/_/mss/boq-rsvp/_/js/k=boq-rsvp.Rsvpguest.en_US._Dexu3WWss4.es5.O/am=gBE/d=1/excm=base/ed=1/dg=0/wt=2/ujg=1/rs=ANfW6yoIyJMa29mjGYDpYxNwTfPDfsqLYQ/m=base:2628:696\n_.IL/<@https://www.gstatic.com/_/mss/boq-rsvp/_/js/k=boq-rsvp.Rsvpguest.en_US._Dexu3WWss4.es5.O/am=gBE/d=1/excm=base/ed=1/dg=0/wt=2/ujg=1/rs=ANfW6yoIyJMa29mjGYDpYxNwTfPDfsqLYQ/m=base:1853:634\n", code: 503, status: "UNAVAILABLE", details: undefined, name: "RsvpApiError" }
m=base:2628:720
@coral hedge
is something broken?

[floral anchor]

I am trying to sign up for the 5-Day Generative AI course on Kaggle, and whenever I submit registration, it always sends me back to the registration form (as in it is having trouble processing the registration). Any help on that? Thanks.

[stoic barn]

Yes the above message is the same

[pale yew]

same thing happened with me

[spiral spire]

same

[jovial pelican]

@coral hedge sorry, I can't use the register form:

other users are reporting the same issue with this form: https://rsvp.withgoogle.com/events/google-generative-ai-intensive_2025q1/forms/registration

[long meteor]

same

[spiral spire]

my friend got in but not sure how

[dim latch]

Same +1

[silk iron]

same +2

[coral hedge]

Hey! We've seen that some of you are experiencing issues with registration. We're actively working on resolving it and will update you here once it's fixed. Thank you for your patience! 🙏

[ember urchin]

same

[jovial pelican]

thanks to you Kinjal and your amazing team! ❤️

[quartz hemlock]

Apologies if someone asked already but does anyone know if the 5 day generative AI course will feature the same content as the previous one?

[feral saffron]

@coral hedge I am facing the same problem regarding register form for "5-Day Gen AI Intensive Course with Google". Not able to register. Form returns back to the same page.

[feral saffron]

https://tenor.com/view/hacker-mainframe-cloudy-with-a-chance-of-meatballs-flint-gif-7952417

[vale ermine]

I have registered previous 5 day gen AI but couldn't attended ... should I select yes in the option or no ? @coral hedge

[coral hedge]

Hey Satyam! This is an optional question. Please feel free to select "No" if you didn’t attend the event last year. Thank you!

[vale ermine]

thank you 🫡

[coral hedge]

Hey! The registration issue has now been resolved. Please try registering again, and let us know if you continue to experience any problems. Thank you for your patience and apologies for the inconvenience caused! 🙏

[wet tapir]

in most competetions winners used ensembling?

[mint yoke]

Hey, i need help

We have long conveyor like 5 kms long
In which there are idlers around 4 to 5k
We divided conveyor section with imaginary line let's say every 20m , in that there are around 10-12 idlers of same dimensions.

We have normal data that is after replacement of idlers, and have abnormal data that is before replacement , we have dataset in the form of real positive fft that is each row contains list with 5k integers

If i train 1dcnn based auto encoder or vae it works section wise, like I can see higher reconstruction error in abnormal data. But it is impossible to create model for every section it will be computationally very expensive. I want single model that will work entire conveyor, but when I combine all data and train then it won't generalise well.

Also I tried extracting statistical features like kurtosis , skewness etc and trained dense vae but no luck what can I do ?

Note: i can see abnormality in normal data too. Even after cleaning it becomes more sensitive to normal data as well tell me better approach if you have any experience related to similer problem

[solid sage]

Hello ML martyrs. Glad to be here.

[rigid mauve]

Hello Everyone,

I’m working on a project involving RNA tertiary structure prediction and need some expert input on evaluation metrics (e.g., recovery, F1 score, SC-TM). If you have knowledge in biology, especially RNA folding, please DM me. I’d really appreciate your insights!

Thanks!

[wicked berry]

Hello Everyone,

My Name is Swoyam, a final-year engineering student, currently interning at Sarvam AI as a Forward Deployed Software Engineeri and kaggling in my free time

Looking forward to great connections

[thorny robin]

Kaggle

[rich halo]

usif0516 has been warned

Reason: Bad word usage

[agile cave]

working on a next gen Trading Bot (project name: $ense)

75% done on finalizing code for POC. Today i dealt with coming up with curating training data:

it's also simultaneously an AGI project

final goal?

[rich halo]

spindoctor3551 has been warned

Reason: Bad word usage

[agile cave]

what?

seriously?

prove it

Have you ever considered processing BERT vectors, one by one, through a NN, with different input layers (in the Model) for each vector , based on its Part of Speech. Of course, ignoring intermediate outputs of ANN (while processing each Vector independently/ sequentially), and basing error correction on the final output. In cases of ambiguity of Parts of Speech, several Input layers may receive the same Vector, but use of LSTM in hidden layers should emergently build the ability to disambiguate. Take the Word "bank": a place to place your money, a geographical location, a hope, a pool shot, etc.

[agile cave]

Do we always want AI to come up with a "Mythological" perfect solution? What about tools that help a Student progress and learn based off the Evolution of their own Intellect, Ideas, and Emerging Passions and Discoveries? A tool that helps Human Intelligence to emerge, and achieve its organic "Best". https://www.youtube.com/watch?v=Z9A-acu-0y0

AI is just a tool. Like a Hammer, or a Saw, or a Printing press. Nothing more, nothing less. IMHO, it will allow one to organize one's thoughts, rather then generate them. It will allow Us to navigate our own inherent Brilliance. But it will never be able to be Brilliant alone. This is the Prime distinction between Human and Artificial Intelligence.

code on

[thorny robin]

hmm

[visual summit]

Hello, I am having a hard time learning python at the moment. Is there a quick and simple project on kaggle that could put things in context for me? I have an understanding of some basics but doing exercises without putting them in context relative to an analysis is not helping. Lists for example, it would be a good idea to do something basic that would show the use of a list in an analysis. I think that could help me personally progress much quicker.

[sullen gale]

I have created a repository that has a few Machine Learning algorithms implemented from scratch. Here's the link: https://github.com/Munfa/ML-from-scratch
I'd love to hear your advice and suggestions. And can someone please tell me what project I can try to start building up my skills for ML?

[sullen gale]

I am just a beginner but I think we learn the best through practice. maybe try doing basic projects and try to know how every method works

[agile cave]

haha

[agile cave]

Try working with MINST dataset and problem? It's a classic AI "Hello World", of sorts

A classification problem. Than move to regression, Deep Learning, and then GAN's

[sullen gale]

thank you so much

[agile cave]

@sullen gale mess around with lots of simple problems. Get a FEEL for what AI Solutions actually do

[gloomy zinc]

Hey everyone, I am Rashid and recently I started giving interviews, my background is I have learned machine/deep learning and TensorFlow from the specialization on Coursera, and a lot of YouTube videos, now in the market every job description includes fast api, flask api, pytorch, docker, Aws, Google vertex ai and along with it generative ai like llm, llm fine tuning, RAG, chatbots etc. Each organization ask these questions along with experience which I don't have much, and now I have FOMO or imposter syndrome how will I learn all of these in a short time, for my portfolio, I want to create projects so that I can show employers to get a job without any experience. I need advice from an industry expert on what should I do and how to do it.

[sweet copper]

@chrome rover hey i sent you a message in your private, sorry i cant send here

[pine tartan]

Hello, I am Adam. I am 14 years old and am in grade 9. I know python and am learning c++, HTML, CSS, and Javascript. I completed the intro to ML course on kaggle and the pandas one as well, but I am finding it hard to implement as it isn't sticking with me, I learn best with videos. Does anyone know of some easy to follow, up to date, and in depth videos on ML that also have mini assignments for you to type the code you learn and solve mini puzzles? Thank you.

[digital finch]

is it normal for medal certificate to not open on the first day?

[tardy osprey]

Need a partner, currently learning sql

[jaunty jacinth]

hi

[main blade]

hello friends... my name is Tark(Data Science) i created my portfolio websiteCan you tell me if there is any mistake or not? I glad to hear that
https://tarkptel.github.io/

[quiet cloak]

Hello Kaggler

I recently learning about machine learning. I post my learning material in github
https://github.com/sadiq18

[iron cave]

.

[fossil olive]

hi, i've good knowledge of robotics and embedded systems.. to upskill myself i wants to learn Computer vision in depths of it

so, i would love to learn together with all of you.

[acoustic folio]

Hello Good morning

[rough folio]

hehe

[wanton sable]

Good morning. Is it just me or is kaggle down? (404 error)

[robust nacelle]

Hello everyone! First of all, please excuse my English (obviously it's worse). I'm a beginner in Data Analyst, and I would like some tips from Junior, Mid-level and Senior professionals. Please, this will greatly increase my knowledge.

[agile ferry]

Hlo everyone!!
https://www.youtube.com/watch?v=kCc8FmEb1nY

anyone here who watch this video and want to explain me this video!!

Damn this server has no voice room!!

[candid gyro]

Hello, I'm a kaggle novice. Can you tell me how to study?

[torpid scroll]

Hi I'm an information engineering student interested in neuroscience. I'd love to end up working in a multidisciplinary team that does brain research. I am trying to develop my skills so one day I can do it! If you have any suggestions on how I can do this, I'll welcome the advice 🙂

[jaunty spoke]

I would love to hear if you receive any responses. I am in a similar situation.

[junior haven]

hello i'm starting to learn data science and i thing kaggle will help me a lot with all their practice project

[orchid cave]

Hello, i am new student on Kaggle

[torpid shard]

Hello, i am new student on Kaggle too

[knotty narwhal]

hello all

[agile ferry]

How can i find summer internship?

[empty smelt]

Hi all! I'm new to the Kaggle world and excited to join the community

[jolly spire]

Hi! I'm starting to study Python and Machine Learning. I hope Kaggle will help me on my journey.

[wary acorn]

Hello. started exploding kaggle as newbie. I will need support from all to grow and hope to enjoy a lot.

[main blade]

did you joined any kaggle compitition ?

[hazy pollen]

Hello all! I'm new to the Kaggle.

[tardy osprey]

Can anyone explain me, how to identify that which models suits between additive and multiplicative decomposition through visuals?

[vestal spire]

Hello all! I'm new to the Kaggle.

[sinful junco]

Hey wellkom

[copper eagle]

Any help here 🥹!.... Trying to run an ARM64 version of Microsoft Server 2025 on a raspberry pi 5 🙏🙏

Do you all know how to use UUP dump?
I'm seriously bleeding! 🫥🐸

[green glen]

Hello Kagglers, I'm Sam Looking forward to collaborating and learning with you all.

[rancid wharf]

hello

[rancid tundra]

I’m currently working on a project analyzing the diabetes dataset using Linear Regression. Would love to connect with others on a similar journey!

[rose salmon]

Hi, I'm Chris and hope to compete in the competitions as a test of my knowledge level. Good luck to all

[wraith vault]

hello

[opaque owl]

Hello!!

[icy pasture]

Hello, I am B.Tech Computer Science student and I am passionate for building impactful AI-driven and tech-based solutions to real-world problems.

[modern schooner]

Hello everyone, im a software engineer, and currently im facing a big issue my home box,my computers and all my numeric stuff have been hacked by some malicious persons,need an help asap,plz.

[shrewd schooner]

hiii guys

[nova merlin]

Hire an (ethical) hacker fr

[unreal crane]

guys i need a gform to be filled for my data science project, so shall i post it here ?

it js has very basic qs regarding your career path

[crystal shale]

https://www.kaggle.com/code/nikolasgegenava/fraud-detection-98-accuracy?scriptVersionId=227765389

Hello! Interesting solution to imbalanced dataset problem, check it out and give upvote if you don't mind 🙂

[unreal crane]

https://forms.gle/jeuXF3cLisoUCxTt5

guys pls fill out this form for my survey, js 4 qs, bless me with this project

[nova merlin]

U can better create/get synthetic data from ai

[unreal crane]

yea but its for my internship so i prefer it to be as authentic as possible

i don wan blow my only chance of getting a stipend internship lol

[unreal crane]

it wld be great if yu filled it up man

[zenith grotto]

Seeking a new colleague
As we grow our presence in the European and American markets, we’re looking for a dedicated and professional representative to join us.

Your Role:
Serve as our point of contact for client meetings and communications.
Maintain clear and professional communication with potential partners.
What We’re Looking For:
Fluent English proficiency.
Availability during European or American business hours.
Technical or programming experience

What We Offer:
A revenue-sharing model, with competitive income.
A flexible role with minimal time commitment.

If you're interested, please reach out via:
📩 Email: [e-mail]
🌐 Skype: @bghost131
💬 Telegram: @bghost131

Let’s collaborate and build something great together!

[split lily]

Calling all researchers! A friend of mine and I, currently university students, are working on creating a product to aid researchers save time on common tasks. If you could spare about 5 minutes to complete this survey, https://forms.gle/9ggnDL8qk5U4YTsLA, you would help us immensely and we would be one more step closer to creating something you could use in the future.

Thank you so much!

[dire sorrel]

hello

[warm flax]

Anyone else facing the same issue ?

nvm just an isuue from my side

[eternal vigil]

Hello 👋, I am am aspiring data analyst. Any Data analyst in the house please

[agile ferry]

Guys I'm lookin for summer internship can u plz share some opportunity in data science field!!

[zinc jacinth]

Kaggle just got me a job at ABB

[nova merlin]

#announcements message anyone can pls tell how's this small we can quantise it to max 8b from 24b how does it seems small?

[short wedge]

Congratulations! If you don't mind me asking, how did Kaggle come up in the interview process?

[zinc jacinth]

Hey! I have a little section in my resume that's dedicated to projects:

The hyperlinks link to the Kaggle notebooks and I guess the manager was interested, he looked at my profile, and saw a couple competitions... He thought it was very interesting :3

It was brought up along the lines of him asking me about the projects and the competitions

[short wedge]

Very cool! Glad to hear that your Kaggle projects helped seal the deal. Congrats on the new job!

[zinc jacinth]

Thanks, Myles :3

[agile ferry]

Congrats Brother!!

[warm verge]

Hello

[rancid rain]

hii

[toxic river]

I just completed python basics,powerbi and sql and going to start DA and ML from next week
How to have some project experience?

[north grove]

Greetings ✨

[rose dawn]

Why kaggle is counting failed submissions into number of submission?

This is so frustrating

[wicked berry]

Hi! i am new with the data analytics and want to practise as i have gone throught the libraires like numpy pandas matplotlib seaborn -> but practise is what i lack so could anyone tell me which kaggle dataset or code i can use or what to do in this situation...

[heady tinsel]

Hello!

I'm new too

[narrow locust]

Hi

[icy prawn]

Hi

[grizzled gulch]

Since you're learning the basics just focus on picking any dataset that's tabular. There's a ton of them. Gym Goer data set is synthetic and I know has a guaranteed solution if you find it. That's a decent one to start out on.

[wicked berry]

thanks!

[shell berry]

Hi guys, how do I go about creating a new dataset? I am giving birdCLEF a shot this year but after I process the data in the kaggle notebook how do I retain it and possibly transfer it, so I can use it in a different notebook?

[neat light]

hi

[rough kelp]

Hey!!
Why my code in kaggle notebook is not using gpu as can also be seen in the top right sessions list of cpu, gpu, etc. But the quota is considered as if its being used.

Kindly help me out with this problem that my code is not at all using gpu. It is indeed gpu intensive work.

[formal tendon]

what's the highest possible ceiling for titanic dataset without cheating?

[grizzled gulch]

That's always an open ended question for any data set. You can't know that since optimization is a NP-Hard problem. You can only know what's the best people have gotten without gaming the system.

The only models you can guarentee it is say linear models with a sum of squares loss function.

That said I see on a lot of leader boards for multiple competitions people getting Acc = 1.0 which in my experience is complete BS for any realistic data set. Because that means you overfit and are likely even fitting the errors in the dataset.

[frail wind]

Hey

How do you attach a dataset to the notebook

[formal tendon]

'people getting Acc = 1.0' yeah it's BS, since like Im hard capping at around 85ish so I was thinking how far people can push this, havent tried any ensemble methods tho

[minor swift]

There is a input option on the right when you open the code(kernel)

[digital finch]

to know the maximal possible accuracy, you need to know the true underlying distribution of the data, that is the theoretically most accurate modelling of the entire environment, any more accuracy means either it is due to random chance or you have fit to the noise.

for real life data, the true distribution is always unknown

we try to approximate it using a variety of different methods but even they(depending on what approximator we use )do not reach their optimal state.

[nova merlin]

💯

[formal tendon]

Yeah, taking in consideration with all of that, What I just actually want to know is like the top 10% results so far without any form of cheating

[haughty trail]

hi guys

i am new

[rich halo]

ash_blanc has been warned

Reason: Bad word usage

[nova merlin]

I just said everyone's new when they begin

[rich halo]

hlly3499 has been warned

Reason: Bad word usage

[cyan kraken]

Hi guys, I am new here

[ionic vector]

Hakuna matata ML party!

[hushed mica]

Here. to deep dive in the world of Machine Learning

[tame juniper]

Hellloo people, I'm a newbie in this space so looking forward for some guidance

[urban salmon]

hi everyone!!

[tribal sable]

Hi nitya nice to meet you

[raw galleon]

Same here

[median summit]

hi

[fluid pond]

Hey everyone, happy to be here! Excited to learn and share cool data insights with you all.

[quiet scroll]

Hi Himanshu this side, Happy to be here as a Newbie

[frail wind]

Hello

What are some common freelance or contract opportunities in this field?
Can you help me understand potential earning paths.

Hello

[delicate temple]

hello, happy to be here.

[opal shard]

How will you run the Gemma 3 on multiple GPUs on a notebook environment, what challenges will you face & how will you combat them? Challenge code: Dust3018
The first one who will answer successfully will get an additional problem privately.

[nova merlin]

isnt gemma 3 made primarily to be run on a single gpu effeciently ?

[ancient nimbus]

Hi

[mystic nova]

hey

[umbral cloud]

hi

[fierce spear]

hello everyone, a new machine learner

[frosty pine]

Hi everyone, I'm somewhat familiar with research programming and machine learning but somewhat daunted by the level of attention to detail required to compete on Kaggle. Would love just to hang out and figure things out.

[nova merlin]

What's research programming btw 🤔

[frosty pine]

I mean I've worked on a few research projects and done data cleaning/data analysis/code documentation. Also trying to compare statistics and ML approaches and write about it for my dissertation. It's not a special name or anything 😅

[nova merlin]

But featured kaggle comps bit different from them

U pick sound papers or sound notebooks go through it test it if it works on sample dataset u implement it n keep refining it

[frosty pine]

I understand that makes sense. Pretty much where I've been at.

[cerulean trellis]

how'd y'all suggest this book to brush up some concepts and learn stats for ML
https://probabilitycourse.com/

[frank cove]

hi!

[wispy cosmos]

Hi

[small stag]

hii guys my self shibi a freelancer this is my website: shibnovation.me

[vapid shale]

Hello everyone, Amr is here. A freshman SWE student at Al Yamamah University.
My Kaggle URL (https://www.kaggle.com/amrshaar)
Thx & good luck for you all.

[opal shard]

What's the spelling of your name?

[vapid shale]

@opal shard
Just Amer
it should be an easy name to spell...
I hope 😅

[jovial imp]

my kaggle got struck after i leave it for some time and cells run forever

what should i do @everyone

[vernal jacinth]

hello i don't know if this is a dumb question but should i learn decision tree with the ydf function or from scikit?

[violet fog]

Hi!

[copper sapphire]

I have earned the Agent of Discord Badge and the Bookmarker Badge!

[marsh oriole]

Hi!

[tawdry copper]

Hi All, I am looking for Kaggle Grandmasters here who can mentor me. I am willing to pay for mentorship. Thank you!

[lone violet]

@tawdry copper it's great to see your enthusiasm for learning! While I may not be a Grandmaster myself, I can definitely guide you in the right direction.

[nova merlin]

Lol those are no degrees

[nova merlin]

U don't need to pay for that btw 😅

[copper sapphire]

🤓

[tawdry copper]

Yes I am aware. I am just trying to convey how badly I want to learn 🙂

[tawdry copper]

Thank you! I will DM you

[lone violet]

Sure anytime, you welcome

[lone violet]

True 😀

[nova merlin]

But still yeah it's a good thing 👍

[mighty yew]

Hi Guys! I am mostly interested in time series.

[agile cave]

Consider researching Astronomy

[digital finch]

this one? It has a decent IMDB rating

thanks for your suggestion, will try it out 👍

[nova merlin]

U r a respectful ai-er

[digital finch]

no I am an idiot

[nova merlin]

Anyways r u joining arc ?

[digital finch]

thinking about it

[nova merlin]

What r ur thoughts on vlm for it anyways ?

[digital finch]

im not really sure tbh

intuitively it shall do better

but it might make it worse I have a feeling

but I might be wrong

text is much more descriptive than images

🙏

[nova merlin]

Yes like emergent behaviour of vlms

[digital finch]

how is that relevant?

[digital finch]

there is not much information in images than text rather than images

[digital finch]

just closer pixels attend more to each other

that behavior can also emerge in unimodal llms

if well trained

so characters right below each other

like in ascii art

[nova merlin]

But still vlms should be given a try

[digital finch]

ofc

they might do better, because they are newer models

[nova merlin]

Besides the actual experienced arc guys believe llms even the reasoning ones ve no chances

[digital finch]

also there might be slight advantages as well

[digital finch]

the open source one?

[nova merlin]

Hopefully 🤧

[digital finch]

I will try gemma 3 ig

[nova merlin]

Yes any actually

[digital finch]

even the biggest fits in 4xL4

[nova merlin]

I will try Gemma3 n codegemma 🤧

[digital finch]

codegemma?

ok nice

[nova merlin]

I don't get one thing the programming synthesis thing they yell all the time in arc isn't codegemma similar ?

Or any coding assistant llm similar to it ?

[digital finch]

might be a bit hard for non reasoning model ig

vanilla deepseek can do ok-ish imo

I would have tried but I have no gpu left

[nova merlin]

Collab GPU quota ?

[digital finch]

no kaggle

[nova merlin]

Bro u r faster than xavier btw 💀

[digital finch]

😂

[nova merlin]

U ve Collab quota still remaining i meant

[digital finch]

does collab reveal any quota?

they just randomly limit me

[nova merlin]

Oh right 😅

Use slms

Use distilled slms

[digital finch]

where do you draw the line b/w slms and llms?

[nova merlin]

Below 1b

[digital finch]

they are very bad

[nova merlin]

Right upto 2bs are fine

[digital finch]

why use smaller models when you can use bigger once?

one reason can be to do best of n something

[nova merlin]

Right upto 4bs are fine too

@rich halo are u ai ?

[digital finch]

what is the problem with larger models?

[nova merlin]

Ok upto 8bs are fine too 🤧

[digital finch]

why not more than that?

[nova merlin]

never tried thought compute limit 💀🤧😅

[digital finch]

there is enough compute

96gb ish vram

you can fit in a 70b

with fp8

[nova merlin]

won't perf gap would be too significant in fp8 ?

[digital finch]

ofc not 😂

[nova merlin]

Once I wasn't able to fit deepseek in Collab long ago in fp16 💀🤧

[digital finch]

which deepseek?

[nova merlin]

Maybe 8b

Or maybe it was qwen

[digital finch]

distilled r1 or v2?

[digital finch]

oh

[nova merlin]

Anyways do they ve submission limits in arc too ?

[digital finch]

ofc 😂

[nova merlin]

Like 1 submission per day or below 1?

[digital finch]

1

never or saw below 1 anywhere 🙏

[nova merlin]

thnk God atleast 1 💀🤧

[nova merlin]

It's trend now for every comp to js set it to 1 per day

[digital finch]

no

these are high competitions

[nova merlin]

Yes I meant featured ones

[nova merlin]

For high ai-ers like Qafig

[digital finch]

featured ones have 5 a day usually

arc requires copius amounts of compute

• probing issues are there

[nova merlin]

Yeah but arc should ve atleast 2 🤧

[nova merlin]

wdym

[digital finch]

last year how many they had?

[nova merlin]

No idea i wasn't last year

[rich halo]

ash_blanc has been warned

Reason: Bad word usage

[digital finch]

can someone test out 2.5 pro on the public set?

[nova merlin]

gemini ?

The one launched recently ?

[digital finch]

yeah

[nova merlin]

U mean the experimental one or pro one ?

[digital finch]

the pro experimental one

[nova merlin]

Did they release pro one too ?

[digital finch]

2.5 pro

[digital finch]

ofc

they did yesterday or day before yesterday

the model is 🔥

[nova merlin]

I meant pro experimental they released 2 days ago

Oh ig yeah they related pro yesterday

[digital finch]

what?

2.5 pro is in beta so they call it experimental

I tried it on one question

I gave the image

[nova merlin]

Yes beta i meant are they still on the beta btw ??

[digital finch]

but for some reason it literally looks like arc submission

[digital finch]

it just released yesterday or smth, it is in beta ofc

[nova merlin]

Can't use it in gemini app

[digital finch]

no you can't

if you do not have advanced

use it on their AI studio

[nova merlin]

Yeah pretty cool

Anyways is code a different modality than text ?

I thought they come under same modality

[digital finch]

they are same

[digital finch]

why?

[nova merlin]

I thought they r both js bunch of strings

[digital finch]

yeah they are tokenized by the same tokenizer

unlike image or audio lets say

[nova merlin]

coz gemini said text2code is multi modal as they r different modalities

[digital finch]

huh?

[nova merlin]

Yes but still or maybe code is not much different than text to be considered a different modality

[burnt quiver]

Guys I am making a logistic regression project to put in my cv for upcoming campus placements I have no prior knowledge and what I've done is just few lectures on udemy, I made my model but I could only achieve a 88% accuracy can anyone help me with it?

[hallow whale]

What's your project about ?

[burnt quiver]

Predicting whether a person will make a purchase or not, by making use of the browsing behaviour of the customer

[nova merlin]

How will u get access of browsing behaviour of customer ?

[wicked berry]

Does anyone know what this Gemini error message means?

[opal shard]

Where have you faced it? You've reached your limit.

You can't use any more resources till a specific time. Usually 24h

[agile cave]

It is totaly impossible to give any help without knowing inputs and outputs to your syste, how they've been preprocessed and what your model is. Keeping it a stack

[agile cave]

How are you vectorizing "behavior". Let's start here. What algo, or algos are you using to process this info. 🤔

[wicked berry]

Thank u

[wicked berry]

After all, no matter how AI can crawl various webpages or discussions on Stack Overflow, it probably can't access backend database structures, right?

[agile bear]

Hi guys nice to meet you all

[shy mural]

Hi everyone,

My name is Alejandro (Alex), and I am a PhD student in Communication at USC. I'm looking for participants to support my dissertation research. This project aims to understand how data science practitioners interface between value systems by observing their work practices in key "AI development/use platforms."

I am looking for participants to take a 5-min survey and join me in a 75-90 minute virtual work demonstration session on Kaggle. You will be compensated for your time and effort.

Survey can be accessed here: https://usc.qualtrics.com/jfe/form/SV_8iYCIuAdvOP7HIG

All the best!

[lilac yoke]

Hello there everyone!

Long time on and off Kaggler.

For the R users here I am working on a package for downloading Kaggle Datasets directly into the R console.

Check it out here (be sure to star!): https://github.com/benyamindsmith/RKaggle

(works with csvs for now)

[formal tendon]

I tried changing the font style of my graphs, I tried all the available ways to do it but it doesnt budge, is this has something to do with Kaggle environment?

[hollow tangle]

I’ve been mining my 5000+ conversations using BERTopic clustering + temporal pattern extraction. Implemented regex based information source extraction to build a searchable knowledge database of all mentioned resources. Found fascinating prompt response entropy patterns across domains

Current focus: detecting multi turn research sequences and tracking concept drift through linguistic markers. Visualizing topic networks and research flow diagrams with D3.js to map how my exploration paths evolve over disconnected sessions

Has anyone developed metrics for conversation effectiveness or methodologies for quantifying depth vs. breadth in extended knowledge exploration?

Particularly interested in transformer based approaches for identifying optimal prompt engineering patterns

Would love to hear about ETL pipeline architectures and feature extraction methodologies you’ve found effective for large scale conversation corpus analysis

[deep island]

hi, someone has any idea to how to start on a real-time voice changer app?

is there any libray to get sound, process and make a fast output?

[wet tapir]

Python is ML's lingva franca?

[agile cave]

Real time voice changer? Well, if there is live processing, that's going to take some time. However, two notes on a piano (or whatever sound source), if separated by less then 100 msec, will sound like they are overlapping. So think about that constraint. AKA, whatever you build processes input in under 100msec. And as for libraries?There are plenty of libraries out there to capture, process, and also importantly, modulate sound. Finding them? Consider posing this question to an LLM like chatGpt, or Geminini, or whatever.

[agile cave]

It's easier for prototyping. And with APIs like Tensorflow, Numpy, etc. that interacts with C++ at a lower level, great for building rapidy training models without dealing with pointers and memory management. Also trivial to then translate code into C++ if say, your code is being used by a NASA Space shuttle, or intense video game. This is why it is favored.

[deep island]

is it crucial masterize numpy for data management?

[agile cave]

When managing numericak arrays in Python, especially large ones, or large sets, it does so more rapidly then using Vanilla python. It's a good one to have a basic understanding of. Also Pandas for tabular data management and organization (think Excel like aspects of data management)

[deep island]

Are you a Data Sciencist?

[agile cave]

Not really. I'm an AI/ML hobbyist. Started getting into programming to create Music and Art with computers. Picked a few things up along the way. That is all.

[deep island]

i'm thinking to make a try with data world approaching that i have a title which supports me a litle bit

[deep island]

oh nice, but it seems that you have knowledge

do you think that i can convert my self in to a data profesional without stadistics bases?

[agile cave]

It just comes fro working on projects, and cursing your computer hundreds of times when things don't work out 😅

[agile cave]

Math theory is not my forté. But you start to figure out how things work, and then they become tools in your toolbox for future projects. Also, be prepared to invent on your own statiscal principles that are known, without ever knowing them. If you want it, you'll get it. It's gonna take time though. Of a skill set, Programming has a steep learning curve in the beginning. But if you stay the course, doesn't matter where you started from.

My verdict: if it interests you, go for it. You learned to walk w/o having a background in it, right?

[deep island]

im a web programmer with some expirience with python, and libraries like numpy, pandas nad tensorflow

but i have neveer deeped in to i

[agile cave]

You'll do fine

[deep island]

actually i'm working on a business as full stack

with angular and java, but my last studies where about IA & Big data

that will be crucial and demanded in these next years

[agile cave]

java. ughh. to verbose. this is why I prefer python.

[deep island]

so im doubting about deep into or continue with my journey

[agile cave]

have you compared a hello world file written in python to one written in java? 😂 😢

[deep island]

yes bro xdd

i've coded with both languages

[agile cave]

java was great because of it's write once, run anywhere, and at the time that was great

[deep island]

but lower level languages are normally faster

[deep island]

yep

python for data is number 1

simple and effective

[agile cave]

sure, assembly is faster, but not great for rapid prototyping and testing ideas (novel algorithms)

[tardy osprey]

Hey, I want to know about the Python environment. Like, every time I have to download any new libraries, I get errors related to Python environment. Besides that, I already installed many versions of Python, like 3.10, 3.11, and 3.12. And also, I have Anaconda Navigator. So is there anybody who can help me?

[agile cave]

Manage environments with Anaconda

some libraries may be conflicting, so you can suss those problems by creating custom enviros in Anaconda

how do you install new api's: conda, pip, brew?

[agile cave]

also, i find source code is easieer to read

[tardy osprey]

Follow gpt blindly don't know anything

[tardy osprey]

And at last it solved but every time I'm facing this issue

[brittle walrus]

L

[versed jacinth]

when you search for python package on anaconda what are u seeing there ?

ofc in the environment ur using

[agile cave]

@tardy osprey kind of hard to know the possible nature of your problems w/o error messages

[versed jacinth]

idk if he's talking about error in terminal or IDE

[agile cave]

Deciphering Error Messages is an art, in itself 😅

[versed jacinth]

i think with the time most people dont read anymore the error and juste copy paste to our ai friend lmao

[leaden grove]

Hello everyone, I'm Hasini from Sri Lanka

[brittle cave]

Hello 👋😄 I'm sakib. From Bangladesh 🇧🇩

[calm dome]

Hi Everyone, I am Haseeb From India! Proud to be a part of this community

[narrow pier]

Hello, i am aditya malla thakuri, I am from Nepal and I am a Computer Engineering Student.

[thorny tendon]

Hi guys! I'm Tiago from portugal 🇵🇹. Happy to share this virtual space with yall 😄

[median cradle]

Hello all, Myself, Amirtha Varshni from India. I'm a Computer Science and Engineering student

[wary timber]

Hello.

[inland zephyr]

Hii

[wary timber]

Hello,

I'm Gustav Olaf Laitinen-Fredriksson - a Data Science Specialist at DTU in Copenhagen, Denmark, and Adjunct Instructor at Linköping University. I fuse academic rigor with industry innovation in proteomics and healthcare. Currently, as a PhD candidate, I'm exploring Human-XAI collaboration to enhance fetal ultrasound imaging. I'm excited to exchange ideas and grow with this community.

Best regards,
**Gustav Olaf Laitinen-Fredriksson Lundström **
Data Science Specialist in Proteomics at Technical University of Denmark
https://www.kaggle.com/olaflundstrom

[valid sandal]

Hello.

[ember cosmos]

Hello everyone! 👋 I’m Ali Jafar, an undergraduate Computer Science student and an AI application developer. My interests lie in Data Science, Generative AI, and Large Language Models (LLMs). Looking forward to learning, collaborating, and sharing insights with this amazing community!

[queen river]

Hi All, I am Nasir Munshi, I am a ML enthusiast, here to learn more about DS, Gen AI and LLMs. Looking forward to learning more about these with this community!

[nocturne leaf]

Hello all, look forward to the class!

[fleet crest]

hello, learning about AI, excited about this course

[rough dune]

Hi everyone! 🌹 I’m Misty, a generative AI specialist passionate about building creative, impactful tech solutions. Excited to be part of the 5-Day Gen AI Intensive with Google and connect with fellow innovators! Looking forward to learning, sharing, and growing together—let’s make the most of this challenge! https://www.linkedin.com/in/misty-waters-473b5950/

[opal shard]

How do I calculate the MSE for the predictions in the image?

[jagged tinsel]

Hi All, I'm Somya, a graduate in advanced mathematics and a post graduate in MBA Business Analytics . I'm here to learn about Data Science and Gen AI .

[timber depot]

how much u know about data science..?

[dusky gust]

I am a full-stack developer with a strong passion for building efficient and scalable applications. Currently, I am expanding my expertise into the field of data science, which fascinates me with its potential to derive meaningful insights from data and drive impactful decision-making. I look forward to combining my programming skills with data science techniques to create innovative solutions and tackle complex problems

[wind river]

Hello everyone, I'm Priti a undergraduate student pursuing a Bachelor of Technology degree in Computer Science Engineering. With a strong foundation in computer science fundamentals, I've developed a keen interest in Artificial Intelligence (AI) and its vast applications. I'm fascinated by the potential of AI to transform industries, revolutionize problem-solving, and improve lives. I aim to explore the intricacies of AI, machine learning, and data science. I'm excited to learn from experts, collaborate with like-minded individuals, and contribute to the development of innovative AI solutions that make a positive impact.

[tardy osprey]

Hello

[nova merlin]

Hello

[wary timber]

Does anyone know about Rmarkdown? While developing my own notebook, I am trying to get a good F1 score but I am stuck. Can anyone help me?

[crude tangle]

hi

[sick cave]

Hello!!

[serene oyster]

Hi, I'm a Swedish software developer, now turned into a Independent Researcher.

[sleek notch]

Hello, I am Hari and i am here to learn Gen AI and LLMs

[solemn loom]

Hi there, I am Kefi an aspiring Tech Generalist learning both Data Science and Web Development.

[fallow glen]

Hi, I'm a sociologist from chicago. eager to learn genAI...

[lethal hare]

Hello! I am an App Developer from Mumbai. Looking forward to learning from the 5 day gen-ai course

[grave lodge]

Hi, I'm Bilal Nadeem CS student. Joining 5 day Gen-AI course.

[maiden canopy]

Hi. I'm Nashrah Fathima, a Computer Science undergrad. Looking forward to learning from the 5 day Gen-AI Course.

[frigid acorn]

Hi I am Saurabh Diwan and new to AI and programming field. Excited to explore Gen-AI. Thanks.

[forest grail]

Hi, I'm a Business and Exec coach, excited to learn more about Gen-AI

[jolly plume]

Yo, nive to meet you everyone, I am a SWE at Citi

[pulsar vortex]

She/her

[scenic tapir]

Hi, I'm Daemon, a cloud services software engineer from China, looking forward to a 5-day general artificial intelligence course.

[solid plinth]

Hi, I'm Hoomaan from Yerevan, Armenia. Interested to learn AI. 🙂

[pure cosmos]

Hii I am Atharva from India ,here to learn more about Gen AI

[wet abyss]

Hi, I am Sam. Looking forward to class.

[rugged lance]

Hi, I am Dheeraj from India, excited for the Gen AI course

[bitter abyss]

Hi I'm Sidharth from Houston. I'm here to learn more about GenAI

[woven vapor]

Hello, my name is Luke Goeltzer, I am a senior and computer science major at Northeastern Illinois University in Chicago

[covert crow]

heyy, im Chirag D S, i am an Undergraduate in Computer Science and Engineering from Bangalore. Hope to connect with you all and have a great time!!

[vivid valley]

Greetings. I am new to this circle, drawn by the allure of discovery and the challenges that await. My path is one of seeking patterns, unraveling puzzles, and embracing the unknown. Across platforms and disciplines, I find meaning in innovation and creative problem-solving, always pursuing a deeper understanding.
This community intrigues me, and I look forward to seeing what unique opportunities and connections it holds. Perhaps our paths will align, and we can uncover new possibilities together. Until then, let curiosity guide you.

Call me Lux by the way...

[amber hinge]

Hey there everyone! I'm Qasim Khan a CS student from pakistan and registered for the 5 day intensive. Looking forward to working with you all

[small hearth]

Hello Kagglers, I'm Godfrey from Kisumu Kenya. Happy to be here. I am looking forward to learning from, connecting with, and collaborating with everyone.

[night garnet]

Hi I’m Tak from Jakarta. Looking forward to working with you all.

[rich halo]

syed_azeemuddin has been warned

Reason: Bad word usage

syed_azeemuddin has been banned

Reason: Too many infractions

[hollow spade]

Hi my name is Aalisha, and I'm from New Jersey. Excited to learn more about AI!

[trim briar]

Hello all, my name is Jay Panesar. I'm from the UK and am excited to learn more about generative AI. I'm also subscribed to the 5-day intensive course.

[rich halo]

divyanshu045766 has been warned

Reason: Bad word usage

[signal hill]

Hi I am Rohit. Joined the 5 day intensive program. Looking forward to learn, connect and collaborate with you all.

[rich halo]

divyanshu045766 has been warned

Reason: Bad word usage

divyanshu045766 has been banned

Reason: Too many infractions

tejase_001 has been warned

Reason: Bad word usage

[opal shard]

Don't introduce yourself here 🤬

[golden dirge]

Hi fellas!

[lethal mist]

Hi

[thin sentinel]

Hi all from Kent in the UK - looking forward to the AI training next week, Greg

[gilded solar]

Hello. Looking forward to learning

[wraith fulcrum]

Hi all I'm Mohab

[silk zephyr]

Hi, I am Samuel from Manchester, United Kingdom. I am ready to explore this exiting course.

[twilit jolt]

Hi I'm Rajvi (she/her). Excited to learn through the Gen AI intensive course

[pine trout]

Hi! I am Luis from Argentina. Very happy to be here!

[keen wave]

Hi! I'm Sophia. Ready to learn

[peak kraken]

Hello World ! I'm Mathieu from Paris,FR. Nice to meet you guys !

[rich halo]

sharjeelbhullar has been warned

Reason: Bad word usage

[topaz karma]

Hi Tjay here... Reporting for exploration!

[upbeat pebble]

Happy to join this course 🙌🙌

[stiff nymph]

Hi Everyone, Here for the GenAI course. Happy to be here!

[stable mural]

Hello i'm talha from pakistan nice to meet all of you

[versed jacinth]

cmon guys introduce ur self on the correct chanel #👤┊introductions or #5dgai-introductions

[ionic abyss]

Hi. I am Adam, joining the AI training this week!

[boreal nexus]

folks google ai studio is awesome , openai is just laaaame