abhishek
@abhi1thakur
Followers
81,750
Following
663
Media
1,726
Statuses
10,608
🤗 I build AutoTrain @huggingface 👨🏽💻 World's First 4x Grand Master @kaggle 🎥 YouTube 100k+: ⭐ GitHub Star
127.0.0.1
Joined December 2009
Don't wanna be here?
Send us removal request.
Explore trending content on Musk Viewer
billie
• 262368 Tweets
Xavi
• 109160 Tweets
علي النبي
• 85839 Tweets
Laporta
• 42623 Tweets
デザフェス
• 35940 Tweets
#يوم_Iلجمعه
• 35120 Tweets
下半身露出
• 26757 Tweets
Scottie
• 22990 Tweets
OUROAD IN CHIANGMAI
• 20911 Tweets
#Aぇいきなりスペース
• 17864 Tweets
紫耀くん
• 16023 Tweets
イッシュ
• 14932 Tweets
تشافي
• 14923 Tweets
Flick
• 13173 Tweets
Isco
• 10874 Tweets
Louisville
• 10683 Tweets
Post your favorite machine learning memes ⬇️ Here's one of my favorites 🤣
98
797
5K
"Attention is all you need" implementation from scratch in PyTorch. A Twitter thread:
1/
56
786
4K
Tesla's self driving models should be open source
57
290
3K
pip install pandas
pip install scikit-learn
pip install torch
pip install tensorflow
pip install transformers
update resume: experience in data analysis and expertise in machine learning, deep learning with a focus on natural language processing 🤣
64
328
2K
📢 A few months ago, I silently made my book [Approaching (almost) any machine learning problem] FREE. So, if you are planning to buy, you can rather read the PDF first and then decide if you really want to buy it :) Or just read it for free. Here it is:
44
415
2K
Here are some coding tutorials on large language models (LLMs): 🧵1/N
31
341
2K
The easiest LLM Fine Tuning UI just Landed! 🚀
Now, ANYONE can fine-tune (almost) any LLM available on Hugging Face Hub by just uploading a CSV and choosing the parameters and by a single click of a button! 💥 Here's how you can do it: 1/N
31
310
2K
I'm planning to start an applied machine learning course. Anyone interested?
133
75
1K
Everyone wants to train models, no one wants to create data. When was the last time you labeled/collected data yourself to train a model? :)
150
145
1K
LLAMA-v2 training successfully on Google Colab's free version! "pip install autotrain-advanced" 💥 Yes, you can also use your local machine!
32
247
1K
Data scientists building a hundred layer neural network without trying simple things like logistic regression which might be a better fit for the problem and perform better 🤣🤣🤣
36
230
1K
One of my techers used to say "If you are not writing, you are not learning".
When it comes to machine learning and Python, I say "If you didn't code, you didn't learn".
You can't just read or copy-paste code and think that you have learned it. You need to get your hands dirty.
30
151
1K
I have an awesome idea for a machine learning model
[100 people joined]
Let's start by labelling some data
[95 people left]
36
100
1K
🧵If you want to learn time series analysis, take a look at these great resources created by Konrad Banachewicz ⬇️ 1/N
12
197
1K
The EASIEST way to finetune LLAMA-v2 (or any other LLM) on local machine!
22
183
1K
🚀 If you are starting with machine learning / deep learning and get a new dataset to work on, either on kaggle or in real-world or just for fun. There are a few things you must always take care of to squeeze the most out of your model and make it awesome: ⬇️⬇️⬇️
1/6
22
239
1K
"A machine learning model is only as good as the data it is trained on". I can't believe so many people in the industry fail to understand this and expect some kind of magic when machine learning is mentioned. Have you ever faced this situation? How did you make them understand?
93
143
1K
Me when my machine learning model works on the very first try without even tuning any hyperparameters 🤣 🤣 🤣
31
154
1K
When you dont know the basics of machine learning and go for implementing a GAN 🤣🤣🤣
34
172
1K
Do you want to learn time-series analysis for free? Check out this thread 🧵 1/12
13
198
1K
🧵For Kaggle's 30 days of machine learning competition, I made several videos and notebooks on getting started. If you are planning to get started with Kaggle competitions, maybe you can give this thread a try 1/10
14
205
966
Introducing !!!! We are revolutionary. We are the best. We have "inventors" and we publish our research on medium. Here is a sample of our code:
32
59
967
joblib is one of my favorite python libraries to run code using multiprocessing (Parallel & delayed). It's just too simple to use and when used right, makes code quite fast. i use it wherever code takes too much time and i know that it can run in parallel and be fast :D
16
129
922
🎉 🤗 There is a huge difference between using a library & developing it. So far, I have just been a user but now I am glad to announce that I am going to be a part of the amazing team which has developed the best SOTA libraries when it comes to NLP: Hugging Face!
56
36
914
Understanding self-attention. A Twitter thread 🧵
To understand what self-attention is and how it works, you need to know only the following terms:
- dot product
- matrix multiplication
- softmax
Note: it's not rocket science. let's keep it simple. 1/
18
166
906
My home setup for machine learning / deep learning:
- Ubuntu machine with 32GB ram, i9, Titan RTX + 3090, 1TB SSD
- Macbook to SSH into Ubuntu machine. VSCode with SSH extension. I code on MacBook and run everything on Ubuntu
- A windows machine for making/editing YouTube videos
41
43
843
On popular demand, I have open-sourced my book: "How To Become A Data Scientist In 30 Days". Grab your free copy from the github repository:
64
124
839
Have you had troubles or having troubles arranging your machine learning projects? This thread should give you some idea on how to arrange machine learning / deep learning projects. See the folder structure: 1/6 🔽
20
150
826
Super-excited to announce that I have reached worldwide rank # 1 in Kaggle's Notebooks category with more than 50 gold medals!!! 🎉🎉🎉 Thank you to everyone who has supported me till now. I love writing notebooks and will continue doing so! 🤗
36
36
822
oooohhhkay, chatGPT seems to have screwed up here....
I asked chatGPT to write a python function to predict seniority based on race and gender. See the result for yourself :/
86
115
802
Everything is moving so fast in the world of machine learning and ai that if you take a break for a few days, you end up missing a lot and have a lot to catch up. I havent taken a break for a long time now but still unable to catch up. Anyone else feel the same?
73
50
801
1/ 🧵 Getting started with data science and machine learning.
- the first step is to know what data science and machine learning mean and is this field worth getting into?
- if you ask me, I would say yes. the world is data-centric.
15
182
774
👉 Want to learn Natural Language Processing by solving problems? Here is a list of my favourite NLP competitions on
@kaggle
to learn from ⬇️ 1/15
16
178
738
On July 1st, 2019, I became the very First Triple Grandmaster on Kaggle.
Competitions ✔️
Kernels ✔️
Discussions ✔️
@kaggle
Thank you for everything! :)
42
52
747
Second month into 2020 and I conquered the 4th (and hopefully final) category on Kaggle! With this I become the World's First 4xGrandMaster on Kaggle!
Competitons ✅
Discussions ✅
Kernels ✅
Datasets ✅
@kaggle
Thank you for everything! :)
57
58
738
I have started an applied PyTorch 101 series (especially for beginners but may also help some intermediate people). Check out the playlist:
6 videos currently, more coming soon & when you are there, don't forget to SUBSCRIBE!!! Learning deep learning ;)
13
125
725
But I want to fine-tune a Llama2 model in free version of Google Colab
Say no more! With AutoTrain Advanced, you can 😱 Now anyone can use colab to finetune LLMs just by uploading data and tuning parameters! 🚀 1/N
14
142
734
Here's all you need to train a 7b Falcon LLM on Google Colab (or your local machine) Enjoy! 🚀
13
110
723
Building a recommendation system in python using deep learning was my last tutorial on youtube. Here is the entire code for that. Short, concise and pretty 🤩 If you missed the tutorial, you can watch it here:
13
112
689
Asking Data Scientists or ML Engineers to show demos in every meeting is absurd. Project Managers who call themselves DS PMs must show some effort to learn a lil bit about DS and ML: nothing technical, just intro & workflow. Its not like frontend where you always "see" something
24
108
702
If I make a competition that focuses on data collection/data labeling and then building machine learning models, who would be interested in taking part?
83
34
683
👉 One of the most important things when doing quick experiments with different models in machine learning or deep learning is to streamline the process! For this to work, you need to structure your training script well. Here is how I like to do it. And you? :)
21
83
686
No AWS? No Azure? No GCP? 😥
No problem! All you need is Hugging Face! 🤗
Now, just by adding one parameter in AutoTrain, you can fine-tune custom LLMs on Hugging Face Spaces and choose GPUs like T4, A10g & even A100! 🚀🚀🚀
Your training will start in a few seconds and space
13
124
663
Today, im making a tutorial that shows how you can deploy machines learning (& deep learning apps) on google cloud platform. any one interested?
43
42
655
Wow! In 50 days you can do python and also R!!! Learn stats, calculus, ml, whatnot!!!!!🤦 and just the titanic project??? Now its time we should all leave titanic alone and do something else! 🤣
Damn. Data Science only takes 50 days. Python takes only takes 5....
221
450
4K
51
112
652
Whoa! 😱 Something's trending on Github # 2!!!! Yes, it's your favourite machine learning book's repository: "Approaching (Almost) Any Machine Learning Problem". Wuhuuu!!! Thank you, everyone!!! 🎉🎉🎉 The book is available in PDF format for FREE. Read it & you like it, buy it ;)
7
80
650
Most Machine Learning articles on medium are really very bad quality and repetitive. Titles are usually clickbaits. Most start with a story which is utter nonsense and totally not required. In some 5-10% content is useful but most are fully useless. Sorry if I hurt feelings.
55
59
639
🚀 I just got access to
@github
Copilot and it's super amazing!!! This is going to save me so much time!! Check out the short video below!
#GitHubCopilot
I think I'll spend more time writing function descriptions now than the code itself :D
30
128
647
Just arrived! Deep Learning with Python 2nd Edition by
@fchollet
. My holiday gift to myself :D
17
28
605
Finetuning, llama2-13b on a custom dataset. ETA 15mins per epoch - powered by AutoTrain 🚀 pip install autotrain-advanced 🔥🔥🔥
25
59
602
If you want to learn machine learning, dont wait for a mentor. Hand-holding is not something you should look for. Start yourself and you will find mentors. They are everywhere. They share tutorials, code, learning path, etc. Without you even knowing, you will have a mentor!
9
53
580
💥 Did you know that there are problems other than MNIST and iris that you can solve (or try to solve) to learn deep learning and computer vision? Here is a list of my favourite Kaggle competitions to learn deep learning and computer vision from ⬇️ 1/13
5
124
572
No more CUDA, python, libraries issues! AutoTrain Advanced comes with it's own docker container! Just pull the docker and finetune LLMs or build your own dreambooth models (more tasks coming soon!)
Also available on pip: pip install autotrain-advanced 🚀
12
97
578
Instruction finetuning (or general finetuning) an LLM has never been easier. All you need is data and a local machine (or google colab, or a cloud machine). Here are a few simple steps to follow and train an LLM without writing any code! 1/N
5
81
572
How to become a data scientist: 1000 likes
Tutorial on approaching a machine learning dataset: 100 likes
This is whats wrong with data science "enthusiasts" these days ¯\_(ツ)_/¯
32
46
552
🚀 AutoNLP is already in beta & will soon open for all. If you want to be the first to try AutoNLP, all you have to do is click on the join button in the link below. ;)
AutoNLP enables:
🔷 Auto model selection
🔷 Auto fine-tune models
🔷 Auto deploy!
8
138
555
🔥We have been talking a lot about AutoNLP recently but only a few have seen it in action till now. So, I have created an exclusive preview video that shows how easy it is to train models using AutoNLP!!! Check it out here:
8
92
552
🔥I'll be releasing walkthrough videos every day for
@kaggle
's 30 days of Machine Learning. The first video of the series will premiere soon. This series will be beginner-friendly! Link to Day-1:
6
86
530
🚨 NEW TUTORIAL ALERT 🚨
The EASIEST way to finetune llama-v2 on local machine with custom dataset!
P.S. the tutorial also works for any other LLM and can also be used on the free version of google colab!
Check it out here: and don't forget to subscribe!
12
111
540
🎉 Introducing Tez: a sweet, simple library that you can use to train PyTorch models fasterrrrr. (yet keeping everything pythonic and clean). Check it out here: and don't forget to ⭐️ the repo :) I will be using Tez for all my future tutorials!
15
50
525
🚀 In 2022, if you want to learn
🐍 Python
💻 Machine Learning / Deep Learning
📈 Data Science
Then follow your passion, stop slacking and start immediately. There are unlimited free resources available online. Choose carefully and happy learning 🙂
18
52
528
Today I am releasing a library that lets you read files (csv, images, audio, anything), build simple and complicated machine learning and deep learning models, evaluate them, stack them, and deploy on gcp, aws, azure. all in just 2 lines of python code. who wants it? 🤣
49
29
521
Time-Series Specialist,
@tng_konrad
, has kindly agreed to collaborate with me on a series of tutorials on time-series analysis. If you are interested in learning about time-series, you cannot miss this! Join MLSpace Discord for Q&A and further info:
15
63
525
That moment when your watch tells you status of training. ;)
12
56
523
i like all these books. especially the one on top 😛😉
Here is a snapshot of a (small) subset of all of my Coding, Data Science and Machine Learning books. This collection would get you close to 98%-99% of all the necessary core skills to be a good Data Scientists. 1/6
129
1K
7K
12
53
510
If you are a beginner starting with
@kaggle
, I recommend the following:
🔷 Start with a couple of playground competitions (< 2 months)
🔷 Read solutions from past competitions & implement them yourself
🔷 Immediately move to an on-going competition w/ prize money (most to learn)
12
72
515
👉 This is how I think a Data Science / Machine Learning Team should work:
- Inception of an idea (business + data scientists)
- Collaborate with the data engineering team to get appropriate data
- Exploratory analysis in notebooks
🧵
1/4
7
98
504
Excited to announce that I am now a Google Developer Expert in Machine Learning 🎉 Special thanks to
@lucamassaron
@ksoonson
@A_K_Nain
@mervenoyann
@martin_gorner
@KshitizRimal
and thanks to everyone else who supported me till now 🤗
30
23
509
➡️How to use deep learning and build a neural network to tackle a tabular dataset that consists of both categorical and numerical variables?
- use entity embeddings for categorical vars
- combine with numerical vars
- magic!
My generic approach is shown below:
10
73
501
Throw a stone and it will land on a data scientist or a data science enthusiast these days. All of them can do basic math, analytics and have ML skills. Try to differentiate yourself. Learn something extra. Take initiatives. Learn the basics. Do things differently!!!!
22
53
487
I like XGBoost. I like Optuna. So, I started a new pet project: AutoXGB. Input data -> output best model -> deploy using FastAPI. Still, a lot left to do but I wanted to share it with everyone. Very simple project: WDYT? Will something like this be useful?
11
55
495
🔥🔥🔥 With the power of Google Colab, Ngrok & FastAPI and the simplicity of ColabCode, you can now deploy your applications on Google Colab or Kaggle Notebooks (or anywhere else) :) Check out ColabCode here:
pip install colabcode ;)
10
80
493
Thinking of making a video on how to deploy machine learning models on kubernetes. Anyone interested?
50
14
485
I've recently been told that using a YAML file to specify parameters and config information (specifically for machine learning stuff, e.g. training a model) is old school. So, could someone please tell me what the cool kids are using these days? :)
72
27
487
How to train LLAMA-v2 on custom dataset in just 50 lines of code:
4
74
486
How to finetune llama2 on custom dataset (on google colab or on a local machine or anywhere) in just one image! 🚀
9
94
486
Will inverting a binary tree help Facebook come back up again? 🤣
11
40
484
🔥New Tutorial: Building a recommendation system using deep learning. Check it out here:
9
70
476
It's a bit disappointing to see how so many people are interested in "how to become a data scientist" but only a very small percentage of those put in the effort and do the actual work required to become a data scientist. ¯\_(ツ)_/¯
36
31
467
I'm happy to announce Hugging Face Competitions! With 🤗 Competitions, you can create public/private competitions with full control of datasets, metrics and evaluation! Check out our first competition: AI or Not 1/N
20
99
462
🔥 Hugging Face's AutoNLP will make choosing a model, adapting the model to your data & deploying a model a piece of cake for almost all NLP problems: classification, NER, summarization, translation, etc. If you want to be part of the beta, register here:
7
92
461
Recently, a MOOC company asked me if I would design a course for them.
My resp was: If i can do short tutorials and share them on my YouTube channel for "free", why should i do it for you? So that you can charge 1000s of students $$$ for a mere certificate?
They didnt reply.
26
24
457
Exciting News!!! 🚀🚀🚀 Now, you can run AutoTrain UI locally and train models on your own local hardware!!! With just a few commands, you can start & run AutoTrain UI anywhere you want & train state of the art models on a variety of tasks without writing a single line of code.
7
78
451
The first step when approaching a machine learning problem is to split the data into folds. For binary or multiclass classification problems, you can blindly use stratified k-fold and it will keep the ratio of labels across all folds consistent. This is how I do it.
9
61
446