Problem Solving with Arman (ML+Algo)
https://youtu.be/kYkIdXwW2AE?si=lKaiSG9Fhr_RZ-Un
YouTube
Yann LeCun's $1B Bet Against LLMs [Part 2]
Huge thanks to KiwiCo for sponsoring today’s video! Go to https://www.kiwico.com/welchlabs and use code WELCHLABS for 50% off your first monthly crate or for 20% off your first Panda Crate!
Welch Labs Book: https://www.welchlabs.com/resources/ai-book-ezrzm…
Welch Labs Book: https://www.welchlabs.com/resources/ai-book-ezrzm…
🧠 Time to Trivia!
If you love random facts, history deep-dives, or just want to test your trivia-skills -> Encyclopædia Britannica has a massive repository of trivia quizzes covering absolutely everything:
🎬 Pop Culture & Movies
🏛️ World History & Mythology
🧬 Science & Tech
🌍 Geography & Nature
👉 Tap here to test your knowledge: https://www.britannica.com/quiz/browse
If you love random facts, history deep-dives, or just want to test your trivia-skills -> Encyclopædia Britannica has a massive repository of trivia quizzes covering absolutely everything:
🎬 Pop Culture & Movies
🏛️ World History & Mythology
🧬 Science & Tech
🌍 Geography & Nature
👉 Tap here to test your knowledge: https://www.britannica.com/quiz/browse
Encyclopedia Britannica
Trivia Quizzes | Britannica
Take these quizzes at Encyclopedia Britannica to test your knowledge on a variety of fun and interesting topics including Animals, Art, Music, Pop Culture, Science, History and more!
Calculate Number of Parameters in Neural Network:
For ML engineers knowing the parameter count is vital because it determines a model's memory footprint and computational requirements, ensuring it can actually run on target hardware like GPUs or mobile devices.
Trainable parameters in a neural network are the internal variables (primarily weights and biases) that the model automatically adjusts during training to learn from data and improve its accuracy.
It serves as a direct measure of model capacity, helping engineers balance the complexity of the architecture to prevent overfitting while maintaining efficient training and inference speeds.
> Problem link (implement using numpy ~ recommended): https://www.deep-ml.com/problems/291
Dense layer:
CNN layer:
Embedding layer:
For ML engineers knowing the parameter count is vital because it determines a model's memory footprint and computational requirements, ensuring it can actually run on target hardware like GPUs or mobile devices.
Trainable parameters in a neural network are the internal variables (primarily weights and biases) that the model automatically adjusts during training to learn from data and improve its accuracy.
It serves as a direct measure of model capacity, helping engineers balance the complexity of the architecture to prevent overfitting while maintaining efficient training and inference speeds.
> Problem link (implement using numpy ~ recommended): https://www.deep-ml.com/problems/291
Dense layer:
trainable_params = input_size * out_size + out_size * int(use_bias)
CNN layer:
trainable_params = kernel_width * kernel_height * in_channels * out_channels + out_channels * int(use_bias)
Embedding layer:
trainable_params = num_embeddings * embedding_dim
Deep-Ml
Calculate Number of Parameters in Neural Network | Deep-ML
Solve "Calculate Number of Parameters in Neural Network" — a medium Deep Learning coding challenge on Deep-ML. Practice your ML skills with hands-on problems.
def test():
try:
ans = "repeat" * 3.14
return ans
except:
return "exceptional"
finally:
return "finally over"
print(test())
#python_code_trivia
Forwarded from Young&&Yandex
Долго сидишь в коде — делай пит-стоп. Предлагаем короткую разминку, можно выполнять упражнения прямо на стуле и не отходя от компа. Только осторожно
Скоро стартанём в настоящих Тренировках по алгоритмам
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
🔥2❤1
Forwarded from Proggy-Buggy Contest (bot for editing and posting)
4 days left 👀
In just four days, hundreds of participants from different countries will start their 42-minute journey through the algorithmic galaxy 🚀
Some are already practicing.
Some are choosing their lucky towel.
Some have registered weeks ago.
If you're still thinking about it — now is the time.
🏆 Prizes
🌍 Participants from around the world
🧠 Algorithmic challenges
🦆 The most unusual programming contest you'll ever join
Register now and see what you can do in 42 minutes:
👉 proggy-buggy.com/register
See you on June 6. And don't panic 😉
In just four days, hundreds of participants from different countries will start their 42-minute journey through the algorithmic galaxy 🚀
Some are already practicing.
Some are choosing their lucky towel.
Some have registered weeks ago.
If you're still thinking about it — now is the time.
🏆 Prizes
🌍 Participants from around the world
🧠 Algorithmic challenges
🦆 The most unusual programming contest you'll ever join
Register now and see what you can do in 42 minutes:
👉 proggy-buggy.com/register
See you on June 6. And don't panic 😉
Proggy-Buggy Contest
4 days left 👀 In just four days, hundreds of participants from different countries will start their 42-minute journey through the algorithmic galaxy 🚀 Some are already practicing. Some are choosing their lucky towel. Some have registered weeks ago. If you're…
yesterday's post: actually -> 3 days left. Contest will be 6th of June 🙌🔥
Proggy-Buggy Towel Contest 2026 халықаралық бағдарламалау жарысы
42 минут. 13 тапсырма.
Алгоритмдік, логикалық және ерекше форматтағы тапсырмалардан тұратын халықаралық бағдарламалау сайысы.
📅 6-маусым, 2026
⏰ 15:00 (Жергілікті уақыт)
👤 Жеке қатысу
🌍 Онлайн бүкіл әлем бойынша + офлайн хабтар
Қазақстандағы оффлайн хабтар:
Aлматы – Байзақов к., 280, Оңтүстік мұнара, DataArt офисы
Aстана – Сығанақ даңғ., 60/2, Абу-Даби Бизнес Центрі, DataArt офисы
Қарағанды – Алалыкин к., 12, «Terricon Valley» IT хабы
🏆 Санаттар:
• Amateurs • Professionals • Superheroes
Неге қатысу керек?
✔ Логика мен problem-solving дағдыларын дамытуға мүмкіндік
✔ Әлемнің түкпір-түкпірінен келген қатысушылармен жарысу
✔ Сыйлықтар мен қатысу сертификатын алу
✔ DataArt Internship Program рекрутерлерінің назарына ілігу мүмкіндігі
🧠 Жарысқа дейін дайындық тапсырмалары қолжетімді болады.
💻 Қатысу тегін
Тіркелу: proggy-buggy.com/register
⚠ Тіркелудің соңғы күні — 2026 жылғы 5 маусым
🟡 Егер офлайн хабқа сүлгімен келсеңіз — ұйымдастырушылардан арнайы сыйлық аласыз.
42 минут. 13 тапсырма.
Алгоритмдік, логикалық және ерекше форматтағы тапсырмалардан тұратын халықаралық бағдарламалау сайысы.
📅 6-маусым, 2026
⏰ 15:00 (Жергілікті уақыт)
👤 Жеке қатысу
🌍 Онлайн бүкіл әлем бойынша + офлайн хабтар
Қазақстандағы оффлайн хабтар:
Aлматы – Байзақов к., 280, Оңтүстік мұнара, DataArt офисы
Aстана – Сығанақ даңғ., 60/2, Абу-Даби Бизнес Центрі, DataArt офисы
Қарағанды – Алалыкин к., 12, «Terricon Valley» IT хабы
🏆 Санаттар:
• Amateurs • Professionals • Superheroes
Неге қатысу керек?
✔ Логика мен problem-solving дағдыларын дамытуға мүмкіндік
✔ Әлемнің түкпір-түкпірінен келген қатысушылармен жарысу
✔ Сыйлықтар мен қатысу сертификатын алу
✔ DataArt Internship Program рекрутерлерінің назарына ілігу мүмкіндігі
🧠 Жарысқа дейін дайындық тапсырмалары қолжетімді болады.
💻 Қатысу тегін
Тіркелу: proggy-buggy.com/register
⚠ Тіркелудің соңғы күні — 2026 жылғы 5 маусым
🟡 Егер офлайн хабқа сүлгімен келсеңіз — ұйымдастырушылардан арнайы сыйлық аласыз.
Tokens-per-Second Throughput from Inference Intervals
Tokens-per-Second (TPS) measures the processing speed of Large Language Models (LLMs), dictating how many chunks of text (tokens) a model can process as input or generate as output in one second.
~ classical interview-like question for NLP / AI engineers on understanding of base metrics for evaluating performance of LLMs;
> Deep-ML problem link (implement in raw-python): https://www.deep-ml.com/problems/1032
Tokens-per-Second (TPS) measures the processing speed of Large Language Models (LLMs), dictating how many chunks of text (tokens) a model can process as input or generate as output in one second.
~ classical interview-like question for NLP / AI engineers on understanding of base metrics for evaluating performance of LLMs;
> Deep-ML problem link (implement in raw-python): https://www.deep-ml.com/problems/1032
Deep-Ml
Tokens-per-Second Throughput from Inference Intervals | Deep-ML
Solve "Tokens-per-Second Throughput from Inference Intervals" — a medium Machine Learning coding challenge on Deep-ML. Practice your ML skills with hands-on problems.