@diepdd.bsky.social
Reposted
My path into AI
The sort of small wins that accumulate into a real career in AI.
When I started grad school AI prof's didn't have space for me in their group and when I ended I had no papers at NeurIPS/ICLR/ICML, yet the process can still work.
www.interconnects.ai/p/my-path-in...
The sort of small wins that accumulate into a real career in AI.
When I started grad school AI prof's didn't have space for me in their group and when I ended I had no papers at NeurIPS/ICLR/ICML, yet the process can still work.
www.interconnects.ai/p/my-path-in...
My path into AI
How I got here. Building a career brick by brick over 8 years.
www.interconnects.ai
May 14, 2025 at 2:29 PM
My path into AI
The sort of small wins that accumulate into a real career in AI.
When I started grad school AI prof's didn't have space for me in their group and when I ended I had no papers at NeurIPS/ICLR/ICML, yet the process can still work.
www.interconnects.ai/p/my-path-in...
The sort of small wins that accumulate into a real career in AI.
When I started grad school AI prof's didn't have space for me in their group and when I ended I had no papers at NeurIPS/ICLR/ICML, yet the process can still work.
www.interconnects.ai/p/my-path-in...
Reposted
May 14, 2025 at 11:14 PM
Reposted
DeepSeek AI's Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures
An in-depth analysis of the DeepSeek-V3/R1 model architecture and its AI infrastructure, highlighting key innovations such as Multi-head Latent Attention (MLA) for enhanced memory efficiency
An in-depth analysis of the DeepSeek-V3/R1 model architecture and its AI infrastructure, highlighting key innovations such as Multi-head Latent Attention (MLA) for enhanced memory efficiency
May 15, 2025 at 2:46 AM
DeepSeek AI's Insights into DeepSeek-V3: Scaling Challenges and Reflections on Hardware for AI Architectures
An in-depth analysis of the DeepSeek-V3/R1 model architecture and its AI infrastructure, highlighting key innovations such as Multi-head Latent Attention (MLA) for enhanced memory efficiency
An in-depth analysis of the DeepSeek-V3/R1 model architecture and its AI infrastructure, highlighting key innovations such as Multi-head Latent Attention (MLA) for enhanced memory efficiency
Reposted
Val Town
Val Town has 25 repositories available. Follow their code on GitHub.
github.com
May 16, 2025 at 4:43 AM
Reposted
Val Town's Townie
Have not tried it, but per them - For building complex full-stack projects - Code, prompt, branch, pull request
Have not tried it, but per them - For building complex full-stack projects - Code, prompt, branch, pull request
May 16, 2025 at 4:43 AM
Val Town's Townie
Have not tried it, but per them - For building complex full-stack projects - Code, prompt, branch, pull request
Have not tried it, but per them - For building complex full-stack projects - Code, prompt, branch, pull request
Reposted
Native GPU (CUDA) programming in Python?
Nvidia releases CUTLASS 4.0, which ships with CuTe DSL, a programming language that is fully consistent with CuTe C++ in its programming model, APIs, abstraction level, and performance.
Nvidia releases CUTLASS 4.0, which ships with CuTe DSL, a programming language that is fully consistent with CuTe C++ in its programming model, APIs, abstraction level, and performance.
May 13, 2025 at 11:20 PM
Native GPU (CUDA) programming in Python?
Nvidia releases CUTLASS 4.0, which ships with CuTe DSL, a programming language that is fully consistent with CuTe C++ in its programming model, APIs, abstraction level, and performance.
Nvidia releases CUTLASS 4.0, which ships with CuTe DSL, a programming language that is fully consistent with CuTe C++ in its programming model, APIs, abstraction level, and performance.
Reposted
The 2025 Zeitgeist is fascinating. We are living in the times where Microsoft will prevent screen capture from team meetings, on side we have tools like Granola recording and transcribing most meetings we're in. www.bleepingcomputer.com/news/microso...
Microsoft Teams will soon block screen capture during meetings
Microsoft is working on adding a new Teams feature that will prevent users from capturing screenshots of sensitive information shared during meetings.
www.bleepingcomputer.com
May 10, 2025 at 9:46 PM
The 2025 Zeitgeist is fascinating. We are living in the times where Microsoft will prevent screen capture from team meetings, on side we have tools like Granola recording and transcribing most meetings we're in. www.bleepingcomputer.com/news/microso...
Reposted
You can try out the newest and greatest Google TPU:
v6e-1 (Trillium) TPUs!! 2x the high bandwidth memory as v5e-1 (32GB) and a whopping peak rating of 918 BF16 TFLOPS (nearly 3x A100)!
on Google Colab.
v6e-1 (Trillium) TPUs!! 2x the high bandwidth memory as v5e-1 (32GB) and a whopping peak rating of 918 BF16 TFLOPS (nearly 3x A100)!
on Google Colab.
May 12, 2025 at 12:35 AM
You can try out the newest and greatest Google TPU:
v6e-1 (Trillium) TPUs!! 2x the high bandwidth memory as v5e-1 (32GB) and a whopping peak rating of 918 BF16 TFLOPS (nearly 3x A100)!
on Google Colab.
v6e-1 (Trillium) TPUs!! 2x the high bandwidth memory as v5e-1 (32GB) and a whopping peak rating of 918 BF16 TFLOPS (nearly 3x A100)!
on Google Colab.
Reposted
Thinking about writing CUDA code (i.e., Nvidia GPU code)?
Cognition Labs trained a model using RL for writing CUDA kernels. It outperforms top reasoning models (o3 & o4-mini) in writing CUDA code.
Blogpost: cognition.ai/blog/kevin-32b
HuggingFace: huggingface.co/cognition-ai...
Cognition Labs trained a model using RL for writing CUDA kernels. It outperforms top reasoning models (o3 & o4-mini) in writing CUDA code.
Blogpost: cognition.ai/blog/kevin-32b
HuggingFace: huggingface.co/cognition-ai...
Cognition | Kevin-32B: Multi-Turn RL for Writing CUDA Kernels
We are an applied AI lab building end-to-end software agents.
cognition.ai
May 9, 2025 at 1:32 AM
Thinking about writing CUDA code (i.e., Nvidia GPU code)?
Cognition Labs trained a model using RL for writing CUDA kernels. It outperforms top reasoning models (o3 & o4-mini) in writing CUDA code.
Blogpost: cognition.ai/blog/kevin-32b
HuggingFace: huggingface.co/cognition-ai...
Cognition Labs trained a model using RL for writing CUDA kernels. It outperforms top reasoning models (o3 & o4-mini) in writing CUDA code.
Blogpost: cognition.ai/blog/kevin-32b
HuggingFace: huggingface.co/cognition-ai...
Reposted
🎨 In case anyone need…
The NIH BioArt Source is an awesome library of *free* professionally drawn illustrations for scientific presentations or figures. Downloadable in HD. Thank you NIH for this invaluable tool 🙏!
Check it out 👇
bioart.niaid.nih.gov
The NIH BioArt Source is an awesome library of *free* professionally drawn illustrations for scientific presentations or figures. Downloadable in HD. Thank you NIH for this invaluable tool 🙏!
Check it out 👇
bioart.niaid.nih.gov
November 23, 2024 at 3:50 PM
🎨 In case anyone need…
The NIH BioArt Source is an awesome library of *free* professionally drawn illustrations for scientific presentations or figures. Downloadable in HD. Thank you NIH for this invaluable tool 🙏!
Check it out 👇
bioart.niaid.nih.gov
The NIH BioArt Source is an awesome library of *free* professionally drawn illustrations for scientific presentations or figures. Downloadable in HD. Thank you NIH for this invaluable tool 🙏!
Check it out 👇
bioart.niaid.nih.gov
Reposted
This was a great development. And I love how this is enable-able via a simple boolean kwarg in @lucidrains.bsky.social's excellent and continually-maintained library github.com/lucidrains/v...
January 23, 2025 at 2:18 PM
This was a great development. And I love how this is enable-able via a simple boolean kwarg in @lucidrains.bsky.social's excellent and continually-maintained library github.com/lucidrains/v...
Reposted
Introducing The AI Scientist-v2, which produced the 1st fully AI-generated paper to pass peer review at a workshop level (at #ICLR2025) ‼️
Tech Report: pub.sakana.ai/ai-scientist...
GitHub: github.com/SakanaAI/AI-...
This work is a proud collaboration between Sakana AI, UBC, and Oxford University.
Tech Report: pub.sakana.ai/ai-scientist...
GitHub: github.com/SakanaAI/AI-...
This work is a proud collaboration between Sakana AI, UBC, and Oxford University.
April 8, 2025 at 7:21 AM
Introducing The AI Scientist-v2, which produced the 1st fully AI-generated paper to pass peer review at a workshop level (at #ICLR2025) ‼️
Tech Report: pub.sakana.ai/ai-scientist...
GitHub: github.com/SakanaAI/AI-...
This work is a proud collaboration between Sakana AI, UBC, and Oxford University.
Tech Report: pub.sakana.ai/ai-scientist...
GitHub: github.com/SakanaAI/AI-...
This work is a proud collaboration between Sakana AI, UBC, and Oxford University.
Reposted
Shout out to "The Kaggle Book" by Luca Massaron and Konrad Banachewicz! Finally got around to ordering it and did not know that some of my Kaggle Notebooks are highlighted in this book (p. 107)!
February 19, 2025 at 7:53 PM
Shout out to "The Kaggle Book" by Luca Massaron and Konrad Banachewicz! Finally got around to ordering it and did not know that some of my Kaggle Notebooks are highlighted in this book (p. 107)!
Reposted
目指せメダリスト!Kaggle実験管理術 着実にコンペで成果を出すためのノウハウ (髙橋正憲,篠田裕之) が、Apple Booksで配信開始されました。
books.apple.com/jp/book/%E7%9...
books.apple.com/jp/book/%E7%9...
目指せメダリスト!Kaggle実験管理術 着実にコンペで成果を出すためのノウハウ
コンピュータ／インターネット · 2025年
books.apple.com
March 11, 2025 at 11:55 PM
目指せメダリスト!Kaggle実験管理術 着実にコンペで成果を出すためのノウハウ (髙橋正憲,篠田裕之) が、Apple Booksで配信開始されました。
books.apple.com/jp/book/%E7%9...
books.apple.com/jp/book/%E7%9...
Reposted
Will give this a shot since I’m a big supporter of reading for learning/growth. Hoping Kaggle comes in handy for the datasets
February 3, 2025 at 6:38 PM
Will give this a shot since I’m a big supporter of reading for learning/growth. Hoping Kaggle comes in handy for the datasets
Reposted
What skills have I really built through online learning?
To reflect on my recent progress in #AI #upskilling, I built a #DataScience notebook on #Kaggle to analyze learning achievements across Microsoft Learn and Google Cloud Skills Boost using #Python and #MachineLearning.
See below
👇
To reflect on my recent progress in #AI #upskilling, I built a #DataScience notebook on #Kaggle to analyze learning achievements across Microsoft Learn and Google Cloud Skills Boost using #Python and #MachineLearning.
See below
👇
April 11, 2025 at 1:39 PM
What skills have I really built through online learning?
To reflect on my recent progress in #AI #upskilling, I built a #DataScience notebook on #Kaggle to analyze learning achievements across Microsoft Learn and Google Cloud Skills Boost using #Python and #MachineLearning.
See below
👇
To reflect on my recent progress in #AI #upskilling, I built a #DataScience notebook on #Kaggle to analyze learning achievements across Microsoft Learn and Google Cloud Skills Boost using #Python and #MachineLearning.
See below
👇
Reposted
Day 8 of #30DayChartChallenge: Histogram. I took a look at one of my favorite books, The Lord of the Rings. I like how you can see the shifting narratives here. Have fun exploring! 🔎
Data: www.kaggle.com/datasets/ash...
Tools: R(tidytext, ggplot) and Affinity Designer.
Data: www.kaggle.com/datasets/ash...
Tools: R(tidytext, ggplot) and Affinity Designer.
April 8, 2025 at 10:22 PM
Day 8 of #30DayChartChallenge: Histogram. I took a look at one of my favorite books, The Lord of the Rings. I like how you can see the shifting narratives here. Have fun exploring! 🔎
Data: www.kaggle.com/datasets/ash...
Tools: R(tidytext, ggplot) and Affinity Designer.
Data: www.kaggle.com/datasets/ash...
Tools: R(tidytext, ggplot) and Affinity Designer.
Reposted
After 7 years of inactivity, today I updated the C++ implementation of my simple serial communication to the latest libserial version.
Code: github.com/araffin/ardu...
Blog post:
medium.com/@araffin/sim...
Code: github.com/araffin/ardu...
Blog post:
medium.com/@araffin/sim...
Simple and Robust {Computer — Arduino} Serial Communication
Communication With the Arduino Made Easy
medium.com
May 5, 2025 at 5:25 PM
After 7 years of inactivity, today I updated the C++ implementation of my simple serial communication to the latest libserial version.
Code: github.com/araffin/ardu...
Blog post:
medium.com/@araffin/sim...
Code: github.com/araffin/ardu...
Blog post:
medium.com/@araffin/sim...
Reposted
June 3, 2024 at 10:28 AM
Reposted
A Hands-On Guide to Fine-Tuning Large Language Models with PyTorch and Hugging Face leanpub.com/finetuning by Daniel Voigt Godoy is the featured book on the Leanpub homepage! leanpub.com #ai #MachineLearning #DeepLearning #NeuralNetworks #gpt #books #ebooks
March 31, 2025 at 7:52 PM
A Hands-On Guide to Fine-Tuning Large Language Models with PyTorch and Hugging Face leanpub.com/finetuning by Daniel Voigt Godoy is the featured book on the Leanpub homepage! leanpub.com #ai #MachineLearning #DeepLearning #NeuralNetworks #gpt #books #ebooks
Reposted
In fact, this is not "from scratch" and it uses pytorch library, however, this explains or tries to explain all the processes of LLM.
www.manning.com/books/build-...
www.manning.com/books/build-...
March 3, 2024 at 12:09 PM
In fact, this is not "from scratch" and it uses pytorch library, however, this explains or tries to explain all the processes of LLM.
www.manning.com/books/build-...
www.manning.com/books/build-...
Reposted
The Hundred-Page Language Models Book: hands-on with PyTorch leanpub.com/theLMbook by Andriy Burkov is the featured book on the Leanpub homepage! leanpub.com #Ai #Gpt #NeuralNetworks #DeepLearning #DataScience #ComputerScience #Linguistics #books #ebooks
Find it on Leanpub!
Find it on Leanpub!
April 22, 2025 at 6:15 PM
The Hundred-Page Language Models Book: hands-on with PyTorch leanpub.com/theLMbook by Andriy Burkov is the featured book on the Leanpub homepage! leanpub.com #Ai #Gpt #NeuralNetworks #DeepLearning #DataScience #ComputerScience #Linguistics #books #ebooks
Find it on Leanpub!
Find it on Leanpub!
Reposted
Making the default pytorch collate be recursive and aggressively try to convert and stack tensors was a mistake.
April 27, 2025 at 11:29 AM
Making the default pytorch collate be recursive and aggressively try to convert and stack tensors was a mistake.