Pinned Tweet
I actually went through each of the 68 computer vision tutorials in keras to understand the common loss functions that were being used to solve the different problems.
Learned a lot, hope it helps someone :)
datamovesme.com/2022/08/03/an-…
Learned a lot, hope it helps someone :)
datamovesme.com/2022/08/03/an-…
1
4
17
We're all over here using LLMs to code and feeling like rockstars.. meanwhile, those same models trip over tasks scientists handle routinely.
@joinHandshake tested 10 frontier models on VIALS, a benchmark of 161 interpretation tasks from real lab workflows.
Nothing cleared 27%.
@joinHandshake tested 10 frontier models on VIALS, a benchmark of 161 interpretation tasks from real lab workflows.
Nothing cleared 27%.
AI promises to cure disease, but can frontier models interpret the images biotech/pharma scientists handle every day?
To evaluate this, we built the ATLAS Visual Life Sciences (VIALS) benchmark: 161 visual interpretation tasks from professional life sciences workflows including , molecular structures, and phylogenies. These are the images/results that experiments actually produce, not polished figures from papers or textbooks.
We benchmarked today’s 10 leading multimodal models. Opus 5, GPT 5.6, and Gemini 3.7 all fail to crack 27% accuracy, and the other models like Kimi K3 & Grok 4.6 perform much worse.
In contrast, PhD-level biomedical scientists find these visual interpretation tasks straightforward. AI that cant interpret such images will have limited utility in life sciences R&D, where such artifacts are central to how scientists reason, communicate, and make research decisions.
We hope VIALS drives progress toward models that can.
0
0
1
155
Kristen Kehrer retweeted
In "Susie’s School Bus Solution", young readers will join Susie and her mom as they learn about the fascinating world of artificial intelligence #AI and computer vision.
Get the book for a young person in your life, or for yourself: amzn.to/43oma64 by @DataMovesHer
Get the book for a young person in your life, or for yourself: amzn.to/43oma64 by @DataMovesHer
3
2
9
1,272
Kristen Kehrer retweeted
Thrilled to announce Express, a conversational data engineering platform, powered by agents, that makes working with data as simple as describing what you need.
The biggest blocker to AI adoption isn’t the models - it’s the complex, time-consuming process of preparing and undreds of systems.
Express solves that! Describe what you need in natural language. Express figures out where the data lives, connects to it, transforms it, and prepares it for use. No coding required. Production-ready in minutes.
The teams that move fastest with AI will be the ones who can unlock their data fastest.
Now, anyone can, with Express.
Try it today → express.dev
The biggest blocker to AI adoption isn’t the models - it’s the complex, time-consuming process of preparing and undreds of systems.
Express solves that! Describe what you need in natural language. Express figures out where the data lives, connects to it, transforms it, and prepares it for use. No coding required. Production-ready in minutes.
The teams that move fastest with AI will be the ones who can unlock their data fastest.
Now, anyone can, with Express.
Try it today → express.dev
15
24
140
22,923
Kristen Kehrer retweeted
In "Susie’s School Bus Solution", young readers will join Susie and her mom as they learn about the fascinating world of artificial intelligence #AI and computer vision.
Get the book for a young person in your life, or for yourself: amzn.to/43oma64 by @DataMovesHer
Get the book for a young person in your life, or for yourself: amzn.to/43oma64 by @DataMovesHer
0
2
8
1,887
People are always asking what it looks like to go end-to-end algo trading with Python. This covers the pipeline and some of the tools. datamovesme.com/blog/the-momen…
5
0
5
1,948
From notebooks to production backtests in algo trading: momentum+value in Python (data→features→MLflow→Prefect→IBKR).
datamovesme.com/blog/the-momen…
#DataScience #Backtesting #Python
datamovesme.com/blog/the-momen…
#DataScience #Backtesting #Python
0
1
0
116
Kristen Kehrer retweeted
One of the most beautiful and unique things of being a software developer:
The more code you write, the more bugs you’ll have to fix in the future.
The more time you spend working, the longer they’ll need you.
Self-generating demand. It’s amazing.
The more code you write, the more bugs you’ll have to fix in the future.
The more time you spend working, the longer they’ll need you.
Self-generating demand. It’s amazing.
43
13
272
23,263
Kristen Kehrer retweeted
How to make a simple algorithmic trading strategy with a 472% return using Python.
A thread. 🧵
A thread. 🧵
16
82
894
110,357
Kristen Kehrer retweeted
🚨BREAKING: A new Python library for algorithmic trading.
Introducing TensorTrade: An open-source Python framework for trading using Reinforcement Learning (AI)
Introducing TensorTrade: An open-source Python framework for trading using Reinforcement Learning (AI)
6
94
847
62,349
Backtesting a trading strategy is one giant experiment.
Yet most traders throw away their experiment history.
I started using MLflow (yes, the ML tool) to track my non-ML backtests, and it’s a game changer for iteration speed + insights.
Here’s how: datamovesme.com/blog/why-you-s…
Yet most traders throw away their experiment history.
I started using MLflow (yes, the ML tool) to track my non-ML backtests, and it’s a game changer for iteration speed + insights.
Here’s how: datamovesme.com/blog/why-you-s…
0
0
0
116
When thinking about your portfolio projects, think about actually putting something in production. Pay the couple dollars a month, and you can always cancel later.
1
0
0
98
With the ability to share screenshots with ChatGPT, you’re able to tell the LLM “hey, you told me to find some button I don’t see, here’s a screenshot, where’s that button”, and ChatGPT may say “I made a mistake, there’s no button, do this instead”.
1
0
0
69
But you’re able to iterate faster than ever before. (Although do lookout for ending up in some never ending loop where you’re getting dumb recommendations and you need to actually dig in yourself, that does happen). But go that last mile, it’s a much more impressive portfolio.
0
0
0
49
Kristen Kehrer retweeted
In "Susie’s School Bus Solution", young readers will join Susie and her mom as they learn about the fascinating world of artificial intelligence #AI and computer vision.
Get the book for a young person in your life, or for yourself: amzn.to/43oma64 by @DataMovesHer
Get the book for a young person in your life, or for yourself: amzn.to/43oma64 by @DataMovesHer
3
3
9
2,362
"Susie's School Bus Solution" just launched!
I had built a computer vision system to detect the school bus and send a text. I finally turned the process of building this into a children's book.
a.co/d/33iGr28
#computervision #machinelearning #python #datascience
I had built a computer vision system to detect the school bus and send a text. I finally turned the process of building this into a children's book.
a.co/d/33iGr28
#computervision #machinelearning #python #datascience
0
1
1
125
In a recent Mavens of Data episode, Nick Singh shared secrets for acing your analytics interview and landing your dream job.
From working at Facebook to writing bestsellers, he shares it all.
From working at Facebook to writing bestsellers, he shares it all.
1
0
0
159
Don't assume - ask smart questions.
Clarify the context of interview questions, especially for open-ended ones.
It shows you think critically and want to solve the right problem.
Clarify the context of interview questions, especially for open-ended ones.
It shows you think critically and want to solve the right problem.
1
0
0
98
Listen to the full episode with Nick Singh now: mavenanalytics.io/mavens-of-data…
What's your biggest takeaway from Nick's advice?
What's your biggest takeaway from Nick's advice?
0
0
0
85