rcnnnghm

[Jim Dowling @jim_dowling]

What a second half from Mayo! The wait is over. My mum was 10 years old when they won last 75 years ago!

[Sean Blanchfield @seanblanchfield]

You can solve most of the @openclaw security issues using Jentic’s credential manager. You’re not pasting API keys or credentials into the context window - the agent never sees the key so can’t leak them.
As a bonus it reduces tool bloat and does managed execution.
We had this built this for enterprises already and jumped to using it with OpenClaw when devs reached out having done this.

[Pau Labarta Bajo @paulabartabajo_]

Training ML models is easy.

Transforming the data these models need is the hard part... until you learn this ↓↓↓

𝗧𝗵𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺

Building a real-world ML system is

> 𝟭𝟬% about training and deploying ML models,
and
> 𝟵𝟬% about transforming the data these models needs to work.

And the thing is, 𝗻𝗼𝘁 all data transformations are the same.

𝗧𝗵𝗲 𝘁𝗮𝘅𝗼𝗻𝗼𝗺𝘆 𝗳𝗼𝗿 𝗗𝗮𝘁𝗮 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻𝘀

In every ML system we can have up to 3 different types of data transformations

1️⃣ 𝗠𝗼𝗱𝗲𝗹-𝗜𝗻𝗱𝗲𝗽𝗲𝗻𝗱𝗲𝗻𝘁 Transformations, for example rolling averages.
> Reusable across models
> Stored in the feature store

2️⃣ 𝗠𝗼𝗱𝗲𝗹-𝗗𝗲𝗽𝗲𝗻𝗱𝗲𝗻𝘁 Transformations, for example feature normalization
> Specific to one model
> Applied in both training and inference

3️⃣ 𝗢𝗻-𝗗𝗲𝗺𝗮𝗻𝗱 Transformations
> Require real-time data
> Used in online inference

Once you understand how and 𝗪𝗛𝗘𝗥𝗘 your data transformation happens, you are in a good position to start building ML software that works.

If you want to learn more about the taxonomy of data transformations read this excellent blog post by the great @jim_dowling

> 🔗 https://t.co/q7gTBCbxag

----
Hi there! It's Pau Labarta Bajo 👋

Every day I share free, hands-on content, on production-grade ML, to help you build real-world ML products.

𝗙𝗼𝗹𝗹𝗼𝘄 𝗺𝗲 so you don't miss what's coming next

[Jim Dowling @jim_dowling]

@Javierdlrm presented the 1st feature store system paper at a top tier conference - SIGMOD. Highlights:
* read data from lakehouse to Python : Hopsworks is 10-57X higher throughput than Databricks, AWS Sagemaker, and GCP Vertex
https://t.co/49n7FRMxYM

[Jan Curn @jancurn]

On Thursday night, @apify was awarded by @Deloitte's #fast50 and @googlecloud as the CE Tech Rocketship: "This special recognition awards companies that have developed high-quality, scalable solutions with the potential to serve the most demanding customers in the market."

We're very grateful for the recognition and will keep working on our growth rate to rank well in the main category next year :)

Big thanks to the Deloitte team for organizing this competition and the wonderful related events.

[Jim Dowling @jim_dowling]

Proud of the team delivering >10X improvement in reading from @hopsworks feature store. Arrow all the way from the lakehouse to Pandas with @duckdb and #arrowflight (@VoltronData) gives us an 8X-45X better performance than Vertex, Databricks, Sagemaker https://t.co/JR6NLvRc7j

[Elizabeth Daly @elizabethmdaly]

Delighted to announce my team at @IBMResearch has an IBM-UCD fellowship PhD opening on human-compatible AI systems with @brianmacnamee and @ml_labs_irl. Fantastic opportunity to get the best of both academic and industry research https://t.co/2OxRlxe5cp

[Cat Noone @imcatnoone]

Hey Twitter! Would any folks living in🇨🇦Canada, 🇮🇪Ireland, or 🇵🇹Portugal with a disability (or with someone who has), be it physical or mental, be up for a chat?

Plz RT or share!

[Linus ✦ Ekenstam @LinusEkenstam]

Honestly so fucking fed up with @RevolutBusiness & @RevolutApp - Been a customer since the very beginning. Had a really nice sales call early on with someone selling me the first business account deals.

I'm locked out for 6 months, without access to funds.

A thread.

[⭐ amy o'connor ⭐ @amyohconnor]

Good evening!

It’s that special time of the year when I round up some lovely/heartwarming/moving/daft/silly things that happened during the year. If ever there was a year where we needed to look for the positives, I think it’s this one.

Let’s go! 💕

[jungly @jungly]

Hey @lenovoitalia @Lenovo, I ordered a laptop on July 27. Two weeks and the laptop still hasn't been shipped. It was supposed to shipped in seven days. Customer service said I might get an update in three days! This is hurting my work!

[Emmanuel Acho @EmmanuelAcho]

Dear white people,
For days you’ve asked me what you can do to help. I’ve finally found an answer.

Let your guard down and listen.

[MachinePix @MachinePix]

Painting Lissajous patterns with a pendulum, by Cole Newman.

[Scott Lynch @scottlynch78]

Reflecting on the fact that Blofeld's base in DIAMONDS ARE FOREVER had an "If in doubt, ASK" sign posted in the control room. So few megalomaniacs get this! Whether your base is an oil rig or a volcano, empowering your minions via an atmosphere of mature inquiry is so important.

[Motherfoclóir: The Legacy @Motherfocloir]

I hope you all had a peaceful May Bank Holiday Weekend

I'll tell you who didn't have a peaceful May Bank Holiday Weekend - the people of Trim in 1998

22 years ago a strange but terrible gang war involving Boyzone's Mikey Graham threatened to take over the town

[Rob 🚀 @RJSzczerba]

I had to watch this a few times before I realized what was happening 🤣🤣🤣 https://t.co/yEbVM3MIOW

[Philip Nolan @philipnolan1]

This man speaks for the nation.

[Cormac O'Malley @cormacpro]

Sure what else would you be at?

[Jeremy Howard @jeremyphoward]

Data science challenge: find all-cause weekly mortality for as many countries as you can. Plot the difference between 2020 and 2019 deaths (y axis) for each week (x axis) for each country (separate series).

This would be our best guess as to covid-19 related deaths.