Santiago Profile picture
Jun 21 5 tweets 2 min read Twitter logo Read on Twitter
I just spoke to myself, and it was weird.

You can now clone your voice using a few seconds of audio, give the robot personality, and deploy a clone of yourself.

Here is a step-by-step process on how to do this in less than two minutes:
I'm using @myshell_ai:

1. Record 60 seconds of audio using your phone
2. Focus on sound quality. Avoid background noises
3. Go to app.myshell.ai/robot-workshop
4. Click on Create a Robot and give it a name
5. Upload your audio. Click on Clone

It will take a few seconds to complete.
When you are done, you'll have a robot mimicking your voice.

But this is just the start.

You can now program its personality using a short description. The platform will take your instructions and generate a lengthy prompt to give the robot some "life."

Feel free to chat with yourself.

I did. It was weird.

It's difficult for me to understand the future ahead of us. These robots will be everywhere!

As technology improves, it will be hard to tell these robots apart from real people.
Thanks to @myshell_ai for partnering with me and showing me how to create one of these robots.

You should definitely try them out: app.myshell.ai/robot-workshop.

It won't cost you anything, and you'll probably freak out a little.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Santiago

Santiago Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @svpino

Jun 22
This new library will 2x your Machine Learning workflow.

It's open-source, and in 30 seconds, it'll find problems with your model you didn't know you had.

I've tried doing this by hand. It takes weeks.

Read on for the details:
The library is Giskard.

It's fully open-source. Star the repository here: .

Attached you can see a report from Giskard.

With a few lines of code, you can get information that would take a lot of work to produce manually. https://t.co/wW4S9cnyVbgithub.com/Giskard-AI/gis…
Giskard works with any Python model, including PyTorch, TensorFlow, HuggingFace, Scikit-Learn, XGBoost, and LangChain.

Some of the things that will detect for you:

• Performance biases
• Data leakage
• Spurious correlations
• Overconfidence issues
• Underconfidence issues
Read 5 tweets
Jun 20
Building AI applications will be one of the most crucial skills for the next 20 years.

If I were starting today, I'd learn these:

• Python
• OpenAI API
• Langchain

Here is the most comprehensive, free Langchain certification that you'll find online:
@activeloopai reached out to share a course they built.

With 50+ lessons and 10+ practical projects, from automated sales to customer support agents, there's plenty of practice.

The first course: "LangChain & Vector Databases In Production."

Enroll at learn.activeloop.ai. Image
The team asked for my feedback, and I had a chance to review it in detail.

It's good!

There is around 5 days' worth of learning material! Each of the 8 modules comes with quizzes, and you get a certificate at the end from the creators of the course.
Read 7 tweets
Jun 16
I accidentally discovered an undocumented behavior from the new GPT models:

[This is NOT SAFE! DO NOT do this at home!]

The attached code works, but it shouldn't. What are "compute_average" and "numbers," and where are they coming from?

Read on for the details: Image
I'm asking the model to "call a Python function."

Surprisingly, the output is a "function_call" request named "python," and the arguments contain the code to solve the problem.

This is unexpected. That's not the way OpenAI describes how "function calling" works. Image
OpenAI introduced function calling earlier this week, but nothing in their documentation discusses this behavior.

In fact, for this to work, I had to specify a fake function as part of the request.

If I don't do that, the model works as expected.

Code: gist.github.com/svpino/e6748c5… Image
Read 4 tweets
Jun 16
This picture doesn't have a soul.

She is lifeless, dead. She is just a bunch of pixels generated by Artificial Intelligence.

These models are getting good, but they can't compete with us: we have something they don't have.

We have a soul. Image
This doesn't mean generative AI is dead.

Not at all!

In the next few years, the way we produce content will change dramatically thanks to AI.

And here is one of the most extraordinary innovations I've seen that will help that transition:

Remember crypto?
@monsterapis built a platform that uses the GPU power from thousands of crypto-mining rigs to run generative AI models.

Instead of wasting it on mining shitcoins, we can now use a massive network of distributed GPUs!

This is awesome!
Read 5 tweets
Jun 14
Machine Learning can make you a ton of money, but that's just the start.

Companies are cutting massive checks to those with the skills to build end-to-end production systems.

The good news?

Very few people in the world know how to do this.

Here is how you can change that: Image
Next week, I'm running the third cohort of the Machine Learning School.

9 hours of live content where you'll learn to train, tune, evaluate, deploy, and monitor machine learning models.

The ml.school is a community with 400+ professionals just like you.
You will learn how to build a production-ready system, starting from scratch and ending with the attached pipeline.

This will be a challenging course.

If you are a software developer or a data scientists, this course is for you.

But the cohort is just the start. Image
Read 5 tweets
Jun 13
OpenAI just introduced Function Calling.

This is a killer feature! It's the most consequential update to their API since they released it.

Here is an example to show you how powerful this is:
Look at this question:

"What's the result of 22 plus 5 in decimal added to the hexadecimal number A?"

It's a complex question requiring adding numbers in different formats.

To solve it, we can combine functions with prompts. Here is what it looks like:
I'm going to define two functions.

1. One that knows how to add two decimal numbers.
2. The other knows how to add two hexadecimal numbers.

You define these functions as part of the Chat Completion call: Image
Read 8 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(