Albert Thomas
@albertcthomas.bsky.social
Research engineer at Huawei. https://albertcthomas.github.io/blog/
Reposted by Albert Thomas
That was incredibly relieving to me to read “motivation was problem-solving, machine learning as a means to an end” in @lawrennd.bsky.social “Atomic human”.
I always felt as an impostor among colleagues who want to solve intelligence, while I just enjoy working on cool stuff.
I always felt as an impostor among colleagues who want to solve intelligence, while I just enjoy working on cool stuff.
June 14, 2025 at 8:45 AM
That was incredibly relieving to me to read “motivation was problem-solving, machine learning as a means to an end” in @lawrennd.bsky.social “Atomic human”.
I always felt as an impostor among colleagues who want to solve intelligence, while I just enjoy working on cool stuff.
I always felt as an impostor among colleagues who want to solve intelligence, while I just enjoy working on cool stuff.
Reposted by Albert Thomas
Many companies won't use Chinese *open* models for long-tail information and generated code security concerns - a major adjustment in how I see the open model ecosystem. Adoption is on the table for entrants amid DeepSeek and Qwen releasing some of the best models on paper.
buff.ly/y6MMoBt
buff.ly/y6MMoBt
What people get wrong about the leading Chinese open models: Adoption and censorship
Narrative violations on licenses, adoption, and censorship.
www.interconnects.ai
May 6, 2025 at 1:46 PM
Many companies won't use Chinese *open* models for long-tail information and generated code security concerns - a major adjustment in how I see the open model ecosystem. Adoption is on the table for entrants amid DeepSeek and Qwen releasing some of the best models on paper.
buff.ly/y6MMoBt
buff.ly/y6MMoBt
TIL about huggingface-cli delete-cache to clean the models or other resources (datasets, ...) you downloaded from huggingface albertcthomas.github.io/blog/removin...
Albert Thomas
albertcthomas.github.io
May 6, 2025 at 11:32 AM
TIL about huggingface-cli delete-cache to clean the models or other resources (datasets, ...) you downloaded from huggingface albertcthomas.github.io/blog/removin...
« When software is open-source, it means it is open-source – that the source is open – nothing more. […]
It does not mean open to contributions;
It does not mean support is offered;
It does not mean you’re entitled to feature requests;
It does not mean the developer owes you their time;
[…] »
It does not mean open to contributions;
It does not mean support is offered;
It does not mean you’re entitled to feature requests;
It does not mean the developer owes you their time;
[…] »
Discussion on the misconceptions of open-source software and the entitlement some users feel towards developers and their contributions.
By @vale.rocks
Found via @erogol.com
vale.rocks/posts/open-s...
By @vale.rocks
Found via @erogol.com
vale.rocks/posts/open-s...
Open-Source is Just That
Don't bite the hand that feeds you.
vale.rocks
April 12, 2025 at 7:58 AM
« When software is open-source, it means it is open-source – that the source is open – nothing more. […]
It does not mean open to contributions;
It does not mean support is offered;
It does not mean you’re entitled to feature requests;
It does not mean the developer owes you their time;
[…] »
It does not mean open to contributions;
It does not mean support is offered;
It does not mean you’re entitled to feature requests;
It does not mean the developer owes you their time;
[…] »
Really enjoying the series of posts by @beenwrekt.bsky.social on overfitting.
On why it’s time to remove the bias-variance tradeoff from machine learning 101.
Overfitting to theories of overfitting
On a plot that radicalized me like no other.
www.argmin.net
February 18, 2025 at 9:39 AM
Really enjoying the series of posts by @beenwrekt.bsky.social on overfitting.
Reposted by Albert Thomas
Just put on line a talk I gave summarizing what I have learned across the years as a maintainer of open source.
It's _opinions_ (been there, done that), but I'm willing to defend them, having stewarded my share of successful open source projects.
speakerdeck.com/gaelvaroquau...
It's _opinions_ (been there, done that), but I'm willing to defend them, having stewarded my share of successful open source projects.
speakerdeck.com/gaelvaroquau...
Open source software: how to live long and go far
An opinionated guide to building open-source software tools
with a focus on Python and science
A talk that I gave when I was stepping down as a lead…
speakerdeck.com
February 6, 2025 at 8:31 PM
Just put on line a talk I gave summarizing what I have learned across the years as a maintainer of open source.
It's _opinions_ (been there, done that), but I'm willing to defend them, having stewarded my share of successful open source projects.
speakerdeck.com/gaelvaroquau...
It's _opinions_ (been there, done that), but I'm willing to defend them, having stewarded my share of successful open source projects.
speakerdeck.com/gaelvaroquau...
Reposted by Albert Thomas
Good, published, benchmarks of machine learning / data science is crucial.
But so hard.
Well-cited "SOTA" methods typically crash often. They tend to be very computational expensive. Both make a systematic study impossible.
Finally, reviewers always ask for more methods, and more "SOTA".
But so hard.
Well-cited "SOTA" methods typically crash often. They tend to be very computational expensive. Both make a systematic study impossible.
Finally, reviewers always ask for more methods, and more "SOTA".
December 1, 2024 at 4:54 PM
Good, published, benchmarks of machine learning / data science is crucial.
But so hard.
Well-cited "SOTA" methods typically crash often. They tend to be very computational expensive. Both make a systematic study impossible.
Finally, reviewers always ask for more methods, and more "SOTA".
But so hard.
Well-cited "SOTA" methods typically crash often. They tend to be very computational expensive. Both make a systematic study impossible.
Finally, reviewers always ask for more methods, and more "SOTA".
Reposted by Albert Thomas
Game on! 👾 for @scikit-learn.bsky.social
experts only: the ✨boss level✨ has arrived 🚀
For seasoned pros ready to master ML:
🔹 Custom algorithms
🔹 MLOps & deployment
🔹 Align ML with business projects
Be among the first to get certified! 👉https://eu1.hubs.ly/H0dZ18x0
#machinelearning #datascience
experts only: the ✨boss level✨ has arrived 🚀
For seasoned pros ready to master ML:
🔹 Custom algorithms
🔹 MLOps & deployment
🔹 Align ML with business projects
Be among the first to get certified! 👉https://eu1.hubs.ly/H0dZ18x0
#machinelearning #datascience
November 26, 2024 at 7:56 AM
Game on! 👾 for @scikit-learn.bsky.social
experts only: the ✨boss level✨ has arrived 🚀
For seasoned pros ready to master ML:
🔹 Custom algorithms
🔹 MLOps & deployment
🔹 Align ML with business projects
Be among the first to get certified! 👉https://eu1.hubs.ly/H0dZ18x0
#machinelearning #datascience
experts only: the ✨boss level✨ has arrived 🚀
For seasoned pros ready to master ML:
🔹 Custom algorithms
🔹 MLOps & deployment
🔹 Align ML with business projects
Be among the first to get certified! 👉https://eu1.hubs.ly/H0dZ18x0
#machinelearning #datascience
Reposted by Albert Thomas
Here’s a little script I made which I use to get a server up and running automatically (after you answering a few questions, including “what’s your name”) in just a few minutes.
You can even fully automate it with a few environment variables.
github.com/AnswerDotAI/...
You can even fully automate it with a few environment variables.
github.com/AnswerDotAI/...
github.com
November 26, 2024 at 9:35 AM
Here’s a little script I made which I use to get a server up and running automatically (after you answering a few questions, including “what’s your name”) in just a few minutes.
You can even fully automate it with a few environment variables.
github.com/AnswerDotAI/...
You can even fully automate it with a few environment variables.
github.com/AnswerDotAI/...
Totally agree. This is a great paper that everyone doing RL should read.
Adding my love letter to
arxiv.org/pdf/2304.01315
Empirical Design in Reinforcement Learning
by
Andrew Patterson, Samuel Neumann, Martha White, Adam White
JMLR 25 (2024) 1-63
#ReinforcementLearning
These aren’t the heroes we deserve, but they are the heroes we need.
arxiv.org/pdf/2304.01315
Empirical Design in Reinforcement Learning
by
Andrew Patterson, Samuel Neumann, Martha White, Adam White
JMLR 25 (2024) 1-63
#ReinforcementLearning
These aren’t the heroes we deserve, but they are the heroes we need.
arxiv.org
November 23, 2024 at 7:45 PM
Totally agree. This is a great paper that everyone doing RL should read.