aflam

First time I see a clear split where the README markets to humans and the AGENTS.md has a clean tutorial for LLMs.

I'll give it a try, MCP apps are full of promises but protocols are so unstable that I wouldn't want to write the boiler plate myself.

Neat! We started working on something similar, using LD_PRELOAD. After setting some env variables we'd see a commit's files layered at some path, on top of artifacts saved for this commit. The goal is to run jobs that expect to access both git files and build artifacts, while avoid duplication of storage. FUSE would be better but users don't have enough permissions.

Libs that use C# but nobody wants to try building on Linux, Windows binaries we get as-is from outside the organization, C++ code relying on MVSC non-standardness, C code relying on a custom toolchain that can't be ported...

Last year I wanted to have a table custom built using those shapes. It's easy with CNC: https://shapescience.xyz/blog/creating-a-super-ellipsis-tabl...

-hh is what a major app at work requires. Engineering indeed provides room for creativity.

ANN is the building block of density estimation. If I recall correctly that's the bottleneck for density-based clustering as its various algorithms then take advantage of faster yet algorithms (sort, union-find, spanning-trees..). While very interesting I am not sure this gentle introduction is the best place to discuss ANN.

B

We had issues reliably using rviz or other visualisation tools on ROS. Something will often break (opengl with docker ? outdated version of ROS or missing packages ?). There is a lot of value in having a text-based way out for simpler debugging.

Thanks for the article.I came across it as part of scikit-tda, but did not investigate. Looks promising! The TDA background is appealing, as well as runtime complexity.

Related issue: my ISP uses as default wifi password part of the router's MAC address... It could be the kind of problem described in the article - but the site seems down.

Odd indeed, especially since it has all the other useful tools one could use to create palettes. Maybe the target audience is creative people, not us just reusing ad nauseum the same effective palette.

Managing those experiences takes effort, otherwise you're sure to end up with the issues you describe. Even when the data makes sense, sometimes people hand you over an unparsable csv with their results, while the conclusions comes from a gut feeling. Sadly not everybody understands standard tools like boxplots, and analyses are often hard to reproduce.

The tooling to make it easy not there yet.

Thanks a lot for the answer! It's very promising.

It's great to see this sort of tool open-sourced. I am excited by new tools enabling better algo/ML engineering workflows.

In addition to the infra management, it's quickly tricky to support qa/viz/tuning/debugging tools for very different sorts of algorithms/outputs/configurations/metrics. Do you see your project going in those directions?

This sort of tooling is a must have the moment your project has more that a couple developers.Does anyone know of other algorithm versioning tools?

At work we built some custom infrastructure to evaluate algorithm (all kinds) performance over time, use various configurations, run comparaisons... I would like to open-source some of it, would people be interested?

Here is nice presentation covering those: (link: https://www.cs.princeton.edu/~rs/talks/AC11-Cardinality.pdf)

Exactly, that is a beautiful mitigation - as long as you can upgrade the memory controller. I really hope vendors are doing this on all new chips!

They did increase their DRAM refresh rate and mentioned it in release notes. There is not much more you can do easily - and it comes at a real cost in power usage.

https://support.apple.com/en-us/HT204934

Data science now has its own tree-sorting questions. Shameless plug, but since the submission is popular, some could find this helpful: https://shapescience.xyz/blog/interview-questions-for-data-s...

IoT does not yet seem to stand for Intel of Tomorrow! At least not when it comes to boards or processors. Maybe there are just more convinced by the potential of vertical solutions like MobilEye's?

I couldn't agree more! This course features has slides, which many will prefer.

Our company designs optical transport systems; we have a hard time en-boarding new software people because this gap in training material... I'm currently writing a blog post on optical modulation, it's difficult to find an approachable language.

The long and convoluted debugging cycle for machine learning really hurts my faith in ML models. This issue - with some practical advice - was the center of this interview (disclaimer: author). https://shapescience.xyz/blog/interview-data-science-methodo...

I'm convinced we lack decent tools for ML debugging: what could they be?

Hublot worked on special flavors of 18k gold with EPFL University and had some press a couple years ago : http://news.discovery.com/tech/unscratchable-gold-111227.htm In French : http://www.lacotedesmontres.com/actu/Hublot-presente-Magic-G...

Now dplyr has pretty much replaced it with better syntax and perf. Take a look at it. https://github.com/hadley/dplyr

Did anyone dare authorize? I find using simple commands like `du` and `tree` sufficient for me...

Entertaining. Now will the blue bird fare better than the blue chips ?