BadInformatics

10 ms·

91.
by BadInformatics 6y ago
That's great to hear. I'd love to see the Bazel monstrosity that is Tensorflow XLA get easier to build because of Nx. Not holding my breath though!
92.
by BadInformatics 6y ago
Pretty disappointing to see it using XLA. Anything Tensorflow is notoriously difficult to build on a local machine and generally considered a dependency nightmare. I would've thought libtorch bindings would be much easier.
93.
by BadInformatics 6y ago
I know of at least one Canadian hospital that's incorporated ML into 100% of their ED triage. Sure it's not some state of the art deep learning architecture, but it's definitely a step above the old crop of heuristic-based sy
94.
by BadInformatics 6y ago
Don't get me wrong, Python is head and shoulders above Java/C# or C/C++ as a teaching language. The complaint is that, as the educator, lab/tutorial time is being wasted on trivial syntax issues that would be non-existen
95.
by BadInformatics 6y ago
Absolutely, I find this approach gets the right mental gears turning as well. The problem comes up most often with take-home assignments though, where students are more likely to copy-and-paste a couple lines from somewhere because you have
96.
by BadInformatics 6y ago
No, though I have plenty of anecdotes from friends who have. I agree that these apply regardless of language. Whitespace errors are arguably some of the more insidious syntax errors though, since most editors will not differentiate between
97.
by BadInformatics 6y ago
I agree WRT block structure, but having TA'd a Python course that "should" doesn't survive contact with a real world classroom. Students will use anything from IDLE to Notepad to PyCharm, and usually shut down completely
98.
by BadInformatics 6y ago
IMO the bundling and curation is the least interesting part of Conda. What's really nice are the build infrastructure, quality checks and proper dependency bounds (looking at you, PyPI). Something like R's CRAN already has those w
99.
by BadInformatics 6y ago
I remember a few years back when Nim was going to be the next big thing here, but unfortunately it never gained that level of traction. Still an interesting language even without bigco levels of support behind it.
100.
by BadInformatics 6y ago
The type system is a lot less sound and expressive than even TypeScript's though, so it doesn't quite offer the same level of security. I find any doesn't tend to stick around long when I use TypeScript, but my Python project
101.
by BadInformatics 6y ago
This is my biggest pet peeve with (dynamic) languages that don't distinguish between declaration and assignment. Python's drive to make type annotations look "nicer" certainly doesn't help either. `list[1, 2, 3]` fe
102.
by BadInformatics 6y ago
I used it for a couple WPF apps (of all things) a couple years back. F# is so nice for anything with a good amount of "business logic".
103.
by BadInformatics 6y ago
The FAQ [1] and this SO answer [2] explain it better than I can, but basically yes. However, the (de)serialization overhead is probably better than most alternative formats you could save to. [1] https://arrow.apache.org/faq
104.
by BadInformatics 6y ago
Not sure I follow, that page indicates that JS support is pretty good for all but the more obscure features (e.g. decimals) and doesn't mention data visualization at all? Anyhow, I've successfully used https://github.co
105.
by BadInformatics 6y ago
Is that a problem in the Parquet format or in PyArrow? My understanding is that Parquet is primarily meant for on-disk storage (hence the default on-disk compression), so you'd read into Arrow for in-memory compression or IPC.
106.
by BadInformatics 6y ago
HDF5 is pretty terrible as a wire format, so it's not a 1-1 comparison to Arrow. Generally people are not going to be saving Arrow data to disk either (though you can with the IPC format), but serializing to a more compact representati
107.
by BadInformatics 6y ago
https://arrow.apache.org/docs/js/ has existed for a while and uses typed arrays under the hood. It's a bit of a chunky dependency, but if you're at the point where that level of throughput is required bu
108.
by BadInformatics 6y ago
There are a fair few MPV GUIs out there. I believe Celluloid is available in the main Debian repos.
109.
How Do We Trust Our Science Code?(hillelwayne.com)
3 points by BadInformatics 6y ago|1 comments
110.
by BadInformatics 6y ago
Having helped to clean up a couple of those messes IBM left behind in various parts of the Canadian public sector, the only silver lining I see is that governments are starting to take in-house dev talent and the digital service model serio
111.
by BadInformatics 6y ago
In my limited experience and anecdata, governments have been so trigger happy in getting rid of and/or marginalizing technical staff that there are few qualified individuals in the procurement office who can properly draft RFPs and eva
112.
What engineering can teach (and learn from) us(hillelwayne.com)
2 points by BadInformatics 6y ago|0 comments
113.
by BadInformatics 6y ago
For sure, and one would be remiss not to mention Dylan, CL/CLOS and Clojure here as well. My quibble was with the claim that multiple dispatch rarely shows up in practice, which you've pretty clearly shown is not the case in R!
114.
by BadInformatics 6y ago
If NLP primitives are all that's keeping you from testing the waters, have a look at https://github.com/chengchingwen/Transformers.jl .
115.
by BadInformatics 6y ago
> is this really a problem that I'm going to come across in the real-world when 90% of our time is spent ingesting a poorly-formatted csv, doing some quick plots and perhaps building a model to test something out Yes, multiple dispa
116.
by BadInformatics 6y ago
> Julia is a growing alternative to Fortran/C/etc for long-running computations, it remains awkward and unpleasant for interactive analysis Did you mean to write short-running scripts? If anything, the Julia dev workflow is bia
117.
Testing out Algebraic effects in OCaml for Animations(gopiandcode.uk)
1 points by BadInformatics 6y ago|0 comments
118.
by BadInformatics 6y ago
For better or worse, most of these articles are US-centric because that's where most of the R&D money for health ML is. The far more reasonable approach is to make sure your data contains as many demographics as possible (not just
119.
by BadInformatics 6y ago
I'm of two minds about this article. It does a reasonable job of enumerating the issues with naively deploying ML in a healthcare setting. However, these articles are becoming a dime a dozen and there is little actionable talk on how
120.
by BadInformatics 6y ago
I think it's a timing thing. There was an earlier post with the Anandtech article late yesterday, but because so much of HN is in NA timezones it was likely smothered by the morning. Love the TechTechPotato breakdown of this BTW :)

More ›