HackerTrans
TopNewTrendsCommentsPastAskShowJobs

j6m8

no profile record

Submissions

Based base64 (now with more steganography)

blog.jordan.matelsky.com
1 points·by j6m8·4 ay önce·0 comments

Ask HN: What do you want in a guitar tablature/chord markup language?

1 points·by j6m8·6 ay önce·0 comments

Show HN: Blueblocks, a letter-packing word game

blueblocks.jordan.matelsky.com
3 points·by j6m8·9 ay önce·0 comments

Show HN: Epistolary – Respond to your emails in handwriting

github.com
81 points·by j6m8·2 yıl önce·22 comments

Show HN: SQLite-like query tool for graphs

github.com
3 points·by j6m8·2 yıl önce·0 comments

Empirical influence functions to understand the logic of fine-tuning

arxiv.org
1 points·by j6m8·2 yıl önce·0 comments

comments

j6m8
·6 ay önce·discuss
https://jordan.matelsky.com
j6m8
·7 ay önce·discuss
This is neat — I do think this is relevant to more than just the software engineering space. See also, healthcare and law (I wrote more at length here, not to derail this comment thread [1]). Our junior training on-ramps for a lot of knowledge-work fields are in some semblance of equilibrium, but it's an unstable one.

[1] https://blog.jordan.matelsky.com/AI-doctors-bum-me-out/
j6m8
·geçen yıl·discuss
To those who are interested in reading more, I recommend Mary Roach's "Stiff" (and all her other work too!) highly highly highly!
j6m8
·geçen yıl·discuss
you're welcome to use/snag any of it you like :) planning to keep that repo growing, so if you think of good synergies for us, I'd love to keep making useful stuff!
j6m8
·geçen yıl·discuss
you could try my https://github.com/j6k4m8/goosepaper :)
j6m8
·2 yıl önce·discuss
no worries, good to know this would be a useful feature! I'll add it to my backlog.

    pip install 'git+https://github.com/j6k4m8/frof/'
and then

    frof myfile.frof
should work!

Was thinking about rewriting it in Go recently... :)
j6m8
·2 yıl önce·discuss
here they'll probably be executed simultaneously, since they both have zero dependencies and the machine can run multiple jobs at the same time. (can be disabled with `--max_jobs=1` or `-p=1`).

Here's another illustrative example:

    A -> B
    B -> C
    Z -> C
In this situation, frof will schedule `Z` to run in a parallel thread ASAP, so it will likely run alongside A... and if Z takes longer to run than A, Z will continue running when A stops and B starts. But C will wait for all other jobs to finish before it can schedule.
j6m8
·2 yıl önce·discuss
You write the file and ALL steps are run in topological order so that a job never runs until its dependencies have run. i.e., in a tool I'll have `build.frof` as a separate frof file than `download-dependencies.frof`, perhaps. (If your preference is that those belong in the same file I'd be down to have PRs that support that! Should be very easy, I'm happy to try implementing this if there's interest.)

So for a file with those contents called `mygraph.frof`, you can (after installing) run `frof mygraph.frof` to kick off the jobs in the current shell (inheriting env vars etc).

[edit] maybe a clarifying example here: https://blog.jordan.matelsky.com/frof-render/
j6m8
·2 yıl önce·discuss
I wrote frof [1] for exactly this purpose :)

Designed to be ultra-simple and with minimal "config-file acrobatics".

It looks like this [edit, formatting]:

    write -> analyze
    build -> analyze

    write:     echo 1 2 3 > data.txt
    build:     compile_tool.sh > tool.sh
    analyze:   tool.sh data.txt
https://github.com/j6k4m8/frof/
j6m8
·2 yıl önce·discuss
That's a question more suited to a microbiologist or bacteriologist than to me, but my educated guess, at least in the electron microscopy case, is that you'll see the bacteria inside the depth of the slices, rather than sitting "atop" the slices. i.e., if you cut open an apple and find half a worm, the worm was in the apple. If you cut open an apple and then see a worm on top of the slice, it's possible it arrived post-cut.
j6m8
·2 yıl önce·discuss
Very hard to clean a blender!

More nitpickfully, one of the big things we care about is if the bacteria are living _harmlessly_ in the brain. i.e., site of microbes, and a lack of inflammation, will answer more than just "are there microbes around".
j6m8
·2 yıl önce·discuss
Microbes are CRAZY. They're everywhere. Thermal vent-friendly microbes. Space-friendly microbes. Vacuum-resilient, heat-resilient, acid-resilient. Microbe-free-environment-friendly microbes [1]. It seems hard to imagine that a blood-brain barrier could really keep the brain sterile.

We're lucky to live in a scientific era during which a "gut microbiome" is taken for granted (heck, even FDA-approved treatments depend on it! Google FMT, but don't click "images" from your work laptop), but it wasn't so long ago that we felt microbes were unlikely to live endogenously and harmlessly anywhere in the body.

There were also some hypotheses (untested, if memory serves) that COVID-19 influenced olfactory neurons through direct infection. Don't tell the blood-brain barrier, but if I were a bacterium, the nasal palate would be my ingress strategy. Or maybe the gums or gut — one of the cranial nerves, certainly. [edit] I should clarify — covid is viral, not bacterial, but it does show that this is a potential entry vector.

The central nervous system is incredibly complicated, and our symbiotic relationship with microbes is extraordinary. I think it does a disservice to bacteria to suppose they DON'T get involved in an organ :)

[1] https://www.space.com/ryugu-asteroid-sample-earth-life-colon...
j6m8
·2 yıl önce·discuss
A few years ago my team mounted what I think was the largest-(to-date) scale search for this in electron microscopy brain tissue volumes [1].

I STRONGLY believe there is a substantial central nervous system microbiome, but (spoiler alert) no evidence found in that search :)

If you're excited about this work, the datasets are all freely available from BossDB [2] — well over a dozen petavoxels of it! I'd be so curious if models these days could pick up on something we missed!

[1]: https://www.biorxiv.org/content/10.1101/2022.07.12.499807v1 [2]: https://bossdb.org
j6m8
·2 yıl önce·discuss
raw data is O(petabytes) (single-digit); synapse-neuron graph will be probably order 100GB. But you also want morphology and locations, since it's not enough to just say "X connects to Y" if you want to know about dynamics!

i'm not hosting this dataset specifically, but check out https://bossdb.org/. my disclaimer and also my brag is that this is my job and research area :) if you're looking for a copy, let's talk! there are easy ways and hard ways :)
j6m8
·2 yıl önce·discuss
preprint coming out soon about this specifically :)

in the meantime, here's a simple tool paper we wrote explaining how you can treat this like a cool graph database challenge [1] and a preprint showing how you could approach that question when your number of samples per animal is close to N=1 [2]. basically..... it's hard! but also.... it's cool!

[1]: https://www.nature.com/articles/s41598-021-91025-5 [2]: https://www.biorxiv.org/content/10.1101/2023.10.16.562590v1....
j6m8
·2 yıl önce·discuss
The reMarkable company has been super adversarial to a lot of these tools, and the file standards and API have been moving goalposts for years. MOST of the tools on that Awesome list are defunct because the primary open source tools for getting data to the reMarkable cloud (rmapi and rmapy) are no longer maintained — the primary maintainers both cite reMarkable's moving target API as the final dealbreaker. SUPER sad.

I've been hoping to write my own now that the dust has settled, but it's definitely a MAJOR project yet to be done by the FOSS community.
j6m8
·2 yıl önce·discuss
ha — that's exactly what I've been working on on a separate reMarkable project posted earlier today [1] :)

[1] https://news.ycombinator.com/item?id=41437740
j6m8
·2 yıl önce·discuss
You can write software for it but reMarkable as a company has been downright adversarial to the open source community and MOST of the tools on that Awesome page are now defunct because reMarkable has obfuscated API endpoints or changed file-standards to prevent third party efforts.
j6m8
·2 yıl önce·discuss
Totally, the reMarkable has handwriting recognition as well — this project was partially borne of the workflow of "write text, auto transcribe, send to a computer, copy paste into an email." How do you like the Boox? My friend loves his!
j6m8
·2 yıl önce·discuss
ha, I learned shorthand in second grade (and yes, it was already an antique by then). I wonder what a multimodal language model would say to shorthand...