davnn
davnn on Hacker News
Posted by
davnn on Sunday September 27, 2026 @01:20PM
from the show-hn dept.
https://github.com/getgat-dev/gat
Hi HN, I built Gat because I wanted one thing from data versioning: check out an old Git commit and recover the datasets, model weights, or assets that belong to it. It's basically a lock file for large files.I’d appreciate any feedback!
Do we know if we are bunch of dynamically adjusting weights?
Posted by
davnn on Tuesday September 15, 2026 @07:01AM
from the fastpaced.com dept.
Posted by
davnn on Tuesday September 08, 2026 @07:52AM
from the show-hn dept.
https://getgat.dev
Hi HN — I was frustrated for a long time using Git LFS and DVC to version large files.
Gat is the boringly simple tool I was looking for: Git manages versioning and Gat manages storage.
No dedicated LFS server, no clean/smudge filters, just a lock file.The workflow is small: `gat add` some files, commit `gat.lock` with Git, then `gat push` the content to your storage.
In a fresh clone, run...
I really appreciated his coding-style, but the bar is quite low on research/ML-algorithms to be fair. I still wonder how he managed to get „trending“ repositories regularly despite the repositories being empty.
What was happening with this account? I was often seeing popular but empty (only title of the paper and maybe a short readme) repositories that were created directly after a paper was published?
Do you have a link to share? Would be highly interested!
I would say uv provides quite a lot of additional features that can be used in various ways to create plain-old venvs. Note, however that uv-pack can also pack a subset of your uv-monorepo for a specific package (there are still some quirks I have to admit..).
My experience was that it‘s surprisingly painful to „just copy“ a venv and especially a uv-created venv. There are a lot of paths to be modified to get the venv working. Copying a venv felt hacky and wrong, that‘s why I built the tool :)
Posted by
davnn on Sunday January 25, 2026 @01:56PM
from the show-hn dept.
https://github.com/davnn/uv-pack
I kept running into the same problem: modern Python tooling, but deployments to air-gapped systems are a pain. Even with uv, moving a fully locked environment into a network-isolated machine was no fun.uv-pack should make this task less frustrating. It bundles a locked uv environment into a single directory that installs fully offline—dependencies, local packages, and optionally a portable...
Posted by
davnn on Thursday October 23, 2025 @05:26PM
from the fastpaced.com dept.
I am using runtime type and shape checking and wrote a tiny library to merge both into a single typecheck decorator [1]. It‘s not perfect, but I haven‘t found a better approach yet.
[1] https://github.com/davnn/safecheck
There appears to be (almost) no true competition in healthcare, therefore no real incentives to improve productivity. Wages in healthcare rise disproportionately without productivity gains (Baumols disease); why invest in digitalization?
> The key insight is that smaller, well-bounded changes are exponentially easier to review thoroughly.
I am not sure if that is the real insight. It appears to me that most people prefer small, well-bounded changes, but it's quite tricky to break down large tasks into small but meaningful changes, isn't it? To me, that appears to be the key.
That's basically where I am coming from :). I know about my addiction.
Working in the ML field, I can't hate Python. But the type system (pre-3.12, of course) cost me a lot of nerves. Hoping for a better post-3.12 experience once all libraries are usable in 3.12+. After that experience, I’ve come to truly appreciate TypeScript’s type system. Never thought I’d say that.
That‘s actually quite similar to the nearness library [1]. The main difference appears to be vicinity‘s focus on simplicity while nearness tries to expose most of the functionality of the underlying backends.
[1] https://github.com/davnn/nearness
That's the point I was trying to make, which is also why I brought up 'pseudo-local' development environments in the article.
Posted by
davnn on Sunday October 06, 2024 @07:01AM
from the fastpaced.com dept.
Maybe you refer to [1]? These are the two options I like to show students who first get in touch with data visualization.
[1] https://www.data-to-viz.com/
Note: In the about page, they provide an extensive list of historical chart classification schemes :).
ChatGPT answer: There are *two* "R"s in the word "strawberry."
There are libraries that support shape checking and I‘ve written a package to combine beartype’s runtime typechecks with jaxtyping‘s shapechecks, see
https://github.com/davnn/safecheckOnce accustomed to shape checking it‘s quite a boost in productivity for us, no more fiddling around with invalid dimensions.
Is 5k an appropriate amount for such a finding? Sounds incredibly cheap for such a large organization. How much would something like this be worth on the black market?
> To me it seems like it should be a top priority.
Maybe most people are content with children, the current state of the art against aging communities?
Not yet, I believe. Revenue from iPhone sales is still quite fundamental to Apple‘s success, it‘s more than triple the revenue from all services combined (not including Google‘s search engine deal).
Python as a glue language (as it‘s mostly used in data intensive applications) for something like MPI should not add too much overhead?
Can you expand on your usage of AI tech?
I would be highly surprised if LLMs would NOT be used as individual math tutors. In fact ChatGPT would likely already be very helpful.