sandkoan

9 ms·

61.
by sandkoan 4y ago
I'd honestly argue that he might not have even needed OpenAI embeddings—any off-the-shelf Huggingface model would've sufficed. Because of attention mechanisms, we no longer so heavily depend on the existence of those "pointy
62.
by sandkoan 4y ago
See also Popper's paradox of tolerance: https://en.wikipedia.org/wiki/Paradox_of_tolerance
63.
by sandkoan 4y ago
Huh, never really noticed that on my end. I'm curious as to what question you asked....
64.
Show HN: Converse with Books as If They Were People(socrates.govindgnana.com)
11 points by sandkoan 4y ago|2 comments
65.
by sandkoan 4y ago
If the rewind feature set—especially the search through screen recording aspect—is the most compelling, then perhaps something I've been working on may be of use. It can basically do everything rewind can (though audio support is a WIP
66.
by sandkoan 4y ago
I've been working on a semantic search engine for code that's particularly useful to those navigating unfamiliar codebases. Just write an English description of whatever you're looking for, and it'll show you the most re
67.
by sandkoan 4y ago
No, the basic model already exists. You could throw together something like this yourself: encode your documents with a Transformers model into a series of vectors. Then you're merely a nearest neighbor search away from finding the mos
68.
Meditations on the Devil and Deep Sea Dilemma(blog.govindgnana.com)
2 points by sandkoan 4y ago|0 comments
69.
by sandkoan 4y ago
For a distributed computing/BitTorrent-style method of running these LLMs, see: https://github.com/bigscience-workshop/petals .
70.
by sandkoan 4y ago
I've actually been building something like this: a desktop app that basically records everything you see and hear, and makes it searchable. Maybe we could team up: my email is govind <dot> gnanakumar <at> outlook <dot&gt
71.
by sandkoan 4y ago
This might be what they're referring to: https://regex101.com/ .
72.
The Tortoise and the Hare, Revisited(blog.govindgnana.com)
1 points by sandkoan 4y ago|0 comments
73.
Git: A Generative Image-to-Text Transformer for Vision and Language(arxiv.org)
2 points by sandkoan 4y ago|0 comments
74.
Spawn, grow, sow: how coral seeding could boost recovery(aims.gov.au)
1 points by sandkoan 4y ago|0 comments
75.
by sandkoan 4y ago
I, too, am an aphant (which I believe affects only 3 to 10 percent of the population), but I understood this to still apply to us: the picture is merely replaced with the small kernel of a concept it represents.
76.
by sandkoan 4y ago
How curious: I have a small children's book I picked up at the California Academy of Sciences entitled, "The ABCs of particle physics," made by a joint collaboration between Fermilab and SLAC [1], and here I find another poem
77.
by sandkoan 4y ago
From what I can see, they download the repo to their server, generate per-function semantic embeddings (which will no doubt cost them at scale) [1], then use some combination of GPT-3 and Codex prompted with something like "Here's
78.
by sandkoan 4y ago
This is good fun. In the same spirit as something I threw together recently [1], that runs broken code (with un/misdefined functions and variables), and repairs it using a combination of stacktraces and comments fed to GPT-3. [1] http
79.
Show HN: Athelas – Automagically Repair Broken Code(huggingface.co)
8 points by sandkoan 4y ago|0 comments
80.
by sandkoan 4y ago
There might be a reason for that: https://nautil.us/issue/47/consciousness/the-kekul-problem .
81.
by sandkoan 4y ago
Looks pretty cool! I've been working on something similar, but was a bit discouraged by market saturation in the universal search space.... How does this compare to something like, say, Glean?
82.
by sandkoan 4y ago
Here's an updated link: https://huggingface.co/spaces/sandkoan/athelas . Athelas reads stacktraces and whatnot to determine how to repair broken code, and returns what it thinks the code should be, taking into
83.
Show HN: Athelas – Self-Healing Code(9a766a8e-f818-4a63.gradio.live)
5 points by sandkoan 4y ago|1 comments
84.
by sandkoan 4y ago
Woah! I've been facing the same problems with pyannote+whisper for diarization+transcription, and, coincidentally, was just experimenting with combining NeMO and whisper. Do you happen to have a repo for this? Would be invaluable. Edit
85.
by sandkoan 4y ago
For the record, Python didn't have a case or when statement. It does now, recently added to version 3.10.
86.
Concurrent Revisions(microsoft.com)
1 points by sandkoan 4y ago|0 comments
87.
by sandkoan 4y ago
Is there a particular reason that web-based 3d editors failed to find a market? Is it just that those working in the modelling space are already well-versed in actual editing tools, so they don't really have a need for something like t
88.
Ask HN: What's the most performant/practical model/API for text extraction?
1 points by sandkoan 4y ago|1 comments
89.
LLMS Can Be Poets Too: AI Writing Assistant and Constrained Text Generation(aclanthology.org)
1 points by sandkoan 4y ago|0 comments
90.
Intel Introduces Real-Time Deepfake Detector(intel.com)
3 points by sandkoan 4y ago|0 comments

More ›