shayanjm

12 ms·

1.
by shayanjm 2y ago
Nested querying is not something that is standard for regular grammars, amongst other aspects introduced in this paper that implicitly require things like memory (again, not standard for regular grammars)
2.
by shayanjm 2y ago
At what point do we drop the term "regular" expressions altogether for stuff like this? This is going to sound pedantic since I know that most popularly-used regex implementations are themselves non-regular, but I feel like we&#x2
3.
Synthetic data generation through geometry and embeddings(watchful.io)
3 points by shayanjm 3y ago|0 comments
4.
Decoding LLM Uncertainties for Better Predictability(uncertainty.demos.watchful.io)
16 points by shayanjm 3y ago|2 comments
5.
by shayanjm 3y ago
Building off our last research post, we wanted to figure out ways to quantify "ambiguity" and "uncertainty" in prompts/responses to LLMs. We ended up discovering two useful forms of uncertainty: "Structural&quo
6.
by shayanjm 3y ago
I think the author of this post probably meant to caveat that what we call "prompt engineering" TODAY might tend towards snake oil, but prompt engineering _doesn't have to_ be snake oil, and it _doesn't have to_ promote
7.
by shayanjm 3y ago
Interestingly we initially thought that prompt length would play a big factor in the performance of this approach. In practice, though, we discovered that it's actually not as big a factor as we predicted. For instance, Prompt #3 was 4
8.
by shayanjm 3y ago
Great question - there are currently (likely) tons of limitations to this approach as-is. We're planning on testing this on more capable models (e.g: integrated gradients on Llama2) to see how the relationship might change, but here ar
9.
Show HN: A surprisingly effective way to predict token importance in LLM prompts(heatmap.demos.watchful.io)
14 points by shayanjm 3y ago|6 comments
10.
Arguments Against Hand Labeling(oreilly.com)
6 points by shayanjm 5y ago|0 comments
11.
Watchful.io – Firehose Filtration Platform(watchful.io)
29 points by shayanjm 10y ago|0 comments
12.
by shayanjm 11y ago
I came here to ask that EXACT question - fighting the good fight
13.
by shayanjm 11y ago
Off the back of a napkin - the key should never be stored anywhere first of all. In the absence of keyring/keychain/etc., it'd be trivial to introduce a masterpassword implementation in the browser client which is XOR'd
14.
by shayanjm 11y ago
As addressed in the post - there are no mitigating factors in the scenario of accidental exposure. The lowest hanging fruit would be a dumb hashing function which uses some master password. If you've been hit with an OS compromise you&
15.
by shayanjm 11y ago
Yes! There are tons of accidentally-uploaded profiles on github, for instance. Search for the readme string and you'll see a number of very dangerous commits.
16.
We can do better – Please fix plaintext credential storage in Chrome(medium.com)
13 points by shayanjm 11y ago|6 comments
17.
by shayanjm 11y ago
Shameless self-plug - I built a "pastebin scraper as a service" platform, and released it here: https://github.com/shayanjm/pasteye Took much inspiration from dumpmon, but distributed it so users can choose t
18.
by shayanjm 11y ago
I remember chatting with you a while back as I was also working on a pastebin scraping project. http://github.com/shayanjm/pasteye Glad to see dumpmon is still going strong :)
19.
by shayanjm 11y ago
Absolutely - this wasn't meant to be a be-all-end-all to lead generation by geographical bounds. This was more of a "I had a cool idea, wanted to test it, and here's how I went about doing that" sort of post. There are a
20.
by shayanjm 11y ago
That's smart - coupling this with the 'min layers' equation could lead to a 100% iterative solution. We could swap this component in and remove the recursive triangle generation :) As mentioned in the post - deduping is cheap
21.
Spherical Trigonometry, Circle Packing, and Lead Generation – A Journey(blog.ablelending.com)
43 points by shayanjm 11y ago|9 comments
22.
by shayanjm 12y ago
Looks like something I'd use every day if executed properly. Some things that I think would make this an awesome app: 1) IFTTT integration would be sweet. "If i get an e-mail about something, remind me to do it by putting it in my
23.
by shayanjm 12y ago
@shayanjm on twitter. I have a half-baked contextual analysis implementation which I could probably spin into a high-volume twitter analysis tool. Was doing NLP analysis on unstructured data (like news articles) and extracting topics + extr
24.
by shayanjm 12y ago
Agreed. I think generating a more filtered list is possible - but would take significantly more time than it took to build the sentiment analysis tool. The results of the more-filtered-list-tool would be quite interesting, though, as you&#x
25.
by shayanjm 12y ago
Kudos on building something cool to solve a pain point, but unsure about the efficacy of this implementation. A list of negative-sentiment tweets about competitor products is certainly a good place to start, but is by no means a list of act
26.
by shayanjm 12y ago
If you have your personal projects listed in your resume - you'd better make damn sure they are at least 80% polished and _SHOW OFF_ your skills. If they are not/do not, don't list them in your resume or make them private.
27.
by shayanjm 12y ago
Not here to say that one style is better than another, although I've used gitflow almost exclusively for a few years and it's served well. Pushed features/hotfixes/etc. out the door in a timely manner, and never felt hel
28.
by shayanjm 12y ago
Most of it is stuff he would have to unlearn before he could then learn it on the job. I would rather hire someone who has no habits than one who has bad ones.
29.
by shayanjm 12y ago
No, I wouldn't hire you as-is. Probably not a first-choice intern either. Some notes on why not: + Your web-sume looks rough. As pointed out by others, there are a number of typos (i.e: "and provide an opporunity") not to m
30.
by shayanjm 12y ago
Thought I'd give a contrarian perspective here. I was hiring for a telehealth startup based in NYC and got some referrals to some recent General Assembly grads. I bit and went ahead and scheduled some interviews. It was a total joke, h

More ›