Viewing profile — shayanjm

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 standa…

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 …

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 o…

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 snak…

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 …

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…

We explored a novel method to gauge the significance of tokens in prompts given to large language models, without needing direct model access. Essentially, we just did an ablation …

I came here to ask that EXACT question - fighting the good fight

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 implementati…

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 …

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.

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 distribu…

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 :)

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 abo…

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 :)…

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…

@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 unstruc…

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-filt…