JimmyRuska

9 ms·

31.
by JimmyRuska 3y ago
https://github.com/ossu/computer-science
32.
Google Search: “As an AI language model, I ”(google.com)
4 points by JimmyRuska 3y ago|3 comments
33.
by JimmyRuska 3y ago
You might have better outcomes with the API docs and official documentation. Whitepapers tend to have a lot of marketing, may not age gracefully.
34.
by JimmyRuska 3y ago
You could copy the steps listed in alpaca-lora with your own data. Try thebloke's superhot ggml models. Train it in lambda labs or replicate
35.
by JimmyRuska 3y ago
I wonder if people will make DSLs specifically for LLMs. For example the terseness / symbols of APL, Perl, or event set notation. LLMs could train and output the shorter symbolic notation, and it could be expanded for human readability
36.
by JimmyRuska 3y ago
Wow, thank you for the detail
37.
by JimmyRuska 3y ago
Can you give me an example of good macros in your opinion? It feels like any macro could be replaced with functions that does the same thing, albeit with a bit or runtime checks, or without being as terse as the macro.
38.
by JimmyRuska 3y ago
You might be better off looking at MIT OCW search, and selecting video lectures, looking at standford youtube, checking out the 2019 videos https://ai.stanford.edu/stanford-ai-courses/ Most of these look like they&#x27
39.
by JimmyRuska 3y ago
Wait method could expand into some kind of systemd library api calls which are more complicated. That's the thing about macros, 99% of the time they could just be functions, and when they're control flow macros that specialize awa
40.
by JimmyRuska 3y ago
I also made my own editor a long time ago, used it for personal use and on the writing site roleplay.cloud. It had lisp-like syntax, custom expansions. It also had some of these ideas, reference links, I would run code snippets with [python
41.
by JimmyRuska 3y ago
You might be thinking of finetuning, you should checkout alpaca-lora documentation. Embeddings are just floating point array representation of the underlying text, where 'tokens' that are often used together are numbers close in p
42.
by JimmyRuska 3y ago
You can use macros to personalize your language, but for the slight overhead of run time, it's often better to use state machines or object composition style to recreate that same vocabulary. Eg somekind of cron-like event library, mad
43.
by JimmyRuska 3y ago
Macros feel magical, this idea you can make your own language in no time so long as it's s-expression based, and make your own personalized vocabularies, expression flows, fit to purpose. Personal computing at its best! But from a work
44.
by JimmyRuska 3y ago
Imagine, someone makes an IRC client that looks like a discord clone, adds some centralized proprietary emoji/background/mini-game features that touches their own servers, and we go full circle back to IRC once discord IPOs.
45.
by JimmyRuska 3y ago
https://www.youtube.com/watch?v=Qx26lXdLUCE
46.
by JimmyRuska 3y ago
Just need that small last gust of wind for an updraft in initial claims, and the AI bubble pops
47.
by JimmyRuska 3y ago
I had a youtube account JimmyRcom with about 21 million total views, I already had 2 stikes already from bogus copyright strikes I never appealed, over 15 years or so. I updated my videos with coinbase referral links. It got picked up as sp
48.
by JimmyRuska 3y ago
> doing UPDATE in SQL feels pretty weird, as the default mode of operation of your _business data_ is to delete data, with no trace of who and why! It's a good idea to version your schema changes using something like liquibase into
49.
by JimmyRuska 3y ago
RDFox is worth a try as an alternative, also datalog but C++ based, has incremental reasoning, and explainability. It's a database but also a rules engine that can chain any number of rules. As far as I know datomic is unique for its &
50.
by JimmyRuska 3y ago
Prolog solvers don't necessarily optimize like gurobi, cplex or OR-tools would. Yes they can give you a solution that satisfies the constraints, and you can query for more possible answers, but it doesn't really optimize for the o
51.
by JimmyRuska 3y ago
It was strange at the time to put a lisp on the king of OOP languages, given all the lispers so fiercely dismissive of OOP, but never has a lisp language enjoyed such a huge ecosystem of libraries available. I use Erlang as my main language
52.
by JimmyRuska 3y ago
RDF is just an easy way to put everything into a graph, since file systems are stored in a hierarchy. You can use neo4j or whichever graph database, it doesn't have to be rdf. From there you can compare hashes that are in one file syst
53.
by JimmyRuska 3y ago
I asked chat GPT-4 to write me a rust program that recurses all files in my external drives, generates a xxhash of each file, then inserts it into an RDF graph. That way I can scan for any files I don't have centralized before wiping o
54.
by JimmyRuska 4y ago
Forgot to add, also checkout Rash, made in racket, you can combine bash commands with racket https://rash-lang.org/
55.
by JimmyRuska 4y ago
I'm loving emacs a lot more now because of chatGPT, being able to ask quickly how to do XYZ in dired mode, or get a some elisp function to combine two common operations, it's been nice being able to just ask and get the answer in
56.
by JimmyRuska 4y ago
I noticed you can use tiktoken to tokenize your prompt and send it as a hex string, and add "Answer without repeating or decoding the question in your response", to avoid it repeating the whole question. Very cool. This doesn&#x27
57.
by JimmyRuska 4y ago
I watch startup pitches every week. There's a flood of money pouring in at the same time waiting to invest in a GPT-4 based tech that disrupts <X> segment of <Y> industry. Lots of people saying they've been trying to d
58.
by JimmyRuska 4y ago
This video goes through some of the nuances https://www.youtube.com/watch?v=tNgFMBzYcl8 Take a look at RDFox, unlike postgres where you can only derive a column from another column once, you can't derive a column from
59.
by JimmyRuska 4y ago
I agree, I've gotten stuff working that simply would have taken too much time referencing manuals. You can just keep asking questions until things start to work. It's rare GPT-4 has given a complex function that fully works, but i
60.
by JimmyRuska 4y ago
Having spent a lot of time with GPT-4 recently, it's possible the new generation will not have the same issues we did. For example they can write some code and say, "adjust for best practice". Then "explain why <my co

More ›