ca_parody · HackerTrans

Is there a cicada in the logo?

Sesco | Quantitative Developer | Pittsburgh, PA - Onsite | Full Time

Sesco is a proprietary energy-trading fund located in Pittsburgh, Pa. We are looking for quantitative developers to work in the automated trading team and help build performant, risk-ajusted trading systems. Contact [email protected] for information.

ca_parody

6 lat temudiscuss

I would also add, for data-science esc tasks (or bulk queries in general) - COPY {} TO is unreasonably fast - often much faster than executing the standard select (especially if that select is being executed by a driver in a slower language).

_

[0] https://www.postgresql.org/docs/9.2/sql-copy.html

ca_parody

6 lat temudiscuss

I am sympathetic to the practical issues & biases at play when discussing the systems we often call meritocracies; especially when those systems result in anti-meritocratic behavior.

However, what is the reader supposed to come away with here as an alternative? Equality of outcome? Should it really be that any given combination of work ethic, dedication, novelty, and a bit of luck will not land you above the mean? I personally am all for improving meritocracies so they live up to the name, but abolishing them due to their implementation failures seems silly and more importantly, the alternatives are dangerous.

ca_parody

6 lat temudiscuss

Sadly one doesn't - unless one happens to have ~300GB of RAM to fit the model into memory and a close personal friend at openai who will share the learned weights with you. Training your an even more expensive endeavor.

Presumably this is how they are justifying the for-a-price API; "its not like you can run it on your home computer anyway". For now, the API is private and geared towards researchers. Still a bit bollocks though.

There are plenty of wrappers [0] around GPT2 though - and those you can probably run on your home workstation.

_ [0] https://pypi.org/project/gpt2-client/

ca_parody

6 lat temudiscuss

Has the 747 transported to most human-distance (people * (km|miles)) of any other make&model transportation machine? I would imagine so...

ca_parody

6 lat temudiscuss

Honestly, for however much this project either (a) is a genuine archeological move for the preservation of information or (b) to get good press, all I genuinely thought when this happened is "aw shucks - wish i fixed those bugs before they zapped it onto film and flew it to santa clause".

ca_parody

6 lat temudiscuss

Forgive me - but how does this avoid the chicken&egg problem here. Without digging through the promo copy, why would one programmatically label training data to do ML on if they have such a program to label data...

ca_parody

6 lat temudiscuss

To me, Go feels condescending to write without generics. I may just not groke the idioms - but putting users in a walled garden that the standard library has the privileged to escape from (i.e. (Map<k, v>)) just seems to go too far in not trusting the programmer. The fact that generics have taken so many years - to still be talking about what [<( to use is beyond me.

There is a difference between becoming C++ and allowing programmers to make fundamental abstractions without interface{} hacks.

ca_parody

6 lat temudiscuss

https://github.com/mratsim/Arraymancer probably has a lot of examples in it's implementation

ca_parody

6 lat temudiscuss

its been a couple years - but mratsim (the same person who authored this ray tracing post) wrote this (which is at least tangential to your question) https://medium.com/@MARatsimbazafy/why-rust-fails-hard-at-sc....

Some of the rust problems mentioned (like clone for arrays) have been fixed. Others, like static generics, have not been (AFAIK).

ca_parody

6 lat temudiscuss

The main reasons I see for nim in scientific computing over python are:

(a) you need performance for applications outside of numpy / pandas & you want to avoid writing a C extension to do so (nim is also actually a good choice for python extensions via nimpy - if you are still using python as the main language of your project/system)

(a2) you want good concurrency (some would say python's async/await is good - but it feels too little too late to me) & you at least want the option of parallelism (like this article takes advantage of)

(b) you want / need the powerful macro system for dsls / custom (compile-type-checkable) language extensions.

There is a good future for nim in scientific computing (frankly, there is a good present) - but for it to be better, people need to be willing to forgo some of the ecosystem (when possible) and to help build that future.

https://github.com/nim-lang/needed-libraries/issues/77

ca_parody

6 lat temudiscuss

They provide NAM, GFS, HERR, etc... they also have proprietary models (CBAM1 & CBMA2 that are higher resolution)

ca_parody

6 lat temudiscuss

I think if that some of these concepts should just be included with the 3.9 Annotated Type

    Annotated[IO[str], unpure]

in 3.9 and above if the authors want to commit to type hints being core to this. I agree that the decorators feel a little wrong

ca_parody

6 lat temudiscuss

Why the lambdas at all in these examples...

  (lambda real_user: real_user.get_balance())

and not

  User.get_balance

(besides subclasses not getting their balance called...)

ca_parody

6 lat temudiscuss

Nim feels more like hyper modern python to me. Haskell is an entirely different paradigm.

comments