2,064,114 events, 1,028,004 push events, 1,644,546 commit messages, 124,668,766 characters
HopHR - Principal Data Scientist - statistical modeling. Vulcan Inc. Careers - Fisheries Data Analyst Intern. Výzkumný pracovník – odborník pro statistické zpracování dat | ResearchJobs.cz. Data Analyst | Manifest Solutions. Data Analyst at HackerEarth . Data Product Operations Analyst - Validity - Career Page. Hey—the New York Times is hiring an election forecaster! « Statistical Modeling, Causal Inference, and Social Science. Special Year on Statistical Machine Learning - Statistical Machine Learning Bootcamp. SkipTheDishes Careers - Data Analyst. Clinical Data Analyst II | PWW Recruiting, LLC. .
Once again, god fucking dammit Derynect
...As well as other assorted changes to streamline things and improve the overall user experience.
[MIRROR] Adds a new Tech to the B.E.P.I.S., Specialized Engineering, and a new Minor Reward. (#1088)
- Adds a new Tech to the B.E.P.I.S., Specialized Engineering, and a new Minor Reward. (#48507)
About The Pull Request
Round 2. image Adds a new tech to be unlocked within the B.E.P.I.S.'s major techs, called "Specialized Engineering". Within that tech are 2 new items, Heat Resistant Rods, and the Tinker's Gloves, as well as a new minor tech, the Survival Pen. Heat Resistant Rods: Have you ever wanted to expand Lavaland Base? Build a lava fortress worthy of your magnificence? Well, now you can! Using Heat Resistant Rods, you can make a catwalk to cross lava tiles, as well as build atop lava, without messy methods like using the RCD! Tinker's Gloves: This just in engineers, insulated gloves have new competition in town! The Tinker's gloves are complicated, overdesigned gloves that, while not very shockproof, allow for faster wall girder construction. No longer will you need to die of old age walling off an area when you're out of RCD ammo! Warning: This product contains no likeness to clockwork gauntlets used by the extinct cult of Rat'Var, and any such similarity is by no means intentional.
Survival Pen: Have you ever been stuck on lavaland, trapped by your survival pod with just a few chairs, some titanium walls, and walls of ash closing in on you? Well THANK GOD you have your Rockbreaker brand Survival Pen! Allows for basic mining operations, and is portable like a standard pen! BUT WAIT! A watcher blocks your path. Thankfully, your expensive, diamond encrusted pen isn't just good for being the world's slowest pickaxe, it's ALSO coated in the one thing watchers crave: DIAMONDS. Toss the pen to draw the watcher's attention elsewhere. Why It's Good For The Game
Starts to fill our B.E.P.I.S. Major Reward Techs, as intended, and fills a niche that doesn't get much play nowadays: Building on Lavaland. This was something I've had in the back of my mind since we did the first tests of the Disaster gamemode a few months ago, so each item here was made with the intention of improving the mining base should the station become un-livable. Lava-Proof Rods are beneficial in that you'll be able to cross and build over lava by the mid-end of the shift. RCDs still work as usual, but this way you won't have to worry about getting specialized engineering equipment as a member of cargo or science. The tinker's gloves are basically a variant of the nitrile gloves, but they grant a speed bonus to adding plating to metal girders, so that some kind of functional alternative to giving every engineer insulated gloves. Survival Pens are quite honestly very niche, but getting dumb pens with extra functionality is a tradition of trade shows all over the world, so it feels right at home as a minor reward.
Also, does a tiny change to the doe sprites, just to look a little bit less old. Changelog
cl add: A new Technology has been implemented as a major reward in the B.E.P.I.S., Specialized Engineering, to appeal to engineering utility and new construction horizons. tweak: Watchers will now actively consume diamond ore left lying around, alongside the new survival pens. /cl
- Adds a new Tech to the B.E.P.I.S., Specialized Engineering, and a new Minor Reward.
Co-authored-by: ArcaneMusic [email protected]
[MIRROR] SEVERAL mask sprite fixes (#1100)
-
SEVERAL mask sprite fixes (#48744)
-
society moment
god im so funny
- do you apes not know what consistency means?
seriously who the fuck made these sprites
- SEVERAL mask sprite fixes
Co-authored-by: Rob Bailey [email protected]
Hello, good day friends.
It has been a long time things, but now we are back with good packages and something lovely that will amazing you in life and I would love everyone in here to part take in my package because is one of the lovely packs ever
Removed a dumb word I used in a comment
Yikes. Somebody brought this to my attention, I didn't even remember that I'd written it. "Spa" or "Spastic" is kind of a south park esque slang term that used to be pretty common in Ireland, which I used without even thinking about it. It's definitely not something I would say anymore, 10 years on, and it's something I shouldn't have said at the time either. I'm sorry :(
(somebody on twitter was asking me about how much cleaning up of the source code I did before launching this. I think this commit kinda answers that)
"9:40am. I am up. What a night, the less I speak of it the better. Thankfully I will forget it soon. Nightmares where I am playing a game and the controls do not do want are the worst.
It is time to chill. Let me rant a bit.
...
No matter how good Dendro is it really is just fiction. For fiction to take the next evolutionary step, eventually the authors will have to ask the pertinent question of how to take the power from the worlds inside to the outer ones.
The Dendro author is really smart for doing the setup that he did - the MC can essentially respawn and has game-like superpowers, so the main tension is who gets his way and the tians are the ones who have their lives at stake. To me this makes the story more credible. It is not like in regular shonen the main character can just die - they are always protected by plot armor. With this setup there is no (as much) need for that so the story has more integrity as result.
9:50am. But eventually the understanding that the game is just a parasite on the players will dawn.
Overcoming this part is as what I see as my own responsibility.
9:55am. Well, let me waste time as usual, and then I will get my start. Today I will aim to get the first part of the prepass done - the inlining of globals.
There is just so much work to be done, that I won't set any deadlines lest I burn out. The Inspired would be able to do this 24/7 until they are done, but I am going to play this game as what I am.
12:55pm. Ok, let me stop entertainment here for a bit and get ready to do the chores. After that I will start this thing. I want to spend some time thinking about the prepass. It is finally time to begin work on it.
1:35pm. Done with chores. Let me finally start this thing. I should focus on work for a while and leave the rest aside.
1:40pm. Focus me, focus. Let me just do this thing. I can do the prepass, no problem. I need to remember what I thinking of yesterday and just do it. It won't take too long at all.
I won't mentally strain myself by trying to do the whole prepass all at once, let me just focus on inlining of globals for the time being.
1:50pm.
module Spiral.PrepassGlobalInlining
open Spiral.Parsing
let global_inline x =
match x with
| RawB -> x
Right. Well, first of all I need to iterate over the global statements. For now, I do not even have type or union, so that only leaves let statements and recursive blocks.
2:05pm.
module Spiral.PrepassGlobalInlining
open Spiral.Parsing
type GIEnv = {values : Map<string,RawExpr>; types : Map<string,RawTypeExpr>}
let global_inline (env : GIEnv) x =
match x with
| RawLet(name,RawInl(arg,body),on_succ) -> ()
| RawRecBlock(l,on_succ) -> failwith ""
| _ -> failwith "Compiler error"
2:10pm. I am thinking about this. I am confused once more. Was this really how I intended to do it?
Er, by that I mean, consider what happens when I compile a module, but not necessarily the whole program.
{values : Map<string,RawExpr>; types : Map<string,RawTypeExpr>}
Rather than getting RawExpr and RawTypeExpr, what I'd get instead is the Expr and TypeExpr.
2:40pm. I am still thinking about it.
It is easy to say, 'just do it', but I do not want to more passes that are necessary around as I will have to cache them later.
I am going to have to merge inlining with the prepass after all.
2:45pm. Ok, I will handle it somehow. It won't be too big of a deal.
Let me take a little break so I can read Kengan Omega."
Solved Day3Part1 Holy fucking shit on a stick did that take a long fucking time
Fucking Poggit... Always a pain in the arse! 🙄
Fucking Poggit... Always a pain in the arse! 🙄
"2:55pm. Let me resume. Where was I?
| RawGlobalInl of int // Has no free vars. Uses global dictionaries during peval for inlining of recursive occurrences.
| RawGlobalInlInlined of VarString * RawExpr // Has no free vars.
I am not going to need this. I've been debating in myself whether to replace | RawGlobalInl of int with | RawGlobalInl of string, but the right answer is to just take it out completely.
| RawModule of Map<string,RawExpr>
| RawModuleOpenInlined of Map<string,RawExpr> * on_succ: RawExpr
This is as well.
| RawTInl of (string * RawTypeTypeExpr) * RawTypeExpr
| RawExpr of RawExpr
Among the raw types, let me take out these two.
I can't deal with it right now.
3pm. Ok, focus me. I can chill later.
First of all, how about I bring the Exprs back?
3:10pm.
// Globals
let mutable tag_prepass = 0
let private tag () = tag_prepass <- tag_prepass + 1; tag_prepass
I admit, I completely forgot that Exprs had these tags for easier comparison.
And until I reminded myself just now, it slipped my mind that I was doing the free vars and stack size calculations interleaved with renaming.
let prepass x =
let free_vars_and_stack_size_dict = Dictionary(HashIdentity.Reference)
let pattern_dict = Dictionary(HashIdentity.Reference)
let renaming_dict = Dictionary(HashIdentity.Reference)
I really, really am getting massive mileage from understanding reference equality.
V0.2 is going to be the version of Spiral where I hammer all of those lessons home.
Now...
3:30pm. I am thinking about it. I am in the flow, but I do not feel like coding just yet.
I was thinking about the inlining, but it had not occurred to me that I would need to merge it with the prepass, so it took me a little off guard.
3:55pm. I am still thinking about it.
I realized just now, arrays are going to have to be nominal types.
Right now, what I am thinking about are how the prepass should be done for the raw typed expressions.
Things like higher order type functions, arrays and primitives.
How do I do inlining of them.
I had not thought about it yet, but now is the time to decide how to do type inbuilts. Should types have ops?
4:30pm. Ok, it turns out there is a lot to think about even before I put in a single line down.
One thing I am also thinking about is getting rid of the builtins.
Instead I could put them directly into the map myself.
One thing I would also want to do is do some inlining in the prepass directly.
For example a + b to !Add(a,b) where applicable. The way this would work is that I'd have...
(inl a b -> !Add(a,b)) a b. This I could reduce in many cases during the prepass into something like let a = a in let b = b in !Add(a,b). And this would speed up partial evaluation by a decent amount.
But there is no shame in leaving that for later.
I can always plug that in when everything else is set.
I should not make it a priority while I am still trying to get the language off the ground.
Later when I have the language working and the test suite operational, then I can do this.
4:40pm. Since it does not seem like I will be writing any code today, let me do a little thing before I leave. Rather than just sit here in this chair like I have for the past 3 hours, it would be more comfortable to think while I am taking a bath.
| RawTInl of (string * RawTypeTypeExpr) * RawTypeExpr
| RawTPrim of PrimitiveType
| RawTArray of RawTypeExpr
I put the type function back and added these two babies. I am really not going to get far without at least the primitive type and arrays.
For other kinds of types, I can consider that question later.
Once I do the fundamentals they won't really be a problem.
4:50pm. Now let me commit here. I probably just call it a day when I get back so most likely this will be my last entry today.
Without a doubt, today did warm me up towards the prepass quite a bit. I'll start hammering it out tomorrow. Unlike the parser, the prepass itself will be relatively straightforward to deal with. It should not take me longer than a few days.
And when I am done with it comes partial evaluation. That is the home stretch.
The code generator itself won't have much in the changes to speak off unlike the other phases.
Most likely, by the time this month is over I will have already started testing.
And on February that will be when I will add the rest of the features and start work on typechecking itself.
That part will take me a few months, but once I've mastered it, the language will be complete.
What I will be doing at that point is two things: moving on to putting in editor support by writing a VS Code plugin, work on making the error handling and reporting graceful, and actually writing the programs in Spiral itself. Plus some bells and whistles like catching of intermediates.
By the end of the year I should be making good use of Spiral, and by the end of next year not only will I have caught up to par with where I was in 2018, but I will have implemented all the ideas that I have currently. That includes those RL agents.
5pm. 2022 might not mark the start of the Singularity, but it will mark the start of myself as a fully fledged programmer capable of using ML to good effect.
In 2018 and 2019 I pretty much beat myself. I will not let myself be clowned again like that time.
Classical logic might be the enemy of understanding and reason, but attacking it won't be constructive. Instead what I will do is attack the very conception of understanding itself. That should be much more profitable. I will bring out the dark arts that gave me the original inspiration to start this quest.
In 2019 I've made absolutely sure - it is not that my math skill being too low is the main problem. I can get math proofs when they are computational in nature and the subject matter is tractable. The issue is that there are very simple algorithms which are essentially intractable and the ML community is spewing bullshit everywhere.
Previously, I regretfully skipped the proofs as I read papers, but this time I will take a piss on them as I move past them.
Ultimately, as a general method for building understanding, writing programs about programs is highly suspicious at best, and at worst fraud (when done axiomatically.)
Randomized testing is the guiding light. If randomized testing is right then ML itself is the main principle of science.
5:10pm. In order to make progress, I need to get rid of the monkey mentality of looking for the right algorithm to plug in and start thinking of my programming as an expression of understanding itself.
Whether it be DL or CFR, I won't be able to reduce them to something more primitive. They themselves are primitives that I need to use to create a language. Not a programming one, but my own personal, internal one.
That is what it means to make the machines learn.
I will end this."
Fixed bug 4903 - Lack of color multiply with alpha (SDL_BLENDMODE_MOD + SDL_BLENDMODE_BLEND) blending mode for all renderers
Konrad
This kind of blending is rather quite useful and in my opinion should be available for all renderers. I do need it myself, but since I didn't want to use a custom blending mode which is supported only by certain renderers (e.g. not in software which is quite important for me) I did write implementation of SDL_BLENDMODE_MUL for all renderers altogether.
SDL_BLENDMODE_MUL implements following equation:
dstRGB = (srcRGB * dstRGB) + (dstRGB * (1-srcA)) dstA = (srcA * dstA) + (dstA * (1-srcA))
Background:
https://i.imgur.com/UsYhydP.png
Blended texture:
https://i.imgur.com/0juXQcV.png
Result for SDL_BLENDMODE_MOD:
https://i.imgur.com/wgNSgUl.png
Result for SDL_BLENDMODE_MUL:
https://i.imgur.com/Veokzim.png
I think I did cover all possibilities within included patch, but I didn't write any tests for SDL_BLENDMODE_MUL, so it would be lovely if someone could do it.
stop fucking my name !!!
fucking idiot stop fucking my life!!!
OH GOD OH FUCK THEY KNOW IM A GRAYTIDE SHITLER NO PLEASE I JUST WANTED THEM TO BE PRINTABLE AT CARGO OHG OD OH FU
Did some crazy shit
made this complicated nested for loop thingy that absolutely fried my brain but it ended up working. Fuck swift.