|
Fil5000 posted:That's not always true, James Hoffman is the biggest coffee dude you've ever seen and just likes to see people enjoying coffee Plus I love that he always goes into every video clearly explaining what the product/technique is good for. He's a true rarity on YouTube.
|
#
?
Mar 17, 2024 07:03
|
|
|
|
| # ? Apr 29, 2024 16:26 |
|
|
TearsOfPirates posted:Plus I love that he always goes into every video clearly explaining what the product/technique is good for. He's a true rarity on YouTube.
|
|
#
?
Mar 17, 2024 09:16
|
|
|
I thought Hoffman was going to be a huge snob till I watched his Nespresso video and it was really insightful in what made it so special. Are we still recommending YouTubers? I stumbled upon abe's projects where he makes little devices. He goes over everything from the design, building and troubleshooting in a concise clear manner with some fun sleight-of-hand tricks https://www.youtube.com/watch?v=wVcerPofkE0
|
|
#
?
Mar 17, 2024 10:02
|
|
|
Sagacity posted:In that case I can also recommend Hames Joffmann's videos. Hames Joffmann is adorable https://www.youtube.com/watch?v=tltBHjmIUJ0
|
|
#
?
Mar 17, 2024 10:25
|
|
|
mobby_6kl posted:Hames Joffmann is adorable I think this is the video that brought him to my attention, and made his lack of snobbery pretty clear - like, he just wants people to like coffee. https://www.youtube.com/watch?v=Z-iNAyu-ejo
|
|
#
?
Mar 17, 2024 11:24
|
|
|
Fragrag posted:I thought Hoffman was going to be a huge snob till I watched his Nespresso video and it was really insightful in what made it so special.
|
|
#
?
Mar 17, 2024 12:45
|
|
|
 Hames Joffman is not James Hoffman:https://www.youtube.com/watch?v=lf2K2PuEUv8
|
|
#
?
Mar 17, 2024 13:21
|
|
|
James Hoffman's entire personality is absolutely just coffee, but it's not really a bad thing for him because he's pretty chill and inviting. He makes good home-brewed coffee feel accessible and unpretentious, in the same way that someone like kenji lopez-alt does for good food.
|
|
#
?
Mar 17, 2024 14:34
|
|
|
Hoffman knows that there is a level of coffee that everyone can enjoy and just wants to nudge people to try something different or new.
|
|
#
?
Mar 17, 2024 19:34
|
|
|
I think the difference we're inching toward here is that Hoffman isn't a coffee guy, he's a coffee expert. The former are just consumer-enthusiasts applying the tech/gearhead approach to refining their own process, while Hoffman is actively exploring and advancing the field as a whole.
|
|
#
?
Mar 18, 2024 02:17
|
|
|
redstormpopcorn posted:I think the difference we're inching toward here is that Hoffman isn't a coffee guy, he's a coffee expert. The former are just consumer-enthusiasts applying the tech/gearhead approach to refining their own process, while Hoffman is actively exploring and advancing the field as a whole. he's the coffee engineer
|
|
#
?
Mar 18, 2024 03:55
|
|
|
He's a coffee achiever!
|
|
#
?
Mar 18, 2024 05:12
|
|
|
he’s a world barista champion (2007, IIRC)
|
|
#
?
Mar 18, 2024 06:31
|
|
|
Hoffman, walking into an unfamiliar room, “someone brewed here”
|
|
#
?
Mar 18, 2024 12:41
|
|
|
Hoffman is the difference that simply having a deep and genuine passion for what you're talking about makes, and he knows that people want to make the best cup of coffee and are willing to spend certain amounts of money so he approaches things without bias and reviews them all honestly.
|
|
#
?
Mar 18, 2024 15:59
|
|
|
TheFluff posted:
Oh poo poo I thought you were just joking with the name, that's hilarious Beve Stuscemi posted:Hoffman, walking into an unfamiliar room, “someone brewed here” Maybe briped though!
|
|
#
?
Mar 18, 2024 18:12
|
|
|
Fil5000 posted:I think this is the video that brought him to my attention, and made his lack of snobbery pretty clear - like, he just wants people to like coffee. And he does try silly things like roasting beans in nitrogen, roasting them in an air fryer and deep-fried coffee beans.
|
|
#
?
Mar 19, 2024 12:43
|
|
|
The new Gamers Nexus video is worth watching just for the "Everybody Loves Nvidia" segment. It's amazing watching Steve die inside more and more every clip. https://www.youtube.com/watch?v=0_zScV_cVug
|
|
#
?
Mar 22, 2024 10:41
|
|
|
That was painful to watch. NVIDIA has gone straight to the moon. Really wish I bought some stock now.
|
|
#
?
Mar 22, 2024 19:34
|
|
|
I used to feel bad that I was priced out by Nvidia. I also feel worse that I bought into the ray tracing marketing with a GPU that wasn’t cut out for it and also lacked VRAM (3070). I think I might be on team red now for the Linux support and the store-brand upscaling that seems to get most of the way there.
|
|
#
?
Mar 22, 2024 19:48
|
|
|
Whitest Russian posted:That was painful to watch. it's incredible how loving idiotic investors are that they're actually fooled by this generative AI garbage
|
|
#
?
Mar 22, 2024 21:21
|
|
|
I mean the bill of goods they're being sold is "this will do a bunch of creative people's work, which means we can push down wages or just hire fewer people" which then feeds into the infinite growth cult that the current stock market is built on and which is hopefully going to come crumbling within the next year. It's already reaching a point where companies are ripping the wiring out of the walls to try and get another growth quarter but it's not sustainable forever, and normal people are starting to notice that whenever "AI" is pushed into a thing they use that thing gets worse.
|
|
#
?
Mar 22, 2024 21:25
|
|
|
Shipon posted:it's incredible how loving idiotic investors are that they're actually fooled by this generative AI garbage The last 15 years of technology has been a drive to introduce change with little to no consideration to if the product produced is actually an improvement. The Generative AI bubble/gamble is certainly better than VR, but it's also being pushed by big tech because they can control it so completely. "Your industry is going to be completely changed because of generative AI, you don't want to be left in the dust do you? Better buy some of our cloud compute before your competitor does."
|
|
#
?
Mar 22, 2024 21:39
|
|
njsykora posted:I mean the bill of goods they're being sold is "this will do a bunch of creative people's work, which means we can push down wages or just hire fewer people" which then feeds into the infinite growth cult that the current stock market is built on and which is hopefully going to come crumbling within the next year. It's already reaching a point where companies are ripping the wiring out of the walls to try and get another growth quarter but it's not sustainable forever, and normal people are starting to notice that whenever "AI" is pushed into a thing they use that thing gets worse. I'm not saying you're wrong, but I'm saying that "it'll collapse any day now" has been what I've heard every year since 2008, and probably earlier than then Ol' Karly M said much the same thing and yet capitalism has yet to die
|
|
|
#
?
Mar 22, 2024 21:39
|
|
|
This isn't end of capitalism stuff though, it's "the big companies are going to end up reporting quarters of decline". There's only so many customers and only so much you can hike prices before it has an impact and people just stop buying. You're already seeing that in videogames where people are buying fewer games but the higher prices were able to hold growth up for a while. It's just expecting the popping of the current bubble.
|
|
#
?
Mar 22, 2024 21:59
|
|
|
It’s also trickling down to web tools that I use that have to rebrand themselves as AI powered for investors and forcing things I absolutely didn’t want or need. I hope it dies fast. I feel like we’re at the point where AI is whatever you want it to be.
|
|
#
?
Mar 22, 2024 22:14
|
|
|
buglord posted:It’s also trickling down to web tools that I use that have to rebrand themselves as AI powered for investors and forcing things I absolutely didn’t want or need. I hope it dies fast. I feel like we’re at the point where AI is whatever you want it to be. There's someone at my work who sent me an email that basically said "Cyrano, can I have the last three years of data for <sensitive business process>? I want to experiment with using an AI model to predict future needs." I CC'd my boss with a note that I'd be happy to help as long as it got cleared, but by the way you're using an in-house model on our own hardware, right? Because there are a lot of concerns about how Google, Microsoft, OpenAI, et al handle their data and I'm not sure if three years of info on <sensitive business process> is something we want to release publicly. Boss, any input? A few hours later I received a VERY curt "after consulting with legal I will not be needing this information."
|
|
#
?
Mar 22, 2024 22:19
|
|
|
I hadn't fully realized how much Nvidia have shifted to milking LLMNN VCs, but it's pretty loving depressing to watch - because it sure as gently caress ain't gonna mean cheaper GPUs for any time soon.buglord posted:I used to feel bad that I was priced out by Nvidia. I also feel worse that I bought into the ray tracing marketing with a GPU that wasn’t cut out for it and also lacked VRAM (3070). I might not be able to get much out of raytracing, but outside of screenshots, I'm not sure I'm losing much. Thinking about it, it reminds me a lot about the early days of 3D acceleration when there were competing APIs, Direct3D hadn't even come on the market yet, and it was still mostly used as a demo for screenshots, rather than something you'd notice while playing a game. I fully expect that it'll eventually find its niche - but for now, until there's a game that supports quadrupling resolution and textures (like Elite Dangerous can do) in a screenshot-mode, where it also applies raytracing, it feels a bit of a gimmick to me.
|
|
#
?
Mar 22, 2024 22:30
|
|
|
I don't think ray tracing was ever really put out there as something that the cards magically made performance without the magic of DLSS. Personally I barely noticed the difference but I mostly play multiplayer shooters so I'm already running my games on low settings on my 3080 for them sweet FPS. I think DLSS was a pretty big deal, it was definitely noticeably worse image quality but it really did up performance at high settings very well for what it was
|
|
#
?
Mar 22, 2024 23:21
|
|
|
Cyrano4747 posted:There's someone at my work who sent me an email that basically said "Cyrano, can I have the last three years of data for <sensitive business process>? I want to experiment with using an AI model to predict future needs." I have run into a few people at work who told me they have used ChatGPT or similar to condense notes and I have gently made them aware that OpenAI is another company and if they're submitting notes with confidential information or PII, they are big-time breaching our policies. Judging by some of the stares I got back, not a lot of people are thinking about it. On the other hand I tried to get permission to feed some of our SQL code into a local LLM on my personal machine to see if it could help unfuck it, and was denied because of confidentiality concerns. Like even if I uploaded our lovely code to the cloud, I'm sure our competitors can't make use of field names like FINAL_REAL_VALUE_2 from the query that creates the table TEST_111111. SpartanIvy fucked around with this message at 02:20 on Mar 23, 2024 |
|
#
?
Mar 23, 2024 02:17
|
|
|
i don't know how people can just blindly trust code generated by chat GPT or whatever, you don't know what it's doing or else you would have written it yourself!!!
|
|
#
?
Mar 23, 2024 02:36
|
|
|
Shipon posted:i don't know how people can just blindly trust code generated by chat GPT or whatever, you don't know what it's doing or else you would have written it yourself!!! It's kinda funny how "uncommented code is bad" has been a regular refrain because it means you don't know how to unpack it because the person that did it didn't tell you, but then people are willing to hang their hats on asking a pile of sand to do the same thing
|
|
#
?
Mar 23, 2024 02:40
|
|
|
My old job was starting to get really on board with the generative AI train when I left, to the point that some of the pre-sales guys were starting to look into feeding decades of SOWs into an LLM to be able to poo poo out new SOWs without involving any of the technical team. Now I work for a university, and ChatGPT is rightfully haram.
|
|
#
?
Mar 23, 2024 04:59
|
|
|
The thing that's so appealing about Generative AI isn't that it's good, it's that it's fast and (relatively) cheap. Even if it's only 5% as competent as a person doing a job, if it takes 1% of their total compensation to do that 5%, it's absolutely worth it to businesses. It also has the fringe benefit of being the perfect scapegoat for any bad decisions management makes since they are black boxes and will never speak up for themselves. Did we decline that mortgage to a family because they are black? No, we did it because the AI said they were a risky borrower. Why did it say that? We don't know. Generative AI is going to erase so many entry level and low level jobs in corporate, just like the introduction of computers did before. The lowest position you're going to see is one of a "manager" but they're not managing people, just different AI's. Their job isn't knowing the business or how to do things, but how to properly ask the AI's to do those things instead. The technology is immature and still not very good now, but the amount of money being poured into it and the speed of advancement is terrifying.
|
|
#
?
Mar 23, 2024 05:14
|
|
|
LLMs might be the only technology smarter than it's proponents, but that's the equivalent of jumping over a limbo pole
|
|
|
#
?
Mar 23, 2024 07:27
|
|
|
Shipon posted:i don't know how people can just blindly trust code generated by chat GPT or whatever, you don't know what it's doing or else you would have written it yourself!!! YMMV, but I never use LLMs for code I can’t write, but rather for obvious stuff and tedium that I’d rather have configured to taste instead of finding something and fitting it in. I’m a pretty good developer, but like, I can dictate a problem using Whisper (my hands are beat up, and I’m clearer when speaking anyway) to GPT4 (for personal uses, at work we use self-hosted models, which are much worse) and get something that I can give a quick once-over before integrating. It’s faster for me to review it and “go back and fix XYZ” than to churn out an API endpoint or an ORM call or whatever. I’ve also found that you can make a “custom GPT” and prompt it with stylistically desirable code in ways that all prompts using that have pre-baked, and it’s had the nice side effect of, say, making it super pedantic about quality logging and error messaging. (For getting that as autocomplete rather than a chat interface, I’ve found that Codeium is really good, better than GitHub Copilot, and free to use.) I don’t let it write stuff like authn/authz code, but I absolutely have used it to generate tests for that sensitive code; I find it helpful to either write the tests myself and tell the LLM to satisfy them, or to write the code myself and have the LLM write tests either beforehand to give some TDD red/green lights for development, or afterwards to capture the behavior of the code I’ve written. This stuff really does work and work well as (competent) senior+ augmentation, and I think that dismissing it because the hypebros are tedious idiots is going to surprise some folks. For code at least (and maybe very little beyond code and voice-to-text transcription, but without question for those) these things are game-changers. I think it will definitely cause a future talent pipeline stall, and I don’t know what we do about that—but it’s not like I’m getting reqs for juniors anyway. tracecomplete fucked around with this message at 07:50 on Mar 23, 2024 |
|
#
?
Mar 23, 2024 07:39
|
|
|
tracecomplete posted:YMMV, but I never use LLMs for code I can’t write, but rather for obvious stuff and tedium that I’d rather have configured to taste instead of finding something and fitting it in. I’m a pretty good developer, but like, I can dictate a problem using Whisper (my hands are beat up, and I’m clearer when speaking anyway) to GPT4 (for personal uses, at work we use self-hosted models, which are much worse) and get something that I can give a quick once-over before integrating. It’s faster for me to review it and “go back and fix XYZ” than to churn out an API endpoint or an ORM call or whatever. I’ve also found that you can make a “custom GPT” and prompt it with stylistically desirable code in ways that all prompts using that have pre-baked, and it’s had the nice side effect of, say, making it super pedantic about quality logging and error messaging. (For getting that as autocomplete rather than a chat interface, I’ve found that Codeium is really good, better than GitHub Copilot, and free to use.) Recently my work gave me github copilot. While I insist everyone should be actually reading what its producing there are some major advantages in just helping not write quite so much rote stuff. Need to filter a collection on two values? You can write out you want do that in a comment line and wait a moment and it'll work that out for you no problem. Typing out some boilerplate that everyone ends up writing all the time ? It'll suggest what it should be, with full context of what names for your current project are. Would I trust it with complex stuff , the actual business logic ? Absolutly Not. You are the programmer, you are responsible and you should be solving the actual problems. But drat if it isn't nice if I can just press tab and copilot just fills in a whole complex object creation in half a second.
|
|
#
?
Mar 23, 2024 08:39
|
|
|
Tzarnal posted:Recently my work gave me github copilot. While I insist everyone should be actually reading what its producing there are some major advantages in just helping not write quite so much rote stuff. Need to filter a collection on two values? You can write out you want do that in a comment line and wait a moment and it'll work that out for you no problem. Typing out some boilerplate that everyone ends up writing all the time ? It'll suggest what it should be, with full context of what names for your current project are. How are you handling licenses from the code that Copilot uses to generate solutions? I'd imagine any GPL code that gets used to generate code makes it a sticky situation.
|
|
#
?
Mar 23, 2024 08:50
|
|
|
Dieting Hippo posted:How are you handling licenses from the code that Copilot uses to generate solutions? I'd imagine any GPL code that gets used to generate code makes it a sticky situation. Theoretically its Not Supposed To Do That. There is actually a toggle in the github project management options for Copilot that lets you toggle is such code is acceptable in a project. In practice I don't know. There certainly have been reports in the past that it was still causing issues with potentially restrictively licensed code being included. That is not really a thing I have control over at my job. Other solutions exists, as I understand it Codeium is generally recommended for not having been trained on any code with potentially licensing issues. Tzarnal fucked around with this message at 09:06 on Mar 23, 2024 |
|
#
?
Mar 23, 2024 09:04
|
|
|
|
| # ? Apr 29, 2024 16:26 |
|
|
Dieting Hippo posted:How are you handling licenses from the code that Copilot uses to generate solutions? I'd imagine any GPL code that gets used to generate code makes it a sticky situation.
|
|
#
?
Mar 23, 2024 14:22
|
|
