mlazos

How can you be hitting cache on what I think are novel LLM prompts …

Yeah but we still don’t have execution vans! I’m not even being sarcastic. China is unequivocally worse even with our recent slide into towards totalitarianism.

In the 2010s all of my exams were still hand written and I went to Brown.. has everything really been computerized that much?

I think the tricky part here is essays, but maybe they should be replaced with an oral exam + essays? A lot of the humanities and classics courses I took were pretty small.

Overall I don’t think these problems are that hard, I just don’t know if the incentives of universities align with fixing them. If their reputation is all that it’s for, grades don’t really matter as long as the $$$ is coming in.

Wtf? What a joke, but I mean the best way to become a billionaire is convince someone with a billion dollars to give it to you. This is actually insane, wow. I guess Qualcomm is desperate? Nobody was bidding for this, but congrats to the team at modular?! I’m actually salty about this because like I don’t feel like mojo was even good after trying it out.

I work on PyTorch torch.compile and it’s a tracing compiler as well. Perhaps this domain is very narrow though; it is also a very not conventional compiler, you’d probably be deeply offended by some of the stuff we do! ;)

The idea of them purposefully wasting my time by having the model act dumber and me having to argue with it without knowing if it’s the prompt or the model was just such an idiotic product decision I can’t believe they shipped that without getting any feedback from users first.

Ah, the ever present “nothing to see here” take. What this government is doing is worse than it’s ever been. At least before when you had your grant it wouldn’t be randomly cancelled at any time.

I believe them to some degree but this trend of posting stuff when it can’t be verified actually needs to end. I’m so tired of this bs marketing.

Oh weird, everything must be fine then, I still see snow sometimes even!

I think one day later the guy got his refund? You all need to chill I feel like. HN is a bubble sometimes

https://github.com/anthropics/claude-code/issues/53262#issue...

By the letter of the law the guy fixing his car should be prosecuted, but like nobody is going to know and it’s not going to happen. In this case it’s pretty obvious the law was broken.

Look! A guy built 95% of slack in 2 weeks! Very skeptical of that btw, but also an organization that justifies every single team by exactly how much $ value they’re generating sounds like hell. How would you ever innovate or try out new ideas? It’s important to quantify what impact your team is generating but there are some cases (e.g. UX) which are really hard to quantify in $ but are still very important for the product

Idk I think they kind of do, tackling climate change is a good idea and at this point I’m ready to lie if it will make ppl do something

Honestly people spoke like this before, just on LinkedIn. Now that ai trained on it we have LinkedIn.. everywhere. Welcome to hell.

I read the tests, it also is really really good to have Claude verify that removing the changes in question break the tests. This brings the quality way way up for me.

Randomness does not imply the uniform distribution

Bro nothing is verifiable, you have to make a call in the end or you can just sit here and be like “oh but what about the speck of sand on the body, GOTCHA”. I used to be like you trying to poke holes in every little thing, but 1 it’s annoying to everyone and 2 you have to eventually take a stance other than “every stance is wrong” your mystical perfectly evidenced war crime is never going to exist, congrats! What do you think actually happened here? The article cites plenty of evidence, eyewitness testimony, make the call. The IDF themselves said they killed them. It’s a family, what do you think justifies that?

Seen this on repeat lately - there will be some war crime that the IDF commits, soldiers or Israeli citizens celebrate it themselves in a TikTok or in Israeli media, then the US media will argue that it didn’t happen or “there’s some information missing”. It’s actually kind of nuts.

Yes driving fast means - execution.

Maybe I’m getting old but I couldn’t tell if this was a joke or not. I think they should explain a little more like what these products actually do?

That’s 100% insider trading. If you use material non-public (including confidential) information to perform the trade it’s illegal. Paying someone to provide confidential info is still insider trading. Paying someone to observe planes (public information) is not insider trading. Researching using publicly available information (even though you don’t share your research) is not insider trading. The key point is the channel from which you receive the information.

This is the absolute biggest grift of the century by the groq team. They never shared actual TCO, and I remember a Seminalaysis article about the power consumption being actually insane - this makes sense because they scale the number of chips to fit a single model when they have no dram. They have good inference latency but there was no way the economics were going to work out. Meanwhile Nvidia with every advantage in the world decides they’re worth 20B? It actually doesn’t make sense at all. The only scenarios the groq system would be worth it is in the exact throughput-optimized scenarios Nvidia already thrives in.

If the neo-nazis celebrate it as a Nazi salute, I don’t think it really matters what others think. In addition to that, seeing it with my own eyes is all the confirmation I need.

On his intelligence SpaceX and Tesla were/are revolutionary companies, but seeing him buy twitter and then send the DOGE five things email makes me feel like maybe it was more of a right place right time sort of thing and not management prowess. I’ll give him credit for hiring the right people which is a skill but other than that his blunders are just too difficult to ignore.

It’s because capitalism assumes a free market with competition. If you allow monopolies to thrive, you will not get those benefits. It’s just that some of these types of markets have different dynamics due to their structure. E.g. natural monopolies where the barrier to entry is huge up front costs. Interestingly the AI startup ecosystem is raising enough money to surpass the barrier of needing a ton of data to train AI.

Yeah where they have every inch of SF mapped, and then still have human interventions. We were promised no more human drivers like 5-7 years ago at this point.

It supports e5m2 and e4m3 right in the doc linked.

It just seems like the rating is a vote. You’d end up with the same problems.

Everyone always thinks this at least in big tech I’ve never heard a PM or exec say a market is not winner take all. It’s some weird corpo grift lang that nothing is worth doing unless its winner take all.

I’m amazed this is even viewed as a “hot take” tbh most of what he said here is pretty high level of abstraction and standard practice for custom hardware. In essence I feel like he’s saying nothing really controversial other than publicly calling out TT for too many abstraction layers (and tbh it’s just in a readme). This is completely fine, he’s a user and this is his experience.

I’m a dev working on torch.compile at meta (previously I worked on ML focused FPGAs) and the approach I would use is build a static graph compiler, use torch.compile (and probably JAX) as graph extraction front-ends and call it a day. I feel like hardware companies don’t know how to handle the flexibility of PyTorch and as a result develop their own APIs which is mistake #1 and virtually makes it impossible to get any market penetration once you head down that path because nobody will ever ever rewrite their models for your hardware when they don’t even know what perf they will get, the risk is just too high. As a result, hardware companies offer inference APIs which hide all of this behind a REST API to basically paper over the lack of generality of the software/hardware interface. This is convenient because then nobody actually knows the perf/$ and they can burn VC money for as long as they want. Whether this is a viable business model or not, we will have to wait until they go public to actually see what their true inference costs are.

To sum it up, start from PyTorch and work your way down to your hardware, this is the only general way if you want to actually sell chips and not just constantly port the model of the day to your hardware.

The idea that cutting research will make a dent in the budget is a fantasy. NSF has a budget of 10 billion. Stop rationalizing gutting crucial programs because of “the deficit” Medicare, social security and the military are the main costs in the US budget. Sure universities are bloated, tackle that problem separately then.

I used this to onboard to the PyTorch team a few years ago. It’s useful for understanding the key concepts of the framework. Torch.compile isn’t covered but the rest of it is still pretty relevant.