NitpickLawyer

26 ms·

571.
by NitpickLawyer 9mo ago
> best of N models like deep think an gpt pro Yeah, these are made possible largely by better use at high context lengths. You also need a step that gathers all the Ns and selects the best ideas / parts and compiles the final output
572.
by NitpickLawyer 9mo ago
> Outsourcing thinking is bad. I keep seeing this and I don't think I agree. We outsource thinking everyday. Companies do this everyday. I don't study weather myself, I check an app and bring an umbrella if it says it's go
573.
by NitpickLawyer 9mo ago
> The result is OK. It has all the features I asked for, and includes document sharing, collaborative editing in real time, support for fonts and line spacing, etc. etc. I could not have paid a developer $170 and got this. The problem, o
574.
by NitpickLawyer 9mo ago
Just don't ask it to produce paperclips...
575.
by NitpickLawyer 9mo ago
The difference is in scaling. The top US labs have oom more compute available than chinese labs. The difference in general tasks is obvious once you use them. It used to be said that open models are ~6mo behind SotA a year go, but with the
576.
by NitpickLawyer 9mo ago
Cool, thanks. Did you try it out, how's the performance? I saw on openrouter that the stealth model was served at ~19t/s. Is it any better on their endpoints?
577.
by NitpickLawyer 9mo ago
> It's avaiable in mine Weird, mine (lite plan) says "Only supports GLM-4.7, GLM-4.6, GLM-4.5, and GLM-4.5-Air" and "Get same-tier model updates" ...
578.
by NitpickLawyer 9mo ago
> Tier 1 transit providers doing port filtering is EXTREMELY alarming. I was admining a small ISP when blaster and its variants hit. Port filtering 139 and the rest was the easiest way to deal with it, and almost over night most of the I
579.
by NitpickLawyer 9mo ago
> [...] prior to reforming society [...] Well, good luck. You have "only" the entire history of human kind on the other side of your argument :)
580.
by NitpickLawyer 9mo ago
#WeDidItMoltbook
581.
by NitpickLawyer 9mo ago
mistral.rs is more like llama.cpp, it's a full inference library written in rust that supports a ton of models and many hardware architectures, not just mistral models.
582.
by NitpickLawyer 9mo ago
There are some good things here: First, we currently have 4 frontier labs, and a bunch of 2nd tier ones following. The fact that we don't have just oAI or just Anthropic or just Google is good in the general sense, I would say. The 4 l
583.
by NitpickLawyer 9mo ago
We have something like that in eu with road markings. Both for clear weather and fog/rain. They mark some of the lines differently, and tell you how many lines you should have between you and the car in front. I think they were first t
584.
by NitpickLawyer 9mo ago
You have no idea what alphaevolve is, yet you try to correct me. This isn't productive, I'm out.
585.
by NitpickLawyer 9mo ago
> soon to be outdated assumptions Wild assumption. Having docs and code in step has never been easier. > soon-to-be-outdated living and constantly modified AGENT file Quite contradictory. > I have absolutely seen far better initial
586.
by NitpickLawyer 9mo ago
> The thing that keeps coming up is that the "anti" crowd is fighting their own deamons, and have kinda lost the plot along the way perfect example
587.
by NitpickLawyer 9mo ago
I have no idea what you're arguing. Alphaevolve is similar to claude code. They are using LLMs in a harness. No idea what you mean with fish, kidneys and so on. Can you please stick to the technical stuff? Otherwise it's just nois
588.
by NitpickLawyer 9mo ago
> First off Alpha Evolve isn't an LLM. It's a system based on LLMs. > AlphaEvolve, an evolutionary coding agent powered by large language models for general-purpose algorithm discovery and optimization. AlphaEvolve pairs th
589.
by NitpickLawyer 9mo ago
Two thoughts here: First, remember when we had LLMs run optimisation passes last year? Alphaevolve doing square packing, and optimising ML kernels? The "anti" crowd was like "well, of course it can automatically optimise some
590.
by NitpickLawyer 9mo ago
> It's crazy how emotionally attached people are to 4o If oAI numbers are to be trusted, there are ~800k people who still use 4o every day.
591.
by NitpickLawyer 9mo ago
> it's not like it just flips. Does this apply to that cool chem trick where a solution goes from black to transparent and back again a few times? I don't know enough to know if that's relevant or not, but I remember seein
592.
by NitpickLawyer 9mo ago
There's an interesting phenomenon I noticed with the "skeptics". They're constantly using what-ifs (aka goalpost moving), but the interesting thing is that those exact same what-ifs were "solved" earlier, but d
593.
by NitpickLawyer 9mo ago
> hiring is difficult and high-end talent is limited. Not only that, but firing talent is also a pain. You can't "hire" 10 devs for 2 weeks, and fire them afterwards. At least you can't keep doing that, people talk
594.
by NitpickLawyer 9mo ago
> It also doesn't work to penalize fraudulent warnings How would you classify fraudulent warnings? "Hey chatgpt, does this text look good to you? LGTM. Ship it".
595.
by NitpickLawyer 9mo ago
And how long will it take before an open model recreates this. The "vibe" consensus before "thinking" models really took off was that open was ~6mo behind SotA. With the massive RL improvements, over the past 6 months I&
596.
by NitpickLawyer 9mo ago
We've been hearing this for 3 years now. And especially 25 was full of "they've hit a wall, no more data, running out of data, plateau this, saturated that". And yet, here we are. Models keep on getting better, at more b
597.
by NitpickLawyer 9mo ago
> You can tell you're in trouble on this thread when everybody starts bringing up the curl bug bounty. I don't know if this is surprising news for people who don't keep up with vuln research, but Daniel Stenberg's cur
598.
by NitpickLawyer 9mo ago
> Prove this statement wrong. If all it takes is "trained on the Internet" and "decompress stored knowledge", then surely gpt3, 3.5, 4, 4.1, 4o, o1, o3, o4, 5, 5.1, 5.x should have been able to do it, right? Claude 2,
599.
by NitpickLawyer 9mo ago
It's a bit disappointing that people are still re-hashing the same "it's in the training data" old thing from 3 years ago. It's not like any LLM could 1for1 regurgitate millions of LoC from any training set... Thi
600.
by NitpickLawyer 9mo ago
This is a much more reasonable take than the cursor-browser thing. A few things that make it pretty impressive: > This was a clean-room implementation (Claude did not have internet access at any point during its development); it depends

More ›