jjcm

10 ms·

181.
by jjcm 7mo ago
I ran it through a rudimentary thinking harness, and it still failed, fwiw: The question is about the best mode of transportation to a car wash located 100 meters away. Since the user is asking for a recommendation, it's importan
182.
by jjcm 7mo ago
> Was blazing fast in responding. I should note this is running on an RTX 6000 pro, so it's probably at the max speed you'll get for "consumer" hardware.
183.
by jjcm 7mo ago
Good call. Right now though traffic is low (1 req per min). With the speed of completion I should be able to handle ~100x that, but if the ngrok link doesn't work defo use the google colab link.
184.
by jjcm 7mo ago
Technically not in this case, or not effectively. The 0 or 1 correspond to a FP16 scaling factor for each group of 128 bits. The value fluctuates between each group of 128.
185.
by jjcm 7mo ago
1 bit with a FP16 scale factor every 128 bits. Fascinating that this works so well. I tried a few things with it. Got it driving Cursor, which in itself was impressive - it handled some tool usage. Via cursor I had it generate a few web pag
186.
by jjcm 7mo ago
It looks like it worked, fwiw. The qwen 27b model distilled on Opus 4.6 has some known issues with tool use specifically: https://x.com/KyleHessling1/status/2038695344339611783 Fascinating.
187.
by jjcm 7mo ago
I want to see the stats for this - which one gets chosen the most?
188.
by jjcm 7mo ago
I disagree with the author's interpretation of many of their points. > Skill Atrophy Very true, but in the same way moving from assembly -> scripting languages degraded the skill of programmers to manage memory. AI is another too
189.
by jjcm 7mo ago
These prediction markets shouldn't be allowed for situations where humans can control the outcome. They do have their place however - one I find particularly interesting is weather prediction markets, primary because they end up having
190.
by jjcm 7mo ago
Out of curiousity, why not distribute this as a webcomponent? It's a perfect use case for it - a semantic object that has built in controls / chrome.
191.
by jjcm 7mo ago
~4 years ago my wife and I planned on buying one EV and one ICE car purely as a hedge on both. We bought the EV first. After ~1mo of driving it we changed our plans and went for 2 EVs. Even driving long distance (we've made a ~3hr driv
192.
by jjcm 7mo ago
Two things stand out to me with this: 1. Drops compute required for training by ~20%. This approach wont just help the ever escalating model sizes larger companies are pushing for, it means things like autoresearch can iterate on new model
193.
by jjcm 7mo ago
This is fantastic, but IMO it misses the most important part of a home security system from a business PoV - the ability to issue an alarm certificate. These are required for insurance discounts, as well as for making certain claims in the
194.
by jjcm 7mo ago
I think this is likely, but in a slightly different way - I think we're going to start seeing more LLMs baked into silicon a la Taalas' ASIC. ie, something like this fake future apple device page: https://speculate-mai.
195.
by jjcm 7mo ago
There's a very simple fix I typically do when it comes to animations: animationCount = 0 animateElement(el) { el.animate({duration: BASE_DURATION / animationCount}) animationCount++ } (formula exag
196.
Nvidia DGX Station(nvidia.com)
6 points by jjcm 7mo ago|3 comments
197.
by jjcm 7mo ago
Grace Blackwell GB300 in a desktop. 748GB of memory. Estimated price is around $100k
198.
by jjcm 7mo ago
The bot problem is serious right now. I've switched to only allowing accounts that have paid at least once to post for my own network. It's a hard barrier (minimum spend is $2 for my site), but it almost completely solves the bot
199.
by jjcm 7mo ago
For now, there's simply no convention or established pattern for this yet. Given that, the speed of change, and the lower barrier of creation now, it's almost always better to create your own for this. Using someone else’s softwar
200.
by jjcm 7mo ago
This is a great step, if only because it enforces more convention for the "right" way to do things by providing a simpler mechanism for this. WRT WebAssembly Components though, I do wish they'd have gone with a different name
201.
by jjcm 8mo ago
"Instant" is really going to age poorly as far as a brand name goes, especially with Taalas ( https://chatjimmy.ai ) proving out that baked silicon models can be truly instant. I was literally posting about this earlie
202.
by jjcm 8mo ago
There's some solid advice in here - especially around performative interactions vs genuine. I was someone who was raised home schooled and it really altered my ability to communicate with my peers, which was something I had to really w
203.
by jjcm 8mo ago
Getting better, but definitely not there yet, nor near Sonnet 4.5 performance. What these open models are great for are for narrow, constrained domains, with good input/output examples. I typically use them for things like prompt expan
204.
by jjcm 8mo ago
I think this will work out nicely for Netflix. It overleverages their competition, putting them in a worse spot, and I suspect traditional media is going to be far less of an demand from a pure screen-time perspective in the next 10 years.
205.
by jjcm 8mo ago
This is the strongest statement in the post: > They have threatened to remove us from their systems if we maintain these safeguards; they have also threatened to designate us a “supply chain risk”—a label reserved for US adversaries, nev
206.
by jjcm 8mo ago
As someone currently working on their own VAE, you reasoning for why you went with WAN 2.1 and your learnings for what you think you did wrong really resonated with me, specifically: > Looking back, we should have just filtered out these
207.
by jjcm 8mo ago
What I really want is a <sandbox> element that can safely run dangerous code, not something that modifies dangerous code. Iframes have significant restrictions as they can’t flow with the DOM. With AI and the increase in dynamic conte
208.
by jjcm 8mo ago
A lot of naysayers in the comments, but there are so many uses for non-frontier models. The proof of this is in the openrouter activity graph for llama 3.1: https://openrouter.ai/meta-llama/llama-3.1-8b-instruct/ac
209.
by jjcm 8mo ago
big +1 on #1. As a tip - most LLMs are unaware it exists due to either knowledge cutoff or not having enough training data for it. As a recommendation, include some examples of it in your AGENTS.md. Here's what I use: -----------------
210.
by jjcm 8mo ago
This is entirely tangential to the article, but I’ve been coding in golang now going on 5 years. For four of those years, I was a reluctant user. In the last year I’ve grown to love golang for backend web work. I find it to be one of the

More ›