ai-christianson

9 ms·

91.
by ai-christianson 1y ago
I like the buzz saw analogy better than the sawdust for oil analogy :)
92.
by ai-christianson 1y ago
I'm the dev of a semi-popular FOSS coding agent --I agree that "vibe coding" as a term is fairly cringe. The thing is, I see AI coding tools, especially agents, as a major force multiplier for individual devs and small teams.
93.
by ai-christianson 1y ago
I think most of the thread is talking about SSR with partial HTML replacement.
94.
by ai-christianson 1y ago
There are AI models that can generate 3D models, e.g. Hunyuan3D. Not quite CAD models, but maybe this could eventually be adapted to that use case. Then there's the possibility of an agent automating an actual CAD program. This has alr
95.
by ai-christianson 1y ago
Presumably I can install this (the web frontend) into k8s itself. Is there a helm chart or kustomize? Would be really cool to install it into k8s and just hit a hosted web endpoint with all the logs and grep/exploration capabilities ku
96.
by ai-christianson 1y ago
This looks awesome --do you cache or store the logs, or is that left up to k8s?
97.
by ai-christianson 1y ago
Very cool. We're using helm currently for our startup. It works, but I'm not super excited about it. We've also played with deploying k8s resources with ansible. It works better than expected, but is a bit clunky. We've
98.
by ai-christianson 1y ago
> So I rewrote the whole thing from scratch using Python So this isn't really codex then?
99.
by ai-christianson 2y ago
Why Kotlin?
100.
by ai-christianson 2y ago
I'm developing an open source coding agent (RA.Aid). I'm using RA.Aid to develop itself (dogfooding,) so I'm constantly running the coding agent. That cost is my peak cost, not average. It's easy to scale back costs to 1
101.
by ai-christianson 2y ago
It's really easy to get to $100 in a day using sonnet 3.7 or o3 in a coding agent. Do that every day for a month and you're already at $3k/month. It's not hard to get to $5k from there.
102.
by ai-christianson 2y ago
Deepseek are the kings currently.
103.
by ai-christianson 2y ago
I've ended up with $5K+ in a month using sonnet 3.7, had to dial it back. I'm much happier with gemini 2.5 pro right now for high performance at a much more reasonable cost (primarily using with RA.Aid, but I've tried it with
104.
by ai-christianson 2y ago
Am Saltman
105.
by ai-christianson 2y ago
Works well in RA.Aid --in fact I'd recommend it as the default model in terms of overall cost and capability.
106.
by ai-christianson 2y ago
> Interesting... I asked o3 for help writing... What tool were you using for this?
107.
by ai-christianson 2y ago
> rather than an economically rational decision Depends on the time horizon
108.
by ai-christianson 2y ago
IMO I much prefer to have my dev tools be completely FOSS so I'm not building my career on skills that are tied to a proprietary sw provider.
109.
by ai-christianson 2y ago
This outage gets me thinking of alternatives. It's nice to have just a folder of music that is completely offline.
110.
by ai-christianson 2y ago
Many OSS contributors are on the payroll of a company --or is that outside the scope of what you're looking for?
111.
by ai-christianson 2y ago
> I would start by making the examples yourself initially What I'm doing right now is this: 1) I have X problem to solve using the coding agent. 2) I ask the agent to do X 3) I use my own brain: did the agent do it correctly
112.
by ai-christianson 2y ago
> The most important thing is doing rigorous evals that are representative of what your users do in your product. Often this is not the same as academic benchmarks. OMFG thank you for saying this. As a core contributor to RA.Aid, optimiz
113.
by ai-christianson 2y ago
Right. You could even run it through code obfuscators and such to create more diverse, realistic examples.
114.
by ai-christianson 2y ago
It's performing well on the aider benchmark too. This one is legit.
115.
by ai-christianson 2y ago
In RA.Aid ( https://github.com/ai-christianson/RA.Aid ) we support normal tool calling, but we also have a our own agent backend called Code Is All You Need. The way it works is we prompt the LLM to generate code to call
116.
by ai-christianson 2y ago
Core contributor of RA.Aid here. We are up to 15 total contributors now and are aiming to be one of the top completely FOSS coding agents. Cool to see we're on your list! Curious to hear feedback on it!
117.
by ai-christianson 2y ago
MLM... uh oh
118.
by ai-christianson 2y ago
> You just copy data over PCIe AFAIK this only happens directly over PCIe if using hacked drivers with p2p enabled (I think tinygrad/tinybox provided these drivers initially.) Otherwise, data goes through system bus/CPU first.
119.
by ai-christianson 2y ago
> I actually didn't think this would work at all without nvlink. It will work, but with less memory bandwidth. If using hacked drivers with p2p enabled, it will depend on the PCIe topology. Otherwise, the data will take a longer pat
120.
by ai-christianson 2y ago
> However today, with the latest, surprisingly good reasoning models like QwQ-32B using up thousands or tens of thousands of tokens in their replies, performance is getting more important than previously and these systems (Macs and even

More ›