tcdent

6 ms·

61.
by tcdent 6mo ago
I appreciate the effort you put into mapping semantics so language constructs can be incorporated into this. You’re probably already seeing that the amount of terminology, how those terms interact with each other, and the way you need to mo
62.
by tcdent 6mo ago
Exactly this. It’s counterintuitive for most people, but the more complexity you add to the systems (the more organic they are), the more sustainably successful they become. Everyone is looking for a simple solution, but simple solutions do
63.
by tcdent 6mo ago
Nah, taking the risk is even more fun when the thing you're modifying holds more value. Chopping the fenders on a Porsche 911 to install a widebody kit does not have the same weight as rolling the seams on an Jeep Cherokee.
64.
by tcdent 6mo ago
I don't think you're wrong, but if you really want to really re-think the approach, building an orchestration layer for Firecracker like every other company in the space is doing is probably not it.
65.
by tcdent 6mo ago
This is a fantastic idea. On a nonzero number of occasions I have priced the cost of running an inference server with a model that is actually usable and the annual cost is astronomical.
66.
by tcdent 6mo ago
And they said turning Lead into Gold was just heresy.
67.
What's Cch? Reverse Engineering Claude Code's Request Signing(a10k.co)
4 points by tcdent 6mo ago|0 comments
68.
by tcdent 6mo ago
5.5 min to train on a PDP/11 you mean to tell me we could have been doing this all along???
69.
by tcdent 7mo ago
The replies lol. "Yes" Proceeds to talk about AI.
70.
by tcdent 7mo ago
DSPy is cool from an integrated perspective but as someone who extensively develops agents, there have been two phases to the workflow that prevented me from adopting it: 1. Up until about six months ago, modifying prompts by hand and incor
71.
by tcdent 7mo ago
Communism. I'm talking about communists.
72.
by tcdent 7mo ago
Not worth it. It is a very significant performance hit. With that said, people are trying to extend VRAM into system RAM or even NVMe storage, but as soon as you hit the PCI bus with the high bandwidth layers like KV cache, you eliminate a
73.
by tcdent 7mo ago
I'm kindof surprised by this take. Did you think the hesitancy of westerners engaging and relying on Chinese labs was due to vibes? There are fundamental cultural differences at play, wether we are comfortable admitting that or not.
74.
by tcdent 7mo ago
I still find it revolting they're writing this stuff in typescript.
75.
by tcdent 8mo ago
Just like you can read source code written by humans (and should if you take this stance) you can also read source code generated by LLMs. Then, when you find something unsavory and feel that your sentiment is warranted, make a contribution
76.
by tcdent 8mo ago
Inference is run on shared hardware already, so they're not giving you the full bandwidth of the system by default. This most likely just allocates more resources to your request.
77.
by tcdent 8mo ago
Major unlock I see coming with the advent of AI is the ability to access knowledge that has dropped out of the sociosphere because of obscurity and language barriers. This is a cool move in the right direction.
78.
by tcdent 8mo ago
Is this even a problem that needs to be solved? How many people have 3d printed guns and used them? Preemptive regulation is absurd.
79.
by tcdent 8mo ago
Skills are for the most part already generated by LLMs. And, if you're implementing them in your own workflow, they're tailored to real-world problems you've encountered. Having a super repo of everyone else's slop is ba
80.
by tcdent 8mo ago
It's super exciting to see another American lab get in the ring. Even if they're not at SOTA on the first release, the fact that they're trying is incredible for open source AI.
81.
by tcdent 8mo ago
Nobody here seems to remember that this was always the plan: release expensive cars to bootstrap the company which allows them to release progressively cheaper cars until everyone can afford one. Not a fanboy, but this seems like it went ex
82.
by tcdent 8mo ago
Could have just called it "clawbot" and maintained some of the hype while eliminating the IP concerns. Instead they chose a completely different name with unrecognizable resonance.
83.
by tcdent 9mo ago
I spent a couple years doing physical stuff professionally. Pro audio system design and install, commercial interior design and fabrication, event production. These pulled from skills I learned from hobbies I did to get away from programmin
84.
by tcdent 9mo ago
Agreed. Seems like a pretty risky optimization that fundamentally changed behavior; like it or not the ordering of vectors is often part of the data structure. Could have just used a prepend to preserve behavior instead pf going down the ra
85.
by tcdent 9mo ago
Furthermore, all of the major LLM APIs reward you for re-sending the same context with only appended data in the form of lower token costs (caching). There may be a day when we retroactively edit context, but the system in it's current
86.
I wish rust had keyword arguments(github.com)
1 points by tcdent 9mo ago|0 comments
87.
by tcdent 9mo ago
This whole thing got blown out of proportion because the devs of third party harnesses that use the oauth API never disclosed that they were already actively sidestepping what is a very obvious message that the oauth API is for Claude Code
88.
by tcdent 9mo ago
This is a fairly well written article which captures the current state of the art correctly. And then goes on to recommend AI Studio is a primary dev tool?! Baffling.
89.
by tcdent 10mo ago
Social movements don't need to be quantifiably better to take off. When the relevant audience is bored enough to be open to something new, it only takes a few influential people to tip the scales. People don't want to be truly rev
90.
by tcdent 10mo ago
This one? https://artificialanalysis.ai/models/gemma-3-27b

More ›