Hacker News: alfiedotwtf

New comment by alfiedotwtf in "How to make your text look futuristic (2016)"

alfiedotwtf — Wed, 13 May 2026 11:20:15 +0000

Not only that, this article had the same feel of an old ANSI scene graffiti tutorial (I think it was made by the ACiD team

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Tue, 12 May 2026 02:02:29 +0000

Ok, you’ve given me the umph to try again. Thanks!

New comment by alfiedotwtf in "If AI writes your code, why use Python?"

alfiedotwtf — Tue, 12 May 2026 02:00:14 +0000

… because model tool calls is non-standard, so Python as the only tool call available works wonders

(Joke but also not a joke)

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 22:49:14 +0000

This might be my only hope until RAM prices come down to human levels again

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 22:48:27 +0000

Ah dang. Hmm, damn this hobby is expensive. Maybe I should just take up drugs instead

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 17:20:44 +0000

Unfortunately I didn't get a Mac with big ram at the time it was cheap, and I'd personally focus on moving away from Apple and going Linux fulltime at work and home (currently Macbook for laptop connected to my big rig, well it's not that big compared to the AI people in here).

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 17:19:09 +0000

I've got a 4060 Ti 12Gb with 128Gb RAM. I was hoping once I could demonstrate to myself that I could run Deepseek v4 Flash locally (even at really slow speeds), then it would be worth my time and money to get something to run it > 20t/s.

... currently testing out Stepfun 3.5 Flash Q4_k_m as a stop gap (unless it blows my socks off first).

New comment by alfiedotwtf in "Running local models on an M4 with 24GB memory"

alfiedotwtf — Mon, 11 May 2026 17:06:34 +0000

What's your opinion with Gemma 4 vs Qwen3.6?

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 16:08:27 +0000

Oh interesting. I've read that Gemma 4 is really good for creative stuff, but I'm mostly interested in agentic coding. Unfortunately, each time I use Gemma 4, I just get it stuck in loops.

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 16:07:16 +0000

I've been keeping an eye on Antirez's Metal fork for llama.cpp, but I totally missed this. Whoa, nice. Giving it a go, thanks!!

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 06:48:29 +0000

… maybe we should just teach models how to get their world knowledge from a local Postgres connection! Then the model can be tiny, and it can query to its little heart desires AND run on commodity hardware TODAY!

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 06:44:25 +0000

If 8 x RTX 6000 is getting you 20s before initial token, how are cloud vendors doing this?

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 06:37:44 +0000

What is everyone running DeepSeek v4 Flash with?!

It’s currently unsupported on Llama.cpp and vllm doesn’t support GPU+CPU MoE, so unless all of you have an array of DGX Sparks in your bedroom, what’s the secret sauce?!

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 06:10:34 +0000

I’m guessing Qwen3.6 for agentic coding and Gemma4 for non-coding stuff?

New comment by alfiedotwtf in "Local AI needs to be the norm"

alfiedotwtf — Mon, 11 May 2026 06:05:09 +0000

This would be nice, but unfortunately the norm at the moment is - release a rushed model that doesn’t work with llama.cpp, but if it does, make sure that the chat template is broken. And even if it did have a perfect chat template, let the model loop endlessly rewriting the same file with same content for hours on end.

It would be nice if model makers could at minimum embrace test harnesses, and stretch goal if they’re going to change underlying formats then at least land compatible readers in the big engines (e.g. llama.cpp and vllm)

New comment by alfiedotwtf in "I’ve banned query strings"

alfiedotwtf — Sun, 10 May 2026 00:22:03 +0000

It may be ridiculous to some, but Chris is totally right - it’s his website and he can do what he wants with it!

New comment by alfiedotwtf in "Accelerating Gemma 4: faster inference with multi-token prediction drafters"

alfiedotwtf — Wed, 06 May 2026 17:42:10 +0000

Maybe it’s just me, but I feel like the LLM crowd are re-discovering Coding and Compression all over again.

New comment by alfiedotwtf in "StarFighter 16-Inch"

alfiedotwtf — Wed, 06 May 2026 17:01:58 +0000

Ask Kim Dot Com about that myth

New comment by alfiedotwtf in "IBM didn't want Microsoft to use the Tab key to move between dialog fields"

alfiedotwtf — Tue, 05 May 2026 22:03:15 +0000

If you have ever used Xero.com to fill out timesheets you’ll personally know that they adhere to IBM’s strict tab standards

New comment by alfiedotwtf in "Agent Skills"

alfiedotwtf — Tue, 05 May 2026 16:46:30 +0000

A lover of Superpowers here, using it for about 2+ months now.

It allows you to explore the problem space upfront, it questions your assumptions, asks more probing questions to confirm what it’s found in the code, and by the time you’re ready to implement, it knows exactly what needs to happen.

Jessie should have called it Socrate’s Methods