HNHacker News
TopNewBestAskShowJobs

robotswantdata

1,075 karma · joined October 15, 2024

submissionscomments
robotswantdata··on How to Train a Gen AI Kick Drum Model on Your Old Linux Desktop with 6GB VRAM
Good modelling
robotswantdata··on How to Train a Gen AI Kick Drum Model on Your Old Linux Desktop with 6GB VRAM
Confused. Why not just make the kick drum from a sine? Seconds
robotswantdata··on Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU
Need to run this on my Xeons with AMX
robotswantdata··on Codex Micro
Ordered. Not sure will beat my streamdeck modules, but YOLO
robotswantdata··on Ask HN: What Are You Working On? (July 2026)
Novel app for live music performance (private for me and another big artist, we may release one day). Very unique
robotswantdata··on Hands-On with the AMD Ryzen AI Halo
12 channel DDR5 6400, 614GB/s peak theoretical
robotswantdata··on Apple Silicon Exec Explains Mac Mini AI Demand and On-Device Future
build a Xeon / epyc 4u server. 12 channel ram.
robotswantdata··on Ask HN: Another "Hacker News" with less AI and more human-focused hacking news?
comp.lang.c
robotswantdata··on GPT‑Live
Any plans to add more international voices and dialects?
robotswantdata··on Small AI Models Gain Traction In places with unreliable networks
https://developers.google.com/edge/gallery

Put that on a spare phone

robotswantdata··on AMD Ryzen AI Halo – $4k AI Dev Kit
Was “only” $2k in its previous form but even in this updated box the mem bandwidth is woefully inadequate. There’s a few models with space for a dedicated GPU for hybrid inference but imo not worth it. Save your money for a Xeon or EPYC build
robotswantdata··on Claude Design System Prompt
Agree, it’s vibe slopped rather than the actual Claude design system prompt
robotswantdata··on Costco is the anti-Amazon
You can put tailwind on the CV
robotswantdata··on 14× faster embeddings: how we rebuilt the ONNX path in Manticore
Intel AMX is very underrated too

https://aws.amazon.com/blogs/compute/accelerate-cpu-based-ai...

robotswantdata··on Wallace the 6 inch f/2.8 telescope, building it, and hiking with it
Not good for planets IMO as 150mm focal length and a small 30mm aperture. Binoculars or small desktop DOB is better with kids for planets. The mini is amazing for deep sky objects for the price
robotswantdata··on Wallace the 6 inch f/2.8 telescope, building it, and hiking with it
Very cool project. Been taking my dwarflab mini on hikes instead of the 12” dob ;) , not the same at all but enjoy getting dark sky and some great astro shots.
robotswantdata··on The truth about being a manager
I edited, but in big mega corps could easily be 5x your junior staff.
robotswantdata··on The truth about being a manager
It doesn’t cover the reality that nearly all your extra pay in a progressive tax system goes to the government. UK is hitting 70% marginal rates , or even 1000% for people with children.
robotswantdata··on The truth about being a manager
Missed the part about 100x the stress and politics for 1.2x the pay.
robotswantdata··on Unlimited OCR: One-shot long-horizon parsing
Very interesting.

The way I understand this works is that the researchers found a clever architectural hack to stop AI from hoarding memory when reading long documents.

Normally, when an AI transcribes a 100 page PDF, it tries to remember every single word it has already ingested. This short-term memory (the KV cache) grows linearly O(N) until the model runs out of VRAM and crashes (or caps it) To avoid this, developers are forced to build janky code that chops PDFs into individual pages, processes them one by one, and glues the text back together.

Unlimited OCR uses Reference Sliding Window Attention (R-SWA) to split the AI's focus into two paths:

Global Reference: The AI keeps full, uncompromised sight of the original document image so it never loses context.

Local Generation: The AI restricts its memory of its own typed text to a tight, moving window (like the last 128 words) and safely forgets the rest.

Will be very interesting for local AI and can’t wait to see what the community builds and extends with it!

robotswantdata··on A 10 year old Xeon is all you need
Granite or sapphire rapids are very under rated for MoE inference loads. But you need a GPU for the KV cache.

Plus many boards also support CXL for RAM expansion over PCI 5!

Source: building a hybrid inference business for regulated industry workloads.

robotswantdata··on MCP is dead?
I agree, codex app’s computer use agent feels sci-fi. Closest we’ve seen yet of a general purpose virtual worker.
robotswantdata··on I'm Getting into Mesh Networks (Meshtastic, MeshCore, and Reticulum)
Zombies
robotswantdata··on I'm Getting into Mesh Networks (Meshtastic, MeshCore, and Reticulum)
Wrong antenna for the frequency?
robotswantdata··on I'm Getting into Mesh Networks (Meshtastic, MeshCore, and Reticulum)
Set up solar nodes last weekend. 200 miles of range now. Nerds, mad ideas. Good times.
robotswantdata··on I'm Getting into Mesh Networks (Meshtastic, MeshCore, and Reticulum)
This is by design. It’s like BBS again
robotswantdata··on 'AI washing': firms are scrambling to rebrand themselves as tech-focused
BPM is now agentic, don’t you know.
robotswantdata··on A scoping review of bicycling interventions’ impacts on well-being
Don’t do PEDs
robotswantdata··on Mozilla to UK regulators: VPNs are essential privacy and security tools
1984 was meant to be a warning, not the UK’s digital infrastructure roadmap
robotswantdata··on I believe there are entire companies right now under AI psychosis
Most labs are shilling “AI worker” dreams to these very companies
← PreviousPage 2 of 6Next →