Previously, I co-founded SpeakerText (speakertext.com), a interactive/social layer on top of online video, and Humanoid (gethumanoid.com), a crowdsourced virtual labor service that uses machine learning to track worker reputation on Mechanical Turk and ensure quality.
In previous lives, I fought forest fires in Montana & Alaska for the US Forest Service, worked as a freelance news reporter for the NY Times, and drove a 911 ambulance in Harlem and the South Bronx.
Submissions
Show HN: Local Motion – Use Cursor Agents and Chat with a Local LLM
github.com
1 points·by MediaSquirrel··0 comments
Show HN: Gist Discover – TikTok for ArXiv Summaries
gist.is
4 points·by MediaSquirrel··1 comments
Scalable GANs with Transformers
arxiv.org
3 points·by MediaSquirrel··0 comments
Cheaper Than Concrete: Robots and the New Stone Age
originals.is
4 points·by MediaSquirrel··0 comments
Lift4D: Harmonizing Single-View 3D Estimation for 4D Reconstruction In-the-Wild
arxiv.org
3 points·by MediaSquirrel··0 comments
Death Is an Engineering Problem
originals.is
3 points·by MediaSquirrel··1 comments
Non-frontal face recognition using GANs and memristor-based classifiers
arxiv.org
2 points·by MediaSquirrel··0 comments
MambAdapter: Lightweight Mamba-Based Adapters for Transfer Learning
arxiv.org
2 points·by MediaSquirrel··0 comments
Can I Buy Your KV Cache?
arxiv.org
36 points·by MediaSquirrel··28 comments
Before You Think: System 0, AI-Mediated Cognition and Cognitive Colonization
arxiv.org
1 points·by MediaSquirrel··0 comments
Reasoning as Pattern Matching: Shared Mechanisms in Human and LLM Reasoning
arxiv.org
1 points·by MediaSquirrel··0 comments
Superficial Beliefs in LLM Decision-Making
arxiv.org
3 points·by MediaSquirrel··0 comments
Show HN: Magenta Real-Time Music Generation Locally on iPhone, Without the GPU
github.com
9 points·by MediaSquirrel··0 comments
Feedback Alignment in Self-Distillation
arxiv.org
2 points·by MediaSquirrel··0 comments
WWDC 2026 – On-Device AI Deep Dive
gist.is
1 points·by MediaSquirrel··0 comments
Show HN: Gemma 4 Multimodal Fine-Tuner for Apple Silicon
github.com
235 points·by MediaSquirrel··28 comments
We Melted iPhones for Science – Generating Real-Time Video with On-Device AI
accelerateordie.com
1 points·by MediaSquirrel··0 comments
[untitled]
1 points·by MediaSquirrel··0 comments
We Melted iPhones for Science
accelerateordie.com
67 points·by MediaSquirrel··44 comments
God Mode – The Brave New World of Agentic Management (Prompts Included)
Dr. Alex Mathiasen, PhD wants to build a pause button for human biology – so that you can live forever. All he needs is enough GPUs to simulate quantum physics.
If you run a smaller whisper-distil variant AND you optimize the decoder to run on Apple Neural Engine, you can get latency down to ~300ms without any backend infra.
The issue is that the smaller models tend to suck, which is why the fine-tuning is valuable.
My hypothesis is that you can distill a giant model like Gemini into a tiny distilled whisper model.
but it depends on the machina you are running, which is why local AI is a PITA.
Here’s the trick: use Gemini Pro deep research to create “Advanced Hacker’s Field Guide for X” where X is the problem that you are trying to solve. Ask for all the known issues, common bugs, unintuitive patterns, etc. Get very detailed if you want.
Then feed that to Claude / Codex / Cursor. Basically, create a cheat sheet for your AI agents.
Memory usage increases quadratically with sequence length. Therefore, using shorter sequences during fine-tuning can prevent memory explosions. On my 64GB RAM machine, I'm limited to input sequences of about 2,000 tokens, considering my average output for the fine-tuning task is around 1,000 tokens (~3k tokens total).
Previously, I co-founded SpeakerText (speakertext.com), a interactive/social layer on top of online video, and Humanoid (gethumanoid.com), a crowdsourced virtual labor service that uses machine learning to track worker reputation on Mechanical Turk and ensure quality.
In previous lives, I fought forest fires in Montana & Alaska for the US Forest Service, worked as a freelance news reporter for the NY Times, and drove a 911 ambulance in Harlem and the South Bronx.