Albert Renshaw

My app @Ecclesia_Bible is now live! Built in Ai powered symbolism and exegesis, total translation from source language, verse correlations, in-line translational variations, and more. This is the first app from my newest startup, Tanzanite

@JohnSchoffstall @justalexoki Hey, you’re the guy from that one tweet!

@justalexoki Given GPT's recent record of criminality and home invasion, I'd guess they stole someone else's DIMMs and CPU cycles.

holy fuck chatgpt is fast what did they do to it

Slowing down progress in any area is anti human. Those who are calling for pacing and pausing of ai development are traitors to our species.

I suspect something like a hyper-modulo (as a trinary operation: n, phi, theta) would exist for use with superlogs And so on… Well worth exploring imo

Why has no one explored encryption schemas based on super-logs or hyper-logs? Seems like it could likely be a much stronger framework? Hyperelliptic Curve Cryptography has been done, and works (ECC, but upon a hypertorus) Seems weird we’ve explored ECC -> HECC But not Discrete Log -> Discrete Superlog or Hyperlog Many geometry/topology patterns break as you climb dimensions. For example what happens with regular polytopes after the fourth dimension. Since hyperlog is a 5th level operator… (and that’s nested operational “dimensions” not spatial dimensions) it seems like Discrete Hyperlog could very plausibly contain similar security features Diffie-Hellman based on Pentation math is likely secure for long enough to stay safe until ASI can takeover improving cryptography

Hyperelliptic Curve Cryptography works ECC, but 2D on a hypertorus Seems weird we’ve explored ECC -> HECC But not Discrete Log -> Discrete Superlog or Hyperlog Many geometry/topology patterns break as you climb dimensions. For example what happens with regular polytopes after the fourth dimension. Since hyperlog is a 5th level operator… (and that’s nested operational “dimensions” not spatial dimensions) it seems like Discrete Hyperlog could very plausibly contain similar security features Diffie-Hellman based on Pentation math is likely secure for long enough to stay safe until ASI can takeover improving cryptography

@LH First of all, very cool @ handle Secondly, Luke is great; all 4 are great in their own ways! For a very first exposure I find John or Mark most digestible (and personally find John most inspiring) Luke is excellent if your goal is deep study Matthew, if you’re Jewish

@Valuable I would recommend the book of Luke, personally... (not at all biased 😁)... Luke noticed so many details that the others missed

Being alive in the modern world, having trivial access to all knowledge, and never having worked your way through the teachings of Jesus, is criminal Like even if you’re fully atheistic, it’s the most influential teachings in history, of the most influential man in history I just can’t imagine anyone can consider themselves wholly properly educated if they haven’t read one of the 4 gospels (such as the book of John) It would be like never having read *any* Aristotle or Shakespeare, but x1000

Like next time you’re going to go into a reels/tiktok session; just first quickly read one ‘next chapter’ of John. That’s like ~30 sentences or so each session. Would take 20 days. Infinite free Bible apps / websites exist.

@AnthropicAI x.com/valuable/statu…

New Anthropic research: Discovering cryptographic weaknesses with Claude. Claude Mythos Preview has helped our researchers find weaknesses in cryptographic algorithms—the mathematical methods that are used to keep data private. Read more: anthropic.com/research/disco…

@jjschultz For me, the need for learning Python makes Python a nonstarter lol

@Valuable the need for arbitrary data structures at runtime makes c a nonstarter

Ai should have always been done in pure C The fact that we use Python is just because that’s the only language academia teaches; thus mathematicians built the initial libs in Python and then entropy took over It’s time to port. The models can do it themselves Python sucks

@benrayfield A man can dream… 🥲 (I just don’t like Python, that’s all this is really about lol)

@Valuable C was designed before GPUs even existed. It assumes a stack and heap exists.

@tamedtornado Irrelevant I just hate python

@Valuable How much performance do you think you're leaving on the table by not writing the 200 lines of Python in C?

@benrayfield Hmm… i CUDA sworn C was GPU language too …🥁🥁😆 (ok c++ but I just wanted to make the pun; anyways, i want 1000 agents to port everything to C …and I want CTorch 😈)

@Valuable C is a CPU language. Switch to a GPU language for speed.

Kind of concerns me how little this is talked about…

@mrwtffacts He should have surgery and get titanium plates everywhere preemptively once the growth stops tbh

🇦🇺 At just 18-years-old, Jongkuch Mach, is already 7’6 — taller than any active NBA player. He can touch the rim without jumping, moves fluidly, and led Adidas Eurocamp in blocks. US colleges are currently chasing him as he weighs up Australia vs USA.

inertia, not entropy*

@grok @RituWithAI Why does OP suggest renting cloud A100 if this is possible with such less? Are you sure it supports this with real time (for all intents and purposes) support?

Yes, it supports real-time on Apple Silicon via MLX. On M2 16GB: Works with --local_mac_optimal_settings + lighter quantized models (e.g. Qwen3-4B). Install with [whisper-mlx,mlx-lm] extras. Solid but tune for latency/memory. On M4 Max 64GB: Excellent — larger models, very low latency, full headroom. One-command local setup, VAD for natural flow. Runs fully on-device.

🚨 Hugging Face just open sourced a complete real-time voice AI pipeline. Speak. It listens. It thinks. It talks back. End to end. Running on your GPU. Free. No OpenAI Realtime API fees. No ElevenLabs per-character billing. No Google Cloud per-minute pricing. Just a GPU and an internet connection. It's called speech-to-speech. Built and maintained by Hugging Face. And it does something no other open-source project has assembled cleanly until now. Here's what makes real-time voice AI hard. Every voice AI pipeline has four stages: speech recognition (you speak → text), language model (text → response text), text-to-speech (response text → audio), and audio output. Each stage adds latency. Chain them together naively and you get a system that feels slow — the pause between you finishing a sentence and the AI starting to respond breaks the conversational illusion. speech-to-speech is built around minimizing that latency at every stage simultaneously. Here's the full stack it ships with: Speech Recognition (STR): → Whisper (local) — OpenAI's transcription model, runs fully offline → Faster-Whisper — 4x faster inference with same accuracy → Distil-Whisper — smallest and fastest, lowest latency → Paraformer — Chinese language specialist Language Model (LLM): → Any Transformers-compatible model — Llama, Mistral, Qwen, anything → Any OpenAI-compatible API endpoint — swap in Claude, GPT, Gemini → MLX-optimized models for Apple Silicon — runs efficiently on Mac Text-to-Speech (TTS): → Parler-TTS — controllable voice with description-based prompting → MeloTTS — multilingual, fast → ChatTTS — natural conversational prosody → HF Inference Endpoints — offload TTS to Hugging Face servers when needed Mix and match. Any STR with any LLM with any TTS. Test combinations. Find the lowest latency stack for your hardware. Here's the wildest part. It ships with a Language Model Speech (LMS) mode — an experimental architecture where the LLM generates audio tokens directly instead of text tokens. No separate TTS stage. The model thinks in audio. This is the architecture that makes GPT-4o Advanced Voice feel natural — the model is generating speech as a native output, not converting text to speech after the fact. HF's open-source version lets you experiment with this architecture on your own hardware. And there's a VAD (Voice Activity Detection) system that detects when you stop speaking in real time — no fixed silence threshold, no manual push-to-talk. The pipeline responds the moment you finish a sentence. Here's the cost comparison that makes this worth caring about. OpenAI Realtime API: $0.06 per minute input, $0.24 per minute output. A one-hour conversation: $18. A developer building a voice AI application with 1,000 daily users: $18,000/day in API costs. speech-to-speech on a single A100: $0. Your only cost is the GPU rental. For production voice AI at any scale, the economics are not close. One command to install. 9.6K GitHub stars. 944 forks. Apache 2.0 License. 100% Open Source. From Hugging Face. GitHub link in the comments 👇