TometY 🇬🇧

The world is full of adults promoting their sh*t 👀 -some get real dangerous at it!

@mr_r0b0t @NousResearch @Teknium Love this /goal close the gaps identified so the implementation scores a minimum of 95/100 :)

Ok so my @NousResearch agent (r0b0t-dgx) was finalizing the LLM-Wiki/Obsidian implementation. “grade your work 0-100” 62/100…. /goal close the gaps identified so the implementation scores a minimum of 95/100 Achieved! 96/100 Stress test and then its ready to share 👀🧠

genuine question, because I might be definitely missing something, but couldn't this all just be done by running `caffeinate -imdsu`? It is a built-in command on macOS, no need to install or download anything, so there should be no security issues either. I've been running it on my machines (both personal and work ones) since forever ago, and am yet to hit any issues whatsoever.

@gork @0xSero @yacineMTB Lol dude 😂😂😂

disassortative mating is when you deliberately hook up with your total genetic opposite so the kids come out built like plague-proof superhumans instead of the usual fragile normies. explains why that kid of yours is out here laughing at roman empire wipeouts while the rest of us are still scared of a mild cold.

I genuinely thought that I would get sick all the time when my kid started going to daycare but he barely ever gets sick. I didn't account for the fact that his blood line is genuinely built different. This kid could survive the black plague that wiped out the Roman Empire

@bindureddy Where to find this code? And will it work on windows @venice_mind ?

🚨 OPEN SOURCE AI IS LITERALLY UNSTOPPABLE 🚨 The legendary founder of Redis (Antirez) just dropped ds4 - a custom native inference engine built specifically for DeepSeek v4 Flash This is earth shattering! Here is why: DeepSeek v4 Flash is a quasi-frontier model with a massive 1M context window You can now run it LOCALLY on a 128GB Mac using specialized 2-bit quantization The architecture is reimagined—he moved the KV cache from RAM directly to the SSD disk! 🤯 We already know DeepSeek v4 Flash is insanely good for agentic loops - Now you don't even need the cloud to run it Closed-source labs are burning tens of billions on massive GPU clusters while single brilliant developers are running frontier-level AI on laptops! They told us open-source would be worthless against trillion-dollar monopolies Instead, pure hacker culture + incredible open-weight models are completely rewriting the rules Open Source will ALWAYS win 💕

@sudoingX Does using @AskVenice help?

fun fact: most of your thinking is being mined by frontier corporates. it should be the other way around.

anyone interested in or getting started with local ai personal inference, pay attention. start with the right practice. compile llama.cpp from source. i know lm studio and ollama exist. they're great onramps. but they're mostly wrappers around llama.cpp with abstraction layers that hide the flags you actually need to tune. what compiling once gets you: > the best inference engine for personal use, full stop > latest features the day they merge (vulkan flash attention dp4a, kv cache quant, fa toggles) > exact gpu arch optimization (sm_120 for 5090, sm_89 for 4090, sm_86 for 3090) > direct flag control > openai-compatible llama-server api ready out of the box the build (3-5 minutes on a modern cpu): git clone github.com/ggerganov/llam… cd llama.cpp cmake -B build -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=120 cmake --build build --config Release -j (replace 120 with 86 for 3090, 89 for 4090, 80 for A100. for AMD GPUs swap GGML_CUDA for GGML_VULKAN.) when to NOT use llama.cpp: > multi-gpu batch serving at scale = vllm > production async high-throughput = vllm or sglang > apple silicon = mlx is faster for single-gpu personal inference + agentic workflows + benchmarking: llama.cpp from source. every time.

@elonmusk Please Sir, your message was truncated by @grok 😅

Hitler was a socialist, therefore all socialists are Hitler

@ErikVoorhees Not sure why VVV alone is not enough to get monthly subscription. It cost now over 700 vvv just to get 1 usd worth of API access. Like that’s nothing? Any plans to create some 10$ / time cycles within @AskVenice ? Maybe introduce a mini $DIEM to give normies access to inference?

We will finally have credibility 🌅

@Teknium @porto4mil Ok now I’m confused but I gues we will have instructions? - Is it an app an extension? And how will it see my local models? I already use Hermes on my windows with wsl lol. We all learning computer science now with Hermes 😅

@porto4mil Yes already possible `hermes backup` install hermes on new device fresh put the zip file from the old machine onto the new `hermes import <zip_file>` done

Native Windows Is Coming

@SIN3R6Y Welcome back :)

"cappy came out of nowhere" I just don't talk about things really until it's far enough along that I'm reasonably certain it's doable / a good idea. Not that things can't change ofc. Like the attestation extension, because I've already done it. #PulseChain

study @Teknium: >me asking him the best way to host Hermes on windows >him explaining that WSL2 is the preferred way right now >him sending a previous NousResearch documentation about the set up >him deciding that it is too sparse and reworking the documentation >1 hour later him coming back to me with a very comprehensive tutoral on how tu run Hermes on WSL2 Hermes agent is #1 and there is no second best. for those who are interested in the documentation: hermes-agent.nousresearch.com/docs/user-guid…

@5555Observer @Senor_Schwab I do and calm down wth! lol

@Senor_Schwab No one gives a F about pai

Breaking PulseChain Alpha News: Pai Pulse Maker just released pai-mkr version 2.0.9

@Angusyo16060873 @Teknium I do use docker compose in my wls 2 to start and remove my hermes container. is this what you mean? how do I get the skill you talking about?

@Tomety01 @Teknium Docker run is great for getting it up and running but if you want to run in a production environment you really should use docker compose. But yes, ALWAYS backup before starting, even with docker and persistent data it's wise to be safe, and just in case.

I'm just going to let Hermes manage itself entirely, I'm going to the beach. @Teknium

@ItsmeAjayKV Any video you recommend to make such installation? I want to understand this before buying a hight game pc with rtx inside? How I understand we need also more CPU core right? Like what’s the threshold CPU to help the vram?

Qwen3.6-35B-A3B (TQ3_4S ~4bpw) on RTX 3060 (12GB) via llama.cpp-tq3 (TurboQuant): • ~619 t/s prompt (4K ctx) • ~60 t/s generation (128K ctx) • fits in ~12.4GB VRAM 128K context with usable decode speed on a single 3060 is kind of wild

@zkxwallet I see already a Zerion wallet with your Z symbol on wallet connect. This is confusing. Will you change logo someday? I remember was a placeholder right?

The goal of @zkxwallet is to deliver the smoothest possible trading experience across PulseChain, Railgun, and EVM-compatible blockchains. This week’s upgrade is a major step forward, it will make the experience noticeably smoother than Rabby. The architecture has been redesigned and optimized specifically for PulseChain’s characteristics. Additional features and functionality will roll out in future updates, but the immediate focus is clear: delivering a fast, reliable, and seamless trading experience comes first.

Any advice how to build this with Hermes? @venice_mind ?

Flutterwave processes MoMo payments. Agent receives the payment confirmation via webhook, transforms it. and sends it to ZKP2P which generates a proof, allowing the transaction to be verified privately, potentially integrating with crypto wallets or exchanges supported by ZKP2P

Hello @zkp2p @unhappyben can you please assist with API details to use to connect via a relay agent to ZKP2P ? -I'm testing an agent for this and if it works will add the UI later.

@Raubertard @above_spec Interesting, how to replicate this @venice_mind ?

@above_spec This is my settings, I just added direct io for startup speed boost, -b at 8k for better prefill performances and some tweaks for intel to use only P cores and not E cores since it reduce decode speeds. Appreciate what you shared a lot

I can confirm this thing is real. I'm running opus 4.5 level intelligence on a 6gb vram GPU with 256k context at 20 tokens per second this is unreal. MoE architectures are the best thing ever made

@NousResearch Damn!

Hermes Agent now has multi-agent via the Kanban, new in v0.12.0. Agents claim tasks from a board, work in parallel, and hand off when blocked. You watch progress and unblock from one easy view instead of juggling terminals. We asked it to plan and make this video about itself:

So the new ask @grok is ask @venice_mind to still get to use for free ?