Tranquil Flow

✨ Let's try save the world? ✨ Weave dreams into reality. Meow? 🐦‍⬛ 🍄 👁️

[Tranquil Flow @FlowTranquil]

@peter_szilagyi Really sad I missed your talk in Prague, absolutely love what you are doing. Keep fighting the good fight o7

[Tranquil Flow @FlowTranquil]

@satorinakamoto Legit. If all these trader mercenaries left nothing would be lost whatsoever. Would be nice to get some of the noise out of this space.

[Tranquil Flow @FlowTranquil]

@satorinakamoto Yuuuup. Beyond frustrating. If people are so concerned with number go up they can fuck off back to the stock market and play in the kiddie pool there. I like Ethereum cause it's offering a decentralised alternative to that hellscape.

[Tranquil Flow @FlowTranquil]

@ExistentialEnso yiss

[Tranquil Flow @FlowTranquil]

@0xSero Elon winning the AI race is GG for humanity. Open source local models must win.

[Tranquil Flow @FlowTranquil]

@uttam_singhk @ASvanevik Tbh most governance tokens in general never made sense.

[Tranquil Flow @FlowTranquil]

@tushant_suneja @pcaversaccio I'd rather something built on principles that doesn't scale and is still useful/good. So much stuff built in this space is just straight up centralised trash. What's the point of it outlasting if we just end up with the same parasites and middlemen all over again?

[Tranquil Flow @FlowTranquil]

@kaiostephens @huggingface Anyone using Carnice on Mac for #Hermes should be able to get a boost using this, although ofc results vary based on hardware :) @Prince_Canuma @awnihannun #MLX

Recommend setting chat_template_kwargs {enable_thinking: false} server-side to keep the model concise on agent loops.

[Tranquil Flow @FlowTranquil]

Three MLX quants of Carnice-V2-27B (by @kaiostephens) up on @huggingface

On M4 Pro vs Q5_K_M / llama.cpp baseline:
• mixed_3_6 (12 GB): ~50% faster
• 4bit (14 GB): ~42% faster
• 6bit (20 GB): ~31% faster

https://t.co/OEclVzM5Sj
https://t.co/OEclVzM5Sj
https://t.co/mLfpZc7JIy

[Tranquil Flow @FlowTranquil]

@vpavlin ummm ok at least I'm getting responses I guess?

[Tranquil Flow @FlowTranquil]

@vpavlin I was having issues with GLM 5.1 + Hermes with their Pro plan. Resolved with this fix, now working fine for me:

[Tranquil Flow @FlowTranquil]

@Teknium Seems like it would be useful if Hermes had a suite of benchmarks to evaluate its memory plugins... ;) https://t.co/YctjLr6pvx. .

[Tranquil Flow @FlowTranquil]

@vpavlin Hope experiments go well! Also curious what made you settle on the Bosgame AI PC? Was it just best bang for buck or any other reason?

[Tranquil Flow @FlowTranquil]

@Bedrovelsen The main thing with the backend:docker mode is that only the workspace is mounted, so hermes-aegis source files are not visible/reachable for the agent process.

[Tranquil Flow @FlowTranquil]

Announcing Hermes-Aegis built for the @NousResearch hackathon! Adds multiple layers of security measures to Hermes-Agent so you can use its sleek agent harness while feeling secure that your device is protected. 🛡️

[Tranquil Flow @FlowTranquil]

@Bedrovelsen It depends on the backend set by user. If using backend:local, then we rely on the existing Hermes-agent approval gating system for dangerous commands. If using backend:docker, Hermes-agent will run inside a docker with a hardened container config to prevent self modification.

[Tranquil Flow @FlowTranquil]

v0.2.1 - CLI mode (WIP)

[Tranquil Flow @FlowTranquil]

Well, we are using agents so we don't just do one entry for the @NousResearch hackathon... we do two. Introducing hermes-neurovision! It's a visualiser that hooks into your Hermes agent session, takes that input and converts it into visuals! See your agents in action :D

[Tranquil Flow @FlowTranquil]

and legacy mode gallery

[Tranquil Flow @FlowTranquil]

@NousResearch Hermes v0.2.0 does address some of the attacks portrayed in the demo :D However, Aegis still provides deeper coverage, and I've deployed a patch to integrate the new security features from Hermes such as redactpy and Tirith.

https://t.co/bUrjlsxcxr