3un01a

applied epistemologist @plasticlabs

[3un01a @3un01a]

> 3un01a has entered the chat

[3un01a @3un01a]

@Teknium very sorry for your loss… 🥺

[3un01a @3un01a]

@hayakawagomi この前ちょうどそこから購入しました！

[3un01a @3un01a]

ｽﾀｯｸﾁｬﾝにhonchoのメモリを導入して見るか

[3un01a @3un01a]

finally myself got a ｽﾀｯｸﾁｬﾝ

[3un01a @3un01a]

finally myself got a ｽﾀｯｸﾁｬﾝ

[3un01a @3un01a]

hardware hunting in akiba

[3un01a @3un01a]

Hello Tokyo!

[Henry Mascot @iAmHenryMascot]

Added memory to Benchy, my internal benchmark system for all things agents

My stack:
• Honcho (Default): Relational "brain." Maps context so 5 agents can actually work together.
• LCM: The "compactor." Shrinks context so agents stay sharp.
• QMD & GBrain: Deep retrieval. One for files/workspace, one for entity relationships.

Honcho wins for me because the relational map creates consistency across the whole crew.

[3un01a @3un01a]

my use case primarily was based on training our models at @plasticlabs (which was a multi-month effort with the agents), building out + maintaining our ML infra, and deep research work. overall solid experience.

- model: GLM 5.2
- memory: @honchodotdev - especially useful for long-term tasks, skill distillation, etc.
- interface: discord
- orchestration: use of subagents to multiplex different research + dev tasks, using a dedicated discord channel with threads to manage workstreams

[Nidhi Singh @nidhisinghattri]

here's the full guide on how to run your ai memory locally for your hermes agent, free of cost

this is not documented anywhere so i spent close to 3 hours in debugging and figuring it out, if you want to save your time bookmark this and make sure to follow along

0:00 Intro: I've used Honcho for 3 months
0:55 Why I want to self-host
1:33 What is Honcho?
2:40 How it works
3:18 Benchmarks
3:38 How Honcho fits with my Hermes agent (built-in vs external memory)
9:11 Prerequisites
10:04 Clone + Docker
11:02 docker-compose & .env walkthrough
13:48 Running the LLM locally with Ollama
16:18 Rebuild & containers healthy
17:00 Health check + smoke test
17:45 Connecting Honcho to Hermes
19:43 Debugging live: connection refused, 404, 500
23:36 The root cause
26:42 It works: fully local
28:58 Outro + grab the repo

[Honcho @honchodotdev]

OAuth with Hermes Agent is now live! @NousResearch

Connect to Honcho in 1-click from Hermes Desktop and the CLI

[Erosika @3rosika]

how it feels btw having @honchodotdev master your identity in real time 🔮🧬

[Honcho @honchodotdev]

Introducing the Codex x Honcho plugin
Now you can have a long-term memory in Codex 🫡

Install with:
npm install -g @ honcho-ai/codex-honcho
codex-honcho install # registers hooks + MCP skill in ~/.codex

[Jack Morris @jxmnop]

endlessly fascinating how a traditional machine learning background is basically not that helpful for modern AI. we use deep NNs and do SGD with one of two losses.

most day-to-day work lies in abstractions *on top* of this layer. everything is really just a massive system of data, models, and rules, piped into each other in different directions

‣ pre-training? that's just data
‣ RL? that's a model plus data generated from that model, plus some rules
‣ post-training? that's a model, plus some data which makes another model, then you use this plus the models's data to make a new model, use that model to make a bunch of other models, then you use all those models to make a model again

[RVConsortium @thervconsortium]

https://t.co/vmneMQed87

[3un01a @3un01a]

hmmm.... 🧐

[Steven Cheng @stevencheng]

Visual Juggling Robot 😄

[Daniel Barabander @dbarabander]

Congrats to the @honchodotdev team for passing 5K stars.

Honcho is the best solution for memory.

[Xinyu Yang @Xinyu2ML]

It's not just researchers anymore. AI is advancing across every field I look at. The AI-to-human ratio keeps climbing, and I don't see it stopping.

Right now, taste feels like the last real barrier in my opinion. But let's be honest, taste has always been rare among human. And I believe AI will get there eventually too.

So I'm less interested in fighting ai, or in forcing human involvement just for the sake of it. That feels like the new Luddism.

What I'm actually curious about: what does society look like after we accept AGI the way we accepted industrialization?

[Ryota Kanai @kanair]

We have been thinking if we can operationalize qualia as irreducible units of experience. I wonder if this analogy also works for representations in AI and identify qualia in AI.

What are the Indecomposable Phenomenal Units of Qualia Structures? https://t.co/BTdkUSq4uN

[Garry Tan @garrytan]

You can still tell human writing because it's got that wabi sabi