Post

Sam Rose
@samwho.dev

Are you using open models on a regular basis? If yes, I’d love to hear about it! - what’s your use-case? - local or through an inference provider? - how has the experience evolved over time? - anything you miss from the closed models, if you’ve used them? Would appreciate reposts for visibility 🙏

Erica Windisch
@ewindisch.ontological.observer

I have a fairly sizable local lab for running open weights. I use it to replace casual chat LLMs, I only use cloud tokens for coding, but would like to transition usage to entirely local inference. Focus of my work is to have a fully local autonomous stack with in-weight learning.

Sam Rose
@samwho.dev

Are you creating your own model for in-weight learning or are you using some existing base model and building on top of that? How often do you update the weights? Do you have problems with catastrophic forgetting?

0 reposts0 likes

Erica Windisch
@ewindisch.ontological.observer

It's a lot of *infrastructure*. Less exciting work has happened (so far) at the model side than making sure weights are check-pointed and distributed via Git / Huggingface, server side sandboxed tool use, OIDC / Oauth scopes and policy, etc. Model-wise, working with Qwen3 dense models.

Erica Windisch
@ewindisch.ontological.observer

I started some work on Qwen3 Omni model architectures and Janus, but priority is a complete agentic loop. The next release will have MCP support with enough tools to instruct the model to re-train itself. After the loop is closed, I'll worry about the *quality* of the results.