Mastodawn

After all, think about the resources you spend and waste. Most token output you might not even read at all, it's just waste. Be mindful about resource consumption. Use #LLM to create your custom tools. They stay also when you do not have access to an LLM.

Show thread

PonyM 3d ago

- Telling the LLM to create project structure that you use often -> teach the LLM to to create good templates (#copier templates for example), you will be faster and the template just works as is.
- Not re-iterate on good practices and learnings -> add them as skills to your collection, then you can make your agents descriptions short by just referencing skills to use for certain tasks

PonyM 3d ago

How to use #AI wrong when coding:

- Keep code examples in Plan files or AGENT.md after code was written -> code is now source of truth. You may want to adapt the "Plan Agent" description
- Using AI to generate well defined and structured output. The AI should write a tool for you to generate it, not generate it itself
- Not using #LSP server integration -> you waste your time fixing what the LSP can tell the Hashtag#LLM to fix right away

Show thread

PonyM 4d ago

@adfichter Was ist die Studi-Quote an Frauen an der ETH Zürich für Informatik?

PonyM 4d ago

CI times using my #homelab #kubernetes #gitlab runner building a #django app using #uv. The build step is a docker rootless build using `buildctl-daemonless.sh build`. Building the image takes as long as the release which is quite astonishing! I also realize: The screen I edit this post is connected to a GPU passed through to my main working VM, and this build probably also ran on the same hardware but inside a k8s worker VM on #proxmox.

PonyM 4d ago

#QGIS Vor ca. 5 Jahren bei einem Geologie-Büro eingeführt als Ablösung von #arcgis. Einen Monat für alle Karte und Symbologien zu transformieren, inkl. Schulungsvideos mit #obs Resultat:

PonyM Feb 28

@kubikpixel Hey was meinsch =)? Für 9.90 pro Jahr

Show thread

PonyM Feb 28

@tsueri Scho ein Zürcher hüt dezue bracht. #didit

PonyM Feb 28

Ironically today since multiple hours the top story on #hackernews is "How to delete my #OpenAI account" with around 1050 votes. I thought, wow this is a alot. Is it compared to other stories? Well since I am doing #data ingest of hackernews since around 3 months, I can actually use one of the plots =)

Show thread

PonyM Feb 28

and last but not least this after Anthropic's fight with DOW: https://x.com/sama/status/2027578652477821175

Sam Altman (@sama) on X

Tonight, we reached an agreement with the Department of War to deploy our models in their classified network. In all of our interactions, the DoW displayed a deep respect for safety and a desire to partner to achieve the best possible outcome. AI safety and wide distribution of

X (formerly Twitter)

Tracks	https://soundcloud.com/red_ribbon
Docs
Code	https://gitlab.com/pony_m
Legends	https://www.youtube.com/channel/UCtWuB1D_E3mcyYThA9iKggQ