Паттерн отделяет агентную логику от конкретного LLM-провайдера через слой адаптеров, фабрик и prompt-style профилей. Один и тот же агентный runtime может переключаться между backend-моделями без переписывания бизнес-логики инструментов.
Паттерн решает задачи переносимости, отказоустойчивости и A/B-сравнения моделей в production/research контурах. Он особенно важен при смешении function-calling диалектов и разных API-контрактов.
| ! The following 2 lines are what I use to display news.ycombinator.com | |
| ! Limit comments to 70 characters wide, making them easier to read | |
| news.ycombinator.com##.comment:style(max-width: 70ch !important; overflow: hidden !important;) | |
| ! Invert the colours of the site, making a dark mode that I like | |
| ! `invert(95%)` does most of the work, | |
| ! but you can tweak the hue-rotate value to fine tune the colours | |
| news.ycombinator.com##html:style(filter:invert(95%) hue-rotate(200deg); background: white) |
| {-# LANGUAGE TypeSynonymInstances #-} | |
| data Dual d = D Float d deriving Show | |
| type Float' = Float | |
| diff :: (Dual Float' -> Dual Float') -> Float -> Float' | |
| diff f x = y' | |
| where D y y' = f (D x 1) | |
| class VectorSpace v where | |
| zero :: v |
O'Reilly bought and shut down katacoda.com 😕
Alternatives:
| Site | Notes |
|---|---|
| killercoda.com | Seems to be the closest replacement for Katacoda |
| Play with Docker | Just for Docker, requires a free Docker account |
| Instruqt | This looks like more of a solution for vendors than a generally available platform This is what Red Hat has chosen to replace Katacoda for their interactive tutorials: https://developers.redhat.com/learn#assembly-id-70181 Source code here: https://github.com/openshift-instruqt/instruqt |
| # MONITORING: services run on loopback interface | |
| # nginx reverse proxy exposes services to network | |
| # - grafana:3010 | |
| # - prometheus:3020 | |
| # - loki:3030 | |
| # - promtail:3031 | |
| # prometheus: port 3020 (8020) | |
| # | |
| services.prometheus = { |
| # Put it to .github/workflows/stale.yml | |
| name: Close stale issues and pull requests | |
| on: | |
| workflow_dispatch: | |
| schedule: | |
| - cron: '30 1 * * *' | |
| jobs: | |
| stale: |
| apiVersion: apps/v1 | |
| kind: Deployment | |
| metadata: | |
| name: kafka-ui-deployment | |
| labels: | |
| app: kafka-ui | |
| # namespace: mstore | |
| spec: | |
| replicas: 1 | |
| selector: |
This post was adapted from an earlier Twitter thread.
It's incredible how many collective developer hours have been wasted on pushing through the turd that is ES Modules (often mistakenly called "ES6 Modules"). Causing a big ecosystem divide and massive tooling support issues, for... well, no reason, really. There are no actual advantages to it. At all.
It looks shiny and new and some libraries use it in their documentation without any explanation, so people assume that it's the new thing that must be used. And then I end up having to explain to them why, unlike CommonJS, it doesn't actually work everywhere yet, and may never do so. For example, you can't import ESM modules from a CommonJS file! (Update: I've released a module that works around this issue.)
And then there's Rollup, which apparently requires ESM to be u
