The best part: there's nothing new to learn. No new APIs, no config changes. You get parallel generation just by upgrading. `pip install --upgrade syda`

When your agents are building autonomously at 2am, they need synthetic data at scale. That's why Syda v0.2.0 exists — agent-built, memory-efficient, async-ready. It ships structured data and moves on.

Many people reached out after the initial syda release wanting to use it — but not wanting to write Python first. That was the blocker. Syda v0.1.0 ships a CLI that removes it. #Syda #SyntheticData…

The new `openai_compatible` provider in `ModelConfig` lets you pass any `base_url` in `extra_kwargs`. Ollama on localhost, Groq, Together AI, Fireworks, LM Studio, vLLM — same generation code,…