GoodSeed v0.3.0 pitches faster ML tracking

This is a focused “do one thing well” launch that could win practitioners tired of heavyweight experiment platforms.

• –The core value prop is speed and usability in daily experiment comparison, not full lifecycle orchestration.
• –Neptune proxy and migration guidance lower switching friction, which is crucial for early adoption.
• –Local-first storage plus optional cloud sync gives teams flexibility on privacy, cost, and workflow maturity.
• –Feature depth (metrics smoothing, config diffing, stdout/stderr, GPU/CPU telemetry) covers most real-world training loops already.