VibeCheck v1: Compact 11M sentiment transformer hits

VibeCheck v1 proves that data diversity is more critical than raw parameter count when building specialized edge-compatible models.

• –**Micro-Sized Efficiency:** At just 11M parameters, it can run on extremely low-power hardware, including CPU-only edge devices.
• –**Smart Data Mix:** Combining IMDb's long-form reviews with SST-2's short-form sentences makes it significantly more versatile than niche-trained predecessors.
• –**Zero-Shot Potential:** The model's ability to pick up "vibes" in non-English languages despite being trained on English suggests the architecture is learning generalized emotional patterns.
• –**Practical Prototyping:** A 30-minute training run yielding usable sentiment analysis is a great benchmark for independent researchers experimenting with from-scratch training.