Update README.md

README.md

@@ -74,24 +74,27 @@ Nandi-Mini-500M introduces several efficiency-focused architectural optimization
 
 #### Shared KV (Shared Key-Value Vectors)
 
-One of the core ideas explored in Nandi-Mini is **Shared KV**, an efficient attention mechanism where Key and Value representations partially share learned vector space representations across attention computation.
+Shared KV is one of the core architectural ideas explored in Nandi-Mini. Instead of storing separate Key and Value vectors, both share the same underlying representation, while a lightweight Key normalization step is applied specifically for attention computation.
 
-This approach is designed to:
+This design reduces KV-cache memory usage by ~50% during inference with only a small increase in compute overhead, since Key normalization and RoPE transformations must be applied dynamically during attention computation.
 
-- Reduce memory overhead during inference
-- Improve parameter efficiency
-- Lower KV-cache footprint for long-context generation
-- Enable faster deployment on resource-constrained hardware
-- Maintain strong quality despite smaller compute budgets
+Nandi supports two KV cache modes:
 
-Shared KV is part of our broader effort toward building deployable foundation models optimized for:
+```json
+"kv_cache_mode": "shared"
+```
 
-- Edge devices
-- On-premise AI systems
-- Low-latency enterprise inference
-- Efficient multilingual serving
+Uses Shared KV, reducing KV-cache memory by ~50% with slightly higher compute overhead.
 
-This is still an active research and optimization area within the Nandi model family, and we plan to share deeper technical details in upcoming engineering blogs.
+```json
+"kv_cache_mode": "vanilla"
+```
+
+Uses standard separate Key-Value caching for maximum inference compatibility and lower compute overhead.
+
+Shared KV is part of our broader focus on deployable foundation models optimized for edge devices, on-premise AI systems, low-latency enterprise inference, and efficient multilingual serving.
+
+This remains an active research area within the Nandi model family, and we plan to share deeper technical details in upcoming engineering blogs.
 
 ---