Update README.md

README.md

@@ -1,7 +1,7 @@
 ---
 title: ZeroEngine V0.1
 emoji: 🚀
-colorFrom: blue
+colorFrom: gray
 colorTo: gray
 sdk: gradio
 sdk_version: 6.5.0
@@ -10,16 +10,10 @@ pinned: false
 license: apache-2.0
 ---
 
-# ZeroEngine System Kernel
-A specialized inference engine optimized for low-resource Hugging Face Spaces (2 vCPUs / 16GB RAM).
+# ZeroEngine V0.1 (Kernel)
+High-performance inference engine for 2-vCPU / 16GB RAM constraints.
 
-## Key Features
-- **Deterministic Partitioning**: Strictly splits 2 vCPUs between two concurrent users.
-- **Resource Gatekeeper**: Prevents OOM crashes with a strict 50% RAM model limit and 200MB system buffer.
-- **Ghosting Queue**: Enables pre-typing and background prompt preparation for queued users.
-- **Persistence Layer**: Tracks model popularity by pushing telemetry JSONs to the HF Hub via `HF_TOKEN`.
-
-## Hardware Specifications
-- **CPU**: 2 vCPUs (shared)
-- **RAM**: 16 GB (Shared)
-- **Optimization**: `llama-cpp` with mmap and single-core pinning per slot.
+## Optimizations
+- **KV-Cache Stitching**: Asynchronous pre-evaluation of queue inputs.
+- **Hard Partitioning**: Dedicated core assignment per concurrent user.
+- **Memory Mapping**: weights mapped via `mmap` to preserve RAM for context.