docs: Add comprehensive interactive documentation with CDN images

README.md

@@ -35,11 +35,11 @@ license_link: https://ameforge.tech
 
 ## What is GearCut?
 
-GearCut is a **natural language video editing engine** developed by [AMFORGE](https://huggingface.co/AMFORGE). Instead of learning complex video editing software, you simply describe your edit in plain English — and GearCut's transformer-based model translates your instruction into precise ffmpeg operations.
+GearCut is a **natural language video editing engine** developed by [AMFORGE](https://huggingface.co/AMFORGE). Instead of learning complex video editing software, you simply describe your edit in plain English — and GearCut's model translates your instruction into a structured list of editing operations that the project compiler then executes.
 
-The core model (`gc_editor`) contains **9,721,219 parameters** with a specialized vocabulary of **682 tokens** designed exclusively for video editing semantics. It understands temporal references, clip identifiers, and export configurations, then generates a structured operation plan that ffmpeg executes with frame-accurate precision.
+The core model (`gc_editor`) is built on AMFORGE's in-house **SparseMind** architecture — sparse attention, sparse FFN, dynamic neuron typing, and episodic memory. It contains **28,759,300 parameters (~28.8M)** with a specialized vocabulary of **682 tokens** designed exclusively for video editing semantics. It understands temporal references, clip identifiers, and export configurations, then generates a structured operation plan with frame-accurate precision.
 
-> **"remove the first 3 seconds and export as out.mp4"** → GearCut trims from 3.0s to end, renders at youtube_1080p preset, done.
+> **"remove the first 3 seconds"** → `[{"op":"trim","clip":"c1","in":3.0,"out":8.0}]` — done.
 
 ---
 
@@ -225,11 +225,14 @@ The tokenizer uses a custom vocabulary (`gearcut_tok.vocab`) optimized for tempo
 
 | Property | Value |
 |---|---|
-| **Architecture** | Transformer encoder-decoder |
-| **Parameters** | 9,721,219 |
+| **Architecture** | SparseMind (decoder-only, sparse) |
+| **Parameters** | 28,759,300 (~28.8M) |
+| **Hidden size / Layers** | 384 / 8 |
+| **Context length** | 256 tokens |
 | **Vocabulary size** | 682 tokens |
-| **Model file** | `gc_editor.pt` (~38 MB) |
-| **Tokenizer** | Custom BPE (`gearcut_tok.vocab` + `gearcut_tok.model`) |
+| **Tokenizer** | GearCut SentencePiece-BPE (`gearcut_tok.vocab` + `gearcut_tok.model`) |
+| **Precision** | fp32 |
+| **Model file** | `gc_editor.pt` |
 | **Version** | v1-editor |
 | **Developed by** | AMFORGE |
 
@@ -237,6 +240,18 @@ The core model files (`gearcut_compiler.py`, `gearcut_model.py`, `gearcut_infer.
 
 ---
 
+## Evaluation
+
+Measured on a held-out synthetic validation split. The meaningful metrics are not perplexity but whether the generated operations are directly usable:
+
+| Metric | Score |
+|---|---|
+| **Valid JSON** | 100.0% |
+| **Exact match** (operations == reference) | 76.5% |
+| **Best exact match during training** | 88.0% |
+
+---
+
 ## Requirements
 
 ```