Update README.md

README.md

@@ -14,6 +14,10 @@ pipeline_tag: text-to-3d
 ---
 
 # 🚀 First INT4 Quantized Efficient-Cube3D - Run on Half the VRAM
+
+[![GitHub](https://img.shields.io/badge/GitHub-efficient--cube3d-black?logo=github)](https://github.com/TrNi/efficient-cube3d)
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://drive.google.com/file/d/1-mWdiDHJIozQnKC-bP9TmaWR0fQLRo2X/view?usp=sharing)
+
 ![](https://img.shields.io/badge/Model_Size-1.26_GB_(82%25↓)-blueviolet)
 ![](https://img.shields.io/badge/Peak_VRAM-11.3_GB_(55%25↓)-brightgreen)
 ![](https://img.shields.io/badge/Setup_Time-6.9s_(97%25↓)-blue)
@@ -35,11 +39,11 @@ enabling 3D shape generation on much smaller, more accessible GPUs.
 </mark>
 
 ### Original BF16 vs Quantized INT4 Comparisons:
-##### A. Easy Categories (7)
+##### A. Easy Categories (3)
 ![Easy categories](assets/easy.gif)
-##### B. Medium Categories (6)
+##### B. Medium Categories (7)
 ![Medium categories](assets/medium.gif)
-##### C. Complex Categories (2)
+##### C. Complex Categories (5)
 ![Complex categories](assets/complex.gif)
 
 
@@ -58,36 +62,36 @@ Quantization method: **RTN W4A16**, group_size=128, via [torchao](https://github
 | `quant_config.json` | tiny | Quantization metadata |
 
 
-## New Benchmarking Dataset (15-categories, 170 prompts)
+## New Benchmarking Dataset (15 categories, 310 prompts)
 
-### Shape Quality (Chamfer Distance, 15 categories, 170 prompts):
+### Shape Quality (Chamfer Distance, 15 categories, 310 prompts):
 
-Median Chamfer Distance: 67.9 × 10⁻³
+Median Chamfer Distance: 67.7 × 10⁻³  
 
-Best categories: `vehicle_land` (41.4), `geometric_primitive` (46.5), `animal_wild` (53.8).  
-Complex categories: `symmetry_topology` (205.8), `abstract_mathematical` (167.9) - high variance.
+Best categories: `animal_domestic` (55.0), `vehicle_land` (52.2), `architecture` (54.0).  
+Complex categories: `symmetry_topology` (113.6), `abstract_mathematical` (107.2) — high variance.
 
 
-| Category | Mean | Std | n |
-|---|---:|---:|---:|
+| Category | Median | Mean | Std | n |
+|---|---:|---:|---:|---:|
 **Easy** (CD × 10⁻³ < 75)
-| vehicle_land | 41.4 | 21.1 | 10 |
-| geometric_primitive | 46.5 | 25.8 | 10 |
-| animal_wild | 53.8 | 21.2 | 10 |
-| animal_domestic | 56.5 | 21.2 | 10 |
-| tool_hardware | 66.7 | 44.7 | 10 |
-| furniture | 70.4 | 34.2 | 10 |
-| musical_instrument | 72.5 | 45.7 | 10 |
+| animal_domestic | 55.0 | 60.4 | 25.5 | 20 |
+| vehicle_land | 52.2 | 61.1 | 39.0 | 20 |
+| architecture | 54.0 | 61.7 | 29.9 | 20 |
 **Medium** (CD × 10⁻³ 75–100)
-| vehicle_air_water | 75.3 | 36.1 | 10 |
-| fine_detail | 79.2 | 54.8 | 10 |
-| visualization_stylized | 85.0 | 46.8 | 30 |
-| electronics | 92.2 | 50.1 | 10 |
-| architecture | 92.8 | 50.0 | 10 |
-| nature_plant | 98.2 | 44.0 | 10 |
+| musical_instrument | 43.1 | 79.0 | 86.3 | 20 |
+| animal_wild | 65.2 | 80.4 | 45.2 | 20 |
+| geometric_primitive | 40.8 | 81.0 | 90.2 | 20 |
+| furniture | 74.9 | 82.5 | 39.8 | 20 |
+| fine_detail | 57.7 | 83.3 | 72.8 | 20 |
+| original_visuals | 71.6 | 79.5 | 47.3 | 30 |
+| vehicle_air_water | 77.8 | 97.5 | 78.9 | 20 |
 **Complex** (CD × 10⁻³ > 100)
-| abstract_mathematical | 167.9 | 165.1 | 10 |
-| symmetry_topology | 205.8 | 242.7 | 10 |
+| electronics | 97.4 | 126.4 | 79.3 | 20 |
+| nature_plant | 111.3 | 132.0 | 69.9 | 20 |
+| tool_hardware | 63.7 | 139.5 | 193.7 | 20 |
+| abstract_mathematical | 107.2 | 147.4 | 124.2 | 20 |
+| symmetry_topology | 113.6 | 176.5 | 169.5 | 20 |
 
 ## Requirements
 
@@ -101,7 +105,7 @@ torchao==0.10.0
 The .pt file is a torchao pickle, torchao enables kernel-supported INT4 inference.
 
 ## Usage
-Please see the Google Colab tutorial. 
+Please see the [![Colab Tutorial](https://img.shields.io/badge/Open_Tutorial-4285F4?style=flat&logo=googlecolab&logoColor=white)](https://drive.google.com/file/d/1-mWdiDHJIozQnKC-bP9TmaWR0fQLRo2X/view?usp=sharing) notebook.
 
 
 ## Quantization details