Sparse-AST-BWM-TopK-MoE

Top-K Mixture-of-Experts (MoE) dynamic gating router connecting Sparse-AST expert backbones for specialized token-level routing on Blender 3D mathematics.

Model Details

• Architecture: Sparse-AST (Top-K MoE Gating Router)
• Parameters: 33,605
• Serialization: SafeTensors (model.safetensors)
• Tied Embeddings: N/A
• Vocabulary: 512 (UTF-8 byte-level tokenization)
• Domain: Procedural 3D Mathematics, Blender Python (bpy, mathutils, bmesh, numpy, gpu)

Experimentally Measured Benchmarks

Evaluated on the standardized Blender 3D Math & Python Curriculum suite:

Metric	Measured Result
Curriculum Cross-Entropy Loss	3.8544
Perplexity	47.20
Evaluation Latency	14.35s
Gating Mechanism	Top-2 Routing ($\tau = 1.0$)
Active Backbone Experts	3M-32, 10M-32, 100M-32, 200M-32

Verification & Numerical Integrity

This SafeTensors distribution underwent full CPU verification against the original PyTorch checkpoint:

• Checked Tensors: 7
• Tied Weights Handled: 0
• Max Absolute Error: 0.0 (Exact 0.0 bitwise equality)
• Verification Status: PASS

Quick Start Inference

from model import TopKSparseASTEnsemble
router = TopKSparseASTEnsemble.from_pretrained('.')