Albert MoE-13 Benchmark Suite v2.0.0

Albert MoE-13 — TIS Benchmark Suite

One-line install and run:

Linux / macOS / Android (Termux)

curl -fsSL https://raw.githubusercontent.com/eriirfos-eng/ternary-intelligence-stack/main/albert-moe-13/benchmarks/install.sh | sh

Windows — PowerShell (open Start → search "PowerShell")

irm https://raw.githubusercontent.com/eriirfos-eng/ternary-intelligence-stack/main/albert-moe-13/benchmarks/install.ps1 | iex

Windows — Command Prompt / cmd.exe (if PowerShell is not your default shell)

powershell -ExecutionPolicy Bypass -Command "irm https://raw.githubusercontent.com/eriirfos-eng/ternary-intelligence-stack/main/albert-moe-13/benchmarks/install.ps1 | iex"

What it measures

#	Benchmark	Output
1	Inference speed	tok/s, ms/tok, GFLOPS
2	@sparseskip routing	active/skipped experts, compute ratio
3	Perplexity	avg cross-entropy, PPL on WikiText-2 (150KB, held-out)

Results exported to albert_bench_results.csv automatically.

Architecture

• 17L × 12E × 256H — native ternary MoE, weights in {-γ, 0, +γ} throughout
• @sparseskip Top-3/12: 75% of experts skipped per decode step (Patent A50296/2026)
• Ternary Traffic Light Routing: per-expert trit execution budget

Platform support

Platform	Method
Linux x86_64	pre-built binary
Windows x86_64	pre-built binary
macOS (any)	builds from source (~5-10 min, requires Xcode CLT)
Linux ARM64	builds from source (~10 min)
Android / Termux	builds from source (~10-20 min)

Notes

• This release packages the v2.0.0 checkpoint (pre-surgery baseline)