# Tied Trit-Planes: Constraining PTQTP to a Uniform Nine-Level Quantizer, with a Persistent Folded Format for Disk-Streamed Mixture-of-Experts Serving

arXiv: https://arxiv.org/abs/2608.08910 · HTML: https://mlsanity.com/paper/2608.08910/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in MoE large language models, one of 12 directions in Mixture-of-experts.
- Builds on 21 of its 32 references that are in the map.
- First posted on arXiv 9 Aug 2026.

## Placed in

- Mixture-of-experts / [MoE large language models](https://mlsanity.com/moe/moe-large-language-models/)

## Builds on (most cited first, 21 of 21)

- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Pointer Sentinel Mixture Models](https://mlsanity.com/paper/1609.07843/) (2016)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [QuIP#: Even Better LLM Quantization with Hadamard Incoherence and Lattice Codebooks](https://mlsanity.com/paper/2402.04396/) (2024)
- [LUT-GEMM: Quantized Matrix Multiplication based on LUTs for Efficient Inference in Large-Scale Generative Language Models](https://mlsanity.com/paper/2206.09557/) (2022)
- [T-MAC: CPU Renaissance via Table Lookup for Low-Bit LLM Deployment on Edge](https://mlsanity.com/paper/2407.00088/) (2024)
- [HOBBIT: A Mixed Precision Expert Offloading System for Fast MoE Inference](https://mlsanity.com/paper/2411.01433/) (2024)
- [QMoE: Practical Sub-1-Bit Compression of Trillion-Parameter Models](https://mlsanity.com/paper/2310.16795/) (2023)
- [Mixture of Quantized Experts (MoQE): Complementary Effect of Low-bit Quantization and Robustness](https://mlsanity.com/paper/2310.02410/) (2023)
- [GuidedQuant: Large Language Model Quantization via Exploiting End Loss Guidance](https://mlsanity.com/paper/2505.07004/) (2025)
- [Accuracy is Not All You Need](https://mlsanity.com/paper/2407.09141/) (2024)
- [QuantMoE-Bench: Examining Post-Training Quantization for Mixture-of-Experts](https://mlsanity.com/paper/2406.08155/) (2024)
- [PTQTP: Post-Training Quantization to Trit-Planes for Large Language Models](https://mlsanity.com/paper/2509.16989/) (2025)
- [DBQ: A Differentiable Branch Quantizer for Lightweight Deep Neural Networks](https://mlsanity.com/paper/2007.09818/) (2020)
- [Statistically-Lossless Quantization of Large Language Models](https://mlsanity.com/paper/2605.02404/) (2026)
- [CAT-Q: Cost-efficient and Accurate Ternary Quantization for LLMs](https://mlsanity.com/paper/2606.26650/) (2026)
- [Unifying Uniform and Binary-coding Quantization for Accurate Compression of Large Language Models](https://mlsanity.com/paper/2506.03781/) (2025)
- [Attend to Your Own Thoughts: Breaking the Barrier for Post-Training Quantization of Reasoning LLMs through the Lens of 1.58-Bit Quantization](https://mlsanity.com/paper/2608.01078/) (2026)
- [Divergent Token Metrics: Measuring degradation to prune away LLM components -- and optimize quantization](https://mlsanity.com/paper/2311.01544/) (2023)
- [Displacement Is Not Direction: Evaluating Fidelity Metrics for Quantized LLM Deployment](https://mlsanity.com/paper/2606.19558/) (2026)
- [BPDQ: Bit-Plane Decomposition Quantization on a Variable Grid for Large Language Models](https://mlsanity.com/paper/2602.04163/) (2026)

## Abstract (arXiv)

PTQTP decomposes LLM weight matrices into two ternary (trit) planes with two free per-group scales. Tying the scales to a fixed ratio of three collapses the decomposition into a single uniform nine-level quantizer, a known balanced-ternary identity. To our knowledge, at the time of writing, this work is the first to impose that identity as a constraint inside PTQTP's solver. The two trit planes then fold losslessly into one 4-bit code plane that we make the persistent serving representation: disk bytes, expert-cache bytes, and kernel input are the same 4.0625-bits/weight blocks, consumed in one integer dot pass. For this conjunction (ratio-3 nine-level code, CPU-SIMD kernels, SSD expert streaming, identical persistent bytes) we likewise found no precedent. We apply this to the routed experts of DeepSeek-V4-Flash-0731, a 284B-A13B mixture-of-experts model, quantizing in one shot from the released MXFP4 expert weights and streaming experts from SSD on a 64 GB laptop. Against a 4.5-bit Q4_K baseline, measured one process per fixture with an expert-lossless anchor arm as reference control, the tied model matches the official serving API on 5/5 fixtures at step 0 (Q4_K: 4/5) and 12/14 captured continuation steps (11/14), scores 86 vs. 84 on a 100-item MMLU subset, decodes 6.7% faster in decode phase, and ships 9% smaller files: no detected fidelity difference at these small evaluation sizes, and every fixture-level difference between the arms traces to a single measured near-tie cell. The tied fit nevertheless shows higher weight-reconstruction error and worse perplexity, a measured dissociation between proxy metrics and reference fidelity. A cumulative trunk-ternarization ladder and bitwise-pinned aarch64/x86-64 kernels complete the report. All code, formats, and evaluation artifacts are open source in the fucina inference stack.