Research map / Architectures and efficiency / Quantization
Highly Efficient and Effective LLMs with Multi-Boolean Architectures
Ba-Hien Tran, Van Minh Nguyen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| increase number of Boolean kernels (ablation) | MBOK · OPT small | approximation error and perplexity | Improved |
| train only the last Boolean kernel instead of the first (ablation) | MBOK · OPT small | weight flips and perplexity | Improved |
| use 3 Boolean kernels instead of 3-bit OPTQ | MBOK · OPT-125M WikiText2 | perplexity: 53.85 -> 29.10 | Improved |
| use 3 Boolean kernels instead of 3-bit OPTQ | MBOK · OPT-6.7B WikiText2 | perplexity: 14.86 -> 11.03 | Improved |
| native Boolean training instead of latent weights (ablation) | MBOK · OPT | convergence speed | Improved |
| use 3 kernels vs MoS 3 experts (ablation) | MBOK · OPT | perplexity | Improved |
| allocate kernels by importance and residual error (fractional bits) (ablation) | MBOK · OPT small | perplexity | Improved |
| use MBOK at 2-bit budget instead of ShiftAddLLM 2.2-bit | MBOK · OPT small C4 | perplexity: 435.84 -> 32.23 | Improved |
| use MBOK instead of BitNet b1.58 finetuning at 1.58 bit | MBOK · OPT small C4 | perplexity: 10199.89 -> reasonable (not given) | Improved |
| use BitBLAS 1-bit kernels instead of FP16 weights | MBOK · LLaMA large linear layers | latency speedup: up to 8.7x | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- QuIP# — QuIP: 2-Bit Quantization of Large Language Models With Guarantees (NeurIPS 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PB-LLM: Partially Binarized Large Language Models (ICLR 2024 · similarity 0.93)
- ARB-LLM: Alternating Refined Binarizations for Large Language Models (ICLR 2025 · similarity 0.92)
- BiLLM: Pushing the Limit of Post-Training Quantization for LLMs (ICML 2024 · similarity 0.91)
- Any-Precision LLM: Low-Cost Deployment of Multiple, Different-Sized LLMs (ICML 2024 · similarity 0.90)
- Treasures in Discarded Weights for LLM Quantization (AAAI 2025 · similarity 0.90)
- Q-Palette: Fractional-Bit Quantizers Toward Optimal Bit Allocation for Efficient LLM Deployment (NeurIPS 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
