RTN (LLM quantization): superseded — cited as a baseline and beaten by newer methods. 2 paper(s) critique it, 10 beat it on benchmarks — #6 of 80 most-superseded. Sub-problem: cluster led by RTN. Newer alternatives in the same sub-problem include STaR-Quant, Timestep-Aware SVDQuant-GPTQ, BWLA, Bit-by-Bit, Benford-Quant.

Is RTN superseded? Critiques, benchmarks & alternatives

What papers say

Verbatim critique sentences, each from a paper that cites RTN as a baseline.

the quantization error increases significantly when the number of bits is small, especially when significant outliers exist.

— CCQ: Convolutional Code for Extreme Low-bit Quantization in LLMs

“uniform grids spend disproportionate capacity on rare large magnitudes while under-resolving dense near-zero regions; the mismatch is exacerbated in layers whose weight magnitudes span multiple decades.”

— Benford's Law as a Distributional Prior for Post-Training Quantization of Large Language Models

Beaten on benchmarks

Head-to-head results where a newer method reports beating RTN. Values are copied from the source paper's tables — verify against the cited paper.

Reasoning-QAT beats RTN
62.38 vs 2.60
Avg. · [Qwen3-4B W3G128]
What Makes Low-Bit Quantization-Aware Training Work for Reasoning LLMs? A Systematic Study
BiLLM beats RTN
15.14 vs 1412020.25
perplexity · [LLaMA-13B 1-bit]
BiLLM: Pushing the Limit of Post-Training Quantization for LLMs
Benford-Quant beats RTN
70.87 vs 755.19
Perplexity · [3 bits, Small models]
Benford's Law as a Distributional Prior for Post-Training Quantization of Large Language Models
Timestep-Aware SVDQuant-GPTQ beats RTN
0.689 vs 0.389
Imaging Quality · [W4A4 quantization]
Timestep-Aware SVDQuant-GPTQ for W4A4 Quantization of Wan2.2-I2V
FAIR-Calib beats RTN
64.64 vs 42.00
Avg. · [Dream-Base]
FAIR-Calib: Frontier-Aware Instability-Reweighted Calibration for Post-Training Quantization of Diffusion Large Language Models
STaR-Quant beats RTN
57.07 vs 44.23
Avg. · [W4A4 (4-bit weight and activation) on LLADA-8B]
STaR-Quant: State-Time Consistent Post-Training Quantization for Diffusion Large Language Models
ADMM-Q beats RTN
11.86 vs 15.27
C4 PPL · [LLaMA-3 8B W4A4KV4 A, g128]
ADMM-Q: An Improved Hessian-based Weight Quantizer for Post-Training Quantization of Large Language Models
ApiQ beats RTN
5.77 vs 6.66
WikiText2 PPL · [LLaMA-2-7B, 3-bit]
Enhancing Ultra-Low-Bit Quantization of Large Language Models Through Saliency-Aware Partial Retraining

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.