How to Parameterize Asymmetric Quantization Ranges for Quantization-Aware Training
Not on the map. Why?
Abstract (arXiv)
This paper investigates three different parameterizations of asymmetric uniform quantization for quantization-aware training: (1) scale and offset, (2) minimum and maximum, and (3) beta and gamma. We perform a comprehensive comparative analysis of these parameterizations' influence on quantization-aware training, using both controlled experiments and real-world large language models. Our particular focus is on their changing behavior in response to critical training hyperparameters, bit width and learning rate. Based on our investigation, we propose best practices to stabilize and accelerate quantization-aware training with learnable asymmetric quantization ranges.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 29 of its 33 references that are in the map. First posted on arXiv 25 Apr 2024.
Builds on
29 of 33 references.
References · 29
- 2020Language Models are Few-Shot Learnerscited by 32,199
- 2023GPT-4 Technical Reportcited by 18,132
- 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
- 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,304
- 2016Pointer Sentinel Mixture Modelscited by 2,992
- 2022Training Compute-Optimal Large Language Modelscited by 2,927
- 2013Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computationcited by 2,031
- 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,838
- 2022SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Modelscited by 1,338
- 2023AWQ: Activation-aware Weight Quantization for LLM Compression and Accelerationcited by 1,234
- 2023Scaling Vision Transformers to 22 Billion Parameterscited by 690
- 2019Learned Step Size Quantizationcited by 604
17 more
- 2023Qwen-Audio: Advancing Universal Audio Understanding via Unified Large-Scale Audio-Language Modelscited by 591
- 2018Quantizing deep convolutional networks for efficient inference: A whitepapercited by 507
- 2018PACT: Parameterized Clipping Activation for Quantized Neural Networkscited by 456
- 2021A White Paper on Neural Network Quantizationcited by 453
- 2023OmniQuant: Omnidirectionally Calibrated Quantization for Large Language Modelscited by 403
- 2023LLM-QAT: Data-Free Quantization Aware Training for Large Language Modelscited by 320
- 2022Estimating the Carbon Footprint of BLOOM, a 176B Parameter Language Modelcited by 264
- 2020LSQ+: Improving low-bit quantization through learnable offsets and better initializationcited by 180
- 2023Memory-Efficient Fine-Tuning of Compressed Large Language Models via sub-4-bit Integer Quantizationcited by 97
- 2023EfficientDM: Efficient Quantization-Aware Fine-Tuning of Low-Bit Diffusion Modelscited by 77
- 2023QLLM: Accurate and Efficient Low-Bitwidth Quantization for Large Language Modelscited by 63
- 2023Pruning vs Quantization: Which is Better?cited by 45
- 2023CBQ: Cross-Block Quantization for Large Language Modelscited by 34
- 2022Neural Network Quantization with AI Model Efficiency Toolkit (AIMET)cited by 24
- 2023Understanding INT4 Quantization for Transformer Models: Latency Speedup, Composability, and Failure Casescited by 23
- 2023Token-Scaled Logit Distillation for Ternary Weight Generative Language Modelscited by 20
- 2018Highly Efficient 8-bit Low Precision Inference of Convolutional Neural Networks with IntelCaffecited by 4
Built on by
Citing papers per year: 2025: 2.
2 papers.