No article text was recorded for this version.
straight-through estimator limitations
Version 1 of 1127 papers141 conceptsfrozen 2026-10-04
Papers in this version
- Blog: Efficient Quantization for Neural Networks with Binary Weights and Low Bitwidth Activations
- Blog: Learning Low-precision Neural Networks without Straight-Through Estimator (STE)
- Blog: BinaryDuo: Reducing Gradient Mismatch in Binary Activation Network by Coupling Binary Activations
- Blog: SURGE: Surrogate Gradient Adaptation in Binary Neural Networks
- Blog: Digital Wireless Image Transmission via Distribution Matching
- Blog: StableQAT: Stable Quantization-Aware Training at Ultra-Low Bitwidths
- Blog: Network Quantization with Element-wise Gradient Scaling
- Blog: Straightening Out the Straight-Through Estimator: Overcoming Optimization Challenges in Vector Quantized Networks
- Blog: Learning Sparse Low-Precision Neural Networks With Learnable Regularization
- Blog: Deep Learning with Low Precision by Half-Wave Gaussian Quantization
- Blog: PV-Tuning: Beyond Straight-Through Estimation for Extreme LLM Compression
- Blog: Integer Discrete Flows and Lossless Compression
- Blog: Multi-Precision Quantized Neural Networks via Encoding Decomposition of {-1, +1}
- Blog: Relaxed Quantization for Discretized Neural Networks
- Blog: Training Quantized Neural Networks With a Full-Precision Auxiliary Module
- Blog: Mixed Precision DNNs: All you need is a good parametrization
- Blog: Learning Discretized Neural Networks under Ricci Flow
- Blog: One-Step Forward and Backtrack: Overcoming Zig-Zagging in Loss-Aware Quantization Training
- Blog: Octo: INT8 Training with Loss-aware Compensation and Backward Quantization for Tiny On-device Learning
- Blog: Robust Training of Neural Networks at Arbitrary Precision and Sparsity
- Blog: Training Quantized Network with Auxiliary Gradient Module
- Blog: Learning Discrete Weights and Activations Using the Local Reparameterization Trick
- Blog: Fast and Slow Gradient Approximation for Binary Neural Network Optimization
- Blog: Hessian-aware Quantized Node Embeddings for Recommendation
- Blog: Network Quantization With Element-Wise Gradient Scaling
- Blog: Improving Quantization-Aware Training of Low-Precision Network Via Block Replacement on Full-Precision Counterpart
- Blog: Improving Quantization-aware Training of Low-Precision Network via Block Replacement on Full-Precision Counterpart
- Blog: QuEST: Stable Training of LLMs with 1-Bit Weights and Activations
- Blog: End-To-End Supervised Product Quantization for Image Search and Retrieval
- Blog: Error-aware Quantization through Noise Tempering
- Blog: Simultaneously Optimizing Weight and Quantizer of Ternary Neural Network Using Truncated Gaussian Approximation
- Blog: MKQ-BERT: Quantized BERT with 4-bits Weights and Activations
- Blog: Propagating Asymptotic-Estimated Gradients for Low Bitwidth Quantized Neural Networks
- Blog: CSQ: Growing Mixed-Precision Quantization Scheme with Bi-level Continuous Sparsification
- Blog: CAGE: Curvature-Aware Gradient Estimation For Accurate Quantization-Aware Training
- Blog: Restructuring Vector Quantization with the Rotation Trick
- Blog: MetaQuant: Learning to Quantize by Learning to Penetrate Non-differentiable Quantization
- Blog: An Empirical study of Binary Neural Networks' Optimisation
- Blog: Restructuring Vector Quantization with the Rotation Trick
- Blog: Learning Frequency Domain Approximation for Binary Neural Networks
- Blog: SCALED : Surrogate-gradient for Codec-Aware Learning of Downsampling in ABR Streaming
- Blog: Hessian-aware Quantized Node Embeddings for Recommendation
- Blog: QGait: Toward Accurate Quantization for Gait Recognition with Binarized Input
- Blog: Gradient Estimation Methods of Approximate Multipliers for High-Accuracy Retraining of Deep Learning Models
- Blog: LongVQ: Long Sequence Modeling with Vector Quantization on Structured Memory
- Blog: AdaSTE: An Adaptive Straight-Through Estimator to Train Binary Neural Networks
- Blog: SYQ: Learning Symmetric Quantization for Efficient Deep Neural Networks
- Blog: Differentiable Soft Quantization: Bridging Full-Precision and Low-Bit Neural Networks
- Blog: Continuous Approximations for Improving Quantization Aware Training of LLMs
- Blog: Improving Quantization-aware Training of Low-Precision Network via Block Replacement on Full-Precision Counterpart
- Blog: Single-Path Bit Sharing for Automatic Loss-Aware Model Compression
- Blog: HadaNets: Flexible Quantization Strategies for Neural Networks
- Blog: Coded Deep Learning: Framework and Algorithm
- Blog: Sparsity-Control Ternary Weight Networks
- Blog: DiVeQ: Differentiable Vector Quantization Using the Reparameterization Trick
- Blog: BiPer: Binary Neural Networks Using a Periodic Function
- Blog: Training Quantised Neural Networks with STE Variants: the Additive Noise Annealing Algorithm
- Blog: Quantized Inverse Design for Photonic Integrated Circuits
- Blog: Fast and Slow Gradient Approximation for Binary Neural Network Optimization
- Blog: Toward Efficient Low-Precision Training: Data Format Optimization and Hysteresis Quantization
Showing 60 of 127 papers.
History & citation
Current · Permanent link
- Addressing Gradient Mismatch in Quantized Neural Networks merged into this direction (74% overlap)
Cite this direction
Latest version (a living page):
@misc{lacuna_straight_through_estimator_limitations_9034b95b,
title = {straight-through estimator limitations},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/straight-through-estimator-limitations/rdr_9034b95bba254e9eb87c07b26f26054f},
note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}This exact version (version 1, frozen text):
@misc{lacuna_straight_through_estimator_limitations_9034b95b_v1,
title = {straight-through estimator limitations},
author = {{Lacuna}},
howpublished = {Lacuna research direction},
year = {2026},
url = {https://lacuna.tiptreesystems.com/direction/straight-through-estimator-limitations/rdr_9034b95bba254e9eb87c07b26f26054f/v/1},
note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}