ResidualQuant: KV Cache Quantization for Looped Transformers with 2-Bit Residuals
cs.LG
Submitted: 2026-10-07
Updated: 2026-10-07
Code: https://github.com/Superone77/FlashLoop
Project page: https://seal-kaist.github.io/projects/ResidualQuant
Terminology
Sources
- Program Synthesis with Large Language Models
- Evaluating Large Language Models Trained on Code
- LoopSpec: Pipelined Self-Speculative Decoding for Looped Transformers
- Training Verifiers to Solve Math Word Problems
- LoopQ: Quantization for Looped Language Models
- Adaptive Computation Time for Recurrent Neural Networks
- Less is More: Recursive Reasoning with Tiny Networks
- Thinking with Looped Flows
- Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models
- Hierarchical Reasoning Model
- SMELT: Scaling Laws for Compute-Matched MoE Looped Transformers
- RecurTrace: Adaptive Latent Reasoning with Loop-Time Memory
- Parallel Loop Transformer for Efficient Test-Time Computation Scaling
- FlashLoop: Fast and Memory-Efficient Looped Transformers via Lazy Updates
- Output-Aware Rotation for INT2 KV-Cache Quantization
- OSCAR: Offline Spectral Covariance-Aware Rotation for 2-bit KV Cache Quantization
- Scaling Latent Reasoning via Looped Language Models
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks