When Precision Meets Position: BFloat16 Breaks Down RoPE in Long-Context Training
Fuente:
arXiv
Saved in:
| Main Authors: | Wang, Haonan, Liu, Qian, Du, Chao, Zhu, Tongyao, Du, Cunxiao, Kawaguchi, Kenji, Pang, Tianyu |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Similar Items
Demystifying the Slash Pattern in Attention: The Role of RoPE
by: Cheng, Yuan, et al.
Published: (2026)
by: Cheng, Yuan, et al.
Published: (2026)
RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, Provably
by: Du, Yufeng, et al.
Published: (2026)
by: Du, Yufeng, et al.
Published: (2026)
LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification
by: Yang, Penghui, et al.
Published: (2025)
by: Yang, Penghui, et al.
Published: (2025)
Think in Parallel, Answer as One: Logit Averaging for Open-Ended Reasoning
by: Wang, Haonan, et al.
Published: (2025)
by: Wang, Haonan, et al.
Published: (2025)
Periodic RoPE for Infinite Context LLMs
by: Huo, Simin
Published: (2026)
by: Huo, Simin
Published: (2026)
Base of RoPE Bounds Context Length
by: Men, Xin, et al.
Published: (2024)
by: Men, Xin, et al.
Published: (2024)
Shuffle the Context: RoPE-Perturbed Self-Distillation for Long-Context Adaptation
by: Li, Zichong, et al.
Published: (2026)
by: Li, Zichong, et al.
Published: (2026)
Understanding the RoPE Extensions of Long-Context LLMs: An Attention Perspective
by: Zhong, Meizhi, et al.
Published: (2024)
by: Zhong, Meizhi, et al.
Published: (2024)
Rotary Positional Embeddings as Phase Modulation: Theoretical Bounds on the RoPE Base for Long-Context Transformers
by: Liu, Feilong
Published: (2026)
by: Liu, Feilong
Published: (2026)
Q-ROAR: Outlier-Aware Rescaling for RoPE Position Interpolation in Quantized Long-Context LLMs
by: Qiao, Ye, et al.
Published: (2025)
by: Qiao, Ye, et al.
Published: (2025)
CoPE: Clipped RoPE as A Scalable Free Lunch for Long Context LLMs
by: Li, Haoran, et al.
Published: (2026)
by: Li, Haoran, et al.
Published: (2026)
RoPE-LIME: RoPE-Space Locality + Sparse-K Sampling for Efficient LLM Attribution
by: Picov, Isaac, et al.
Published: (2026)
by: Picov, Isaac, et al.
Published: (2026)
ReRoPE: Repurposing RoPE for Relative Camera Control
by: Li, Chunyang, et al.
Published: (2026)
by: Li, Chunyang, et al.
Published: (2026)
LightTransfer: Your Long-Context LLM is Secretly a Hybrid Model with Effortless Adaptation
by: Zhang, Xuan, et al.
Published: (2024)
by: Zhang, Xuan, et al.
Published: (2024)
Frayed RoPE and Long Inputs: A Geometric Perspective
by: Wertheimer, Davis, et al.
Published: (2026)
by: Wertheimer, Davis, et al.
Published: (2026)
RoPE Attention Can Be Trained in Almost Linear Time
by: Cao, Yang, et al.
Published: (2024)
by: Cao, Yang, et al.
Published: (2024)
Scaling Laws of RoPE-based Extrapolation
by: Liu, Xiaoran, et al.
Published: (2023)
by: Liu, Xiaoran, et al.
Published: (2023)
Spiral RoPE: Rotate Your Rotary Positional Embeddings in the 2D Plane
by: Liu, Haoyu, et al.
Published: (2026)
by: Liu, Haoyu, et al.
Published: (2026)
When Attention Sink Emerges in Language Models: An Empirical View
by: Gu, Xiangming, et al.
Published: (2024)
by: Gu, Xiangming, et al.
Published: (2024)
Rethinking RoPE: A Mathematical Blueprint for N-dimensional Positional Embedding
by: Liu, Haiping, et al.
Published: (2025)
by: Liu, Haiping, et al.
Published: (2025)
Resonance RoPE: Improving Context Length Generalization of Large Language Models
by: Wang, Suyuchen, et al.
Published: (2024)
by: Wang, Suyuchen, et al.
Published: (2024)
RoPECraft: Training-Free Motion Transfer with Trajectory-Guided RoPE Optimization on Diffusion Transformers
by: Gokmen, Ahmet Berke, et al.
Published: (2025)
by: Gokmen, Ahmet Berke, et al.
Published: (2025)
Circuit Complexity Bounds for RoPE-based Transformer Architecture
by: Chen, Bo, et al.
Published: (2024)
by: Chen, Bo, et al.
Published: (2024)
Diffusion MRI Transformer with a Diffusion Space Rotary Positional Embedding (D-RoPE)
by: Kung, Gustavo Chau Loo, et al.
Published: (2026)
by: Kung, Gustavo Chau Loo, et al.
Published: (2026)
Positional versus Symbolic Attention Heads: Learning Dynamics, RoPE Geometry, and Length Generalization
by: Urrutia, Felipe, et al.
Published: (2026)
by: Urrutia, Felipe, et al.
Published: (2026)
Jordan-RoPE: Non-Semisimple Relative Positional Encoding via Complex Jordan Blocks
by: Zhang, Yaobo
Published: (2026)
by: Zhang, Yaobo
Published: (2026)
PhaseWrap-RoPE: Bounded Phase-Wrap Positional Scoring Evidence and Hardware-Validation Tooling
by: Fredriksen, Daniel Eric
Published: (2026)
by: Fredriksen, Daniel Eric
Published: (2026)
PhaseWrap-RoPE: Bounded Phase-Wrap Positional Scoring Evidence and Hardware-Validation Tooling
by: Fredriksen, Daniel Eric
Published: (2026)
by: Fredriksen, Daniel Eric
Published: (2026)
PhaseWrap-RoPE: Bounded Phase-Wrap Positional Scoring Evidence and Hardware-Validation Tooling
by: Fredriksen, Daniel Eric
Published: (2026)
by: Fredriksen, Daniel Eric
Published: (2026)
PhaseWrap-RoPE: Bounded Phase-Wrap Positional Scoring Evidence and Hardware-Validation Tooling
by: Fredriksen, Daniel Eric
Published: (2026)
by: Fredriksen, Daniel Eric
Published: (2026)
PhaseWrap-RoPE: Bounded Phase-Wrap Positional Scoring Evidence and Hardware-Validation Tooling
by: Fredriksen, Daniel Eric
Published: (2026)
by: Fredriksen, Daniel Eric
Published: (2026)
PhaseWrap-RoPE: Bounded Phase-Wrap Positional Scoring Evidence and Hardware-Validation Tooling
by: Fredriksen, Daniel Eric
Published: (2026)
by: Fredriksen, Daniel Eric
Published: (2026)
PhaseWrap-RoPE: Bounded Phase-Wrap Positional Scoring Evidence and Hardware-Validation Tooling
by: Fredriksen, Daniel Eric
Published: (2026)
by: Fredriksen, Daniel Eric
Published: (2026)
Adaptive 3D-RoPE: Physics-Aligned Rotary Positional Encoding for Wireless Foundation Models
by: Zhang, Chenyu, et al.
Published: (2026)
by: Zhang, Chenyu, et al.
Published: (2026)
RAP: KV-Cache Compression via RoPE-Aligned Pruning
by: Xin, Jihao, et al.
Published: (2026)
by: Xin, Jihao, et al.
Published: (2026)
Untwisting RoPE: Frequency Control for Shared Attention in DiTs
by: Mikaeili, Aryan, et al.
Published: (2026)
by: Mikaeili, Aryan, et al.
Published: (2026)
A Circular Argument : Does RoPE need to be Equivariant for Vision?
by: van de Geijn, Chase, et al.
Published: (2025)
by: van de Geijn, Chase, et al.
Published: (2025)
On the token distance modeling ability of higher RoPE attention dimension
by: Hong, Xiangyu, et al.
Published: (2024)
by: Hong, Xiangyu, et al.
Published: (2024)
LinearARD: Linear-Memory Attention Distillation for RoPE Restoration
by: Yang, Ning, et al.
Published: (2026)
by: Yang, Ning, et al.
Published: (2026)
Rotate Both Ways: Time-and-Order RoPE for Generative Recommendation
by: Wei, Xiaokai, et al.
Published: (2025)
by: Wei, Xiaokai, et al.
Published: (2025)
Similar Items
-
Demystifying the Slash Pattern in Attention: The Role of RoPE
by: Cheng, Yuan, et al.
Published: (2026) -
RoPE Distinguishes Neither Positions Nor Tokens in Long Contexts, Provably
by: Du, Yufeng, et al.
Published: (2026) -
LongSpec: Long-Context Lossless Speculative Decoding with Efficient Drafting and Verification
by: Yang, Penghui, et al.
Published: (2025) -
Think in Parallel, Answer as One: Logit Averaging for Open-Ended Reasoning
by: Wang, Haonan, et al.
Published: (2025) -
Periodic RoPE for Infinite Context LLMs
by: Huo, Simin
Published: (2026)