Fathom: Per-Query Read Depth for Sparse Decoding over Offloaded KV Caches

arXiv:2609.17652 · cs.LG, cs.CL, cs.DC · Submitted 2026-09-15 · Read on arXiv

cs.LG, cs.CL, cs.DC

Submitted: 2026-09-15

Updated: 2026-09-18

Comments: 19 pages, 11 figures, 21 tables. Code and results: https://github.com/vivekkalyanarangan30/fathom

Code: https://github.com/vivekkalyanarangan30/fathom

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Sparsity — 48× in this paper · explained in 4 episodes
• Decode — 21× in this paper · explained in 2 episodes
• Calibration — 19× in this paper · explained in 7 episodes
• KV cache — 17× in this paper · explained in 3 episodes
• Quantization — 11× in this paper · explained in 12 episodes
• Fidelity — 10× in this paper · explained in 2 episodes
• Ablation — 5× in this paper · explained in 1 episode
• Accuracy — 5× in this paper · explained in 1 episode
• Sparse Attention — 5× in this paper · explained in 5 episodes
• In International Conference on Machine Learning — 3× in this paper
• In International Conference on Learning Representations — 2× in this paper
• Transformer — 2× in this paper · explained in 2 episodes
• Core — 1× in this paper · explained in 2 episodes
• Downstream Tasks — 1× in this paper · explained in 2 episodes
• Trace — 1× in this paper · explained in 2 episodes
• Transformer Models — 1× in this paper · explained in 2 episodes

Sources

• RetrievalAttention: Accelerating Long-Context LLM Inference via Vector Retrieval
• The Llama 3 Herd of Models
• Qwen3 Technical Report
• Sparse Attention as a Range Searching Problem: Towards an Inference-Efficient Index for KV Cache
• Post-Training Sparse Attention with Double Sparsity

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: Local-Consistency Optimisation for Bayesian Neural Networks
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks