# RadixMLP -- Intra-batch Deduplication for Causal Transformers

arXiv: https://arxiv.org/abs/2601.15013 · HTML: https://mlsanity.com/paper/2601.15013/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 13 of its 20 references that are in the map.
- First posted on arXiv 21 Jan 2026.

## Builds on (most cited first, 13 of 13)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning](https://mlsanity.com/paper/2307.08691/) (2023)
- [GLU Variants Improve Transformer](https://mlsanity.com/paper/2002.05202/) (2020)
- [SGLang: Efficient Execution of Structured Language Model Programs](https://mlsanity.com/paper/2312.07104/) (2023)
- [FlashInfer: Efficient and Customizable Attention Engine for LLM Inference Serving](https://mlsanity.com/paper/2501.01005/) (2025)
- [NanoFlow: Towards Optimal Large Language Model Serving Throughput](https://mlsanity.com/paper/2408.12757/) (2024)
- [Hydragen: High-Throughput LLM Inference with Shared Prefixes](https://mlsanity.com/paper/2402.05099/) (2024)
- [Preble: Efficient Distributed Prompt Scheduling for LLM Serving](https://mlsanity.com/paper/2407.00023/) (2024)
- [BatchLLM: Optimizing Large Batched LLM Inference with Global Prefix Sharing and Throughput-oriented Token Batching](https://mlsanity.com/paper/2412.03594/) (2024)
- [DeFT: Decoding with Flash Tree-attention for Efficient Tree-structured LLM Inference](https://mlsanity.com/paper/2404.00242/) (2024)
- [Boosting Distributed Training Performance of the Unpadded BERT Model](https://mlsanity.com/paper/2208.08124/) (2022)

## Abstract (arXiv)

Batch inference workloads for causal transformer models frequently process sequences that share common prefixes, such as system prompts, few-shot examples, or shared queries. Standard inference engines treat each sequence independently, redundantly recomputing identical MLP activations for every copy of the shared prefix. We introduce RadixMLP, a technique that exploits the position-wise nature of MLPs, LayerNorms, linear projections, and embeddings to eliminate this redundancy. RadixMLP dynamically maps batches to a prefix trie, gathering shared segments into a compressed representation for position-wise computation and scattering results back only at attention boundaries. RadixMLP is stateless and operates within a single forward pass. In end-to-end serving benchmarks on MS~MARCO v1.1 with Qwen3 models (0.6B to 8B parameters), RadixMLP achieves 1.44-1.59× speedups in realistic reranking workloads, with up to 5× speedups on synthetic benchmarks with longer shared prefixes. Our code is available at https://github.com/michaelfeil/radix-mlp.