# FastE: Readout-Triggered Token Compression for LLM Embedding Inference

arXiv: https://arxiv.org/abs/2609.08407 · HTML: https://mlsanity.com/paper/2609.08407/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in KV cache optimization, one of 12 directions in Efficient LLM inference and serving.
- Builds on 13 of its 23 references that are in the map.
- First posted on arXiv 8 Sep 2026.

## Placed in

- Efficient LLM inference and serving / [KV cache optimization](https://mlsanity.com/inference/kv-cache-optimization/)

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

- [Efficient Streaming Language Models with Attention Sinks](https://mlsanity.com/paper/2309.17453/) (2023)
- [Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models](https://mlsanity.com/paper/2506.05176/) (2025)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [SnapKV: LLM Knows What You are Looking for Before Generation](https://mlsanity.com/paper/2404.14469/) (2024)
- [NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models](https://mlsanity.com/paper/2405.17428/) (2024)
- [LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders](https://mlsanity.com/paper/2404.05961/) (2024)
- [PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction](https://mlsanity.com/paper/2410.17247/) (2024)
- [Qwen3-VL-Embedding and Qwen3-VL-Reranker: A Unified Framework for State-of-the-Art Multimodal Retrieval and Ranking](https://mlsanity.com/paper/2601.04720/) (2026)
- [Pooling And Attention: What Are Effective Designs For LLM-Based Embedding Models?](https://mlsanity.com/paper/2409.02727/) (2024)
- [Interpretable Text Embeddings and Text Similarity Explanation: A Survey](https://mlsanity.com/paper/2502.14862/) (2025)
- [RTPrune: Reading-Twice Inspired Token Pruning for Efficient DeepSeek-OCR Inference](https://mlsanity.com/paper/2605.00392/) (2026)
- [The Truth Lies Somewhere in the Middle (of the Generated Tokens)](https://mlsanity.com/paper/2605.09969/) (2026)
- [Practical token pruning for foundation models in few-shot conversational virtual assistant systems](https://mlsanity.com/paper/2408.11799/) (2024)

## Abstract (arXiv)

In this study, we identify depth-dependent prefix redundancy in final-readout LLM embedding models, notably across representative backbones including Qwen3-Embedding and Qwen3-VL-Embedding. We find that removing prefix states is substantially more damaging in shallow layers than at greater depth, showing that prefix states become increasingly compressible as the prefix and readout states propagate through the network. To this end, we introduce FastE, a training-free, plug-and-play method. FastE uses a shared fixed threshold on batch-mean readout-prefix alignment as a lightweight online heuristic for selecting when compression occurs, and ranks prefix states by the attention scores they receive from the readout position to determine which states are retained in subsequent layers. Our evaluations demonstrate FastE's ability to substantially reduce computational costs: on NarrativeQA with Qwen3-Embedding-0.6B, it reduces decoder-backbone FLOPs by 40.11% while retaining 99.53% of Full Forward nDCG@10. Across five text embedding benchmarks, two backbone scales, and three cross-modal retrieval tasks, the quality-efficiency trade-off is directly customizable through the maximum removal ratio without retraining. We believe FastE offers practical value for scalable embedding generation in retrieval, indexing, clustering, and multimodal representation systems.