---
cited_by:
  - '[[thoughts/inference/architectures#5-four-things-called-replay|README §5]]'
  - '[[thoughts/inference/architectures#6-reading-order|README §6]]'
  - '[[thoughts/inference/deepseek-v4.1-state#33-decoder-side-bounded-replay-58132-59532-59894|dsv41 §3.3]]'
  - '[[thoughts/inference/deepseek-v4.1-serving#4-pr-reading-order|dsv41 §4]]'
  - '[[thoughts/inference/replayssm#4-which-models-use-it|replayssm §4]]'
date: '2026-11-11'
kind: pr
local:
  - file:///Users/aarnphm/workspace/vllm/.buildkite/test_areas/models_basic.yaml
  - file:///Users/aarnphm/workspace/vllm/vllm/config/cache.py
merged: '2026-09-29'
number: 58132
source:
  - https://github.com/vllm-project/vllm/blob/4b2e1cfa7b8094f1a9547aee41c39e05d2ce5c97/vllm/models/deepseek_v41/nvidia/model_state.py
  - https://github.com/vllm-project/vllm/blob/4b2e1cfa7b8094f1a9547aee41c39e05d2ce5c97/vllm/models/deepseek_v41/nvidia/model.py
  - https://github.com/vllm-project/vllm/blob/4b2e1cfa7b8094f1a9547aee41c39e05d2ce5c97/vllm/models/deepseek_v41/decoder_replay_layers.py
  - https://github.com/vllm-project/vllm/blob/4b2e1cfa7b8094f1a9547aee41c39e05d2ce5c97/vllm/models/deepseek_v41/nvidia/vl_model.py
  - https://github.com/vllm-project/vllm/blob/4b2e1cfa7b8094f1a9547aee41c39e05d2ce5c97/.buildkite/test_areas/models_basic.yaml
  - https://github.com/vllm-project/vllm/blob/4b2e1cfa7b8094f1a9547aee41c39e05d2ce5c97/vllm/config/cache.py
state: merged
tags:
  - vllm
  - vllm-pr
  - new-architectures
title: '#58132 [Model] Decoder-side SWA bounded replay for DeepSeek-V4.1'
topic: new-architectures
url: https://github.com/vllm-project/vllm/pull/58132
created: '2026-11-11'
modified: '2026-11-11'
published: '2026-11-11'
pageLayout: default
slug: hinterland/papers/maps/prs/58132
permalink: https://aarnphm.xyz/hinterland/papers/maps/prs/58132.md
generator:
  quartz: v4.6.0
  hostedProvider: Cloudflare
  baseUrl: aarnphm.xyz
full: https://aarnphm.xyz/llms-full.txt
---
Decoder-side SWA bounded replay for DeepSeek-V4.1, on top of the encoder side ([#56227](https://github.com/aarnphm/aarnphm.github.io/issues/56227)). The layers past the last KV-source layer (21-39 of 40 on V4.1-Flash) own only sliding-window KV, so in eager prefill steps they run on each request’s… · full dump: [[thoughts/inference/vllm-prs/58132|#58132]]

