From 70dcf78816f3cd81f51a2a81b76d0effca3485d2 Mon Sep 17 00:00:00 2001
From: Lum0s517 <132797314+Lum0s517@users.noreply.github.com>
Date: Tue, 7 Jul 2026 19:16:48 +0800
Subject: [PATCH] update paper link to arXiv version (#110)

* Update README.md

* Update fake_inference.py

* Update fake_inference.py

* update paper link to arXiv version

* update paper link to arXiv version

* Update fake_inference.py
---
 README.md                 | 6 +++---
 README.zh-CN.md           | 6 +++---
 scripts/fake_inference.py | 1 -
 3 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/README.md b/README.md
index dbbc33a..d38cf48 100644
--- a/README.md
+++ b/README.md
@@ -41,9 +41,9 @@ package installation.
 - [June 2026] Wall-X 1.1.0 updates the open-source training and inference
   stack for Wall-OSS-0.5, including the public serving/evaluation runtime,
   DMuon training support, and install-time CUDA operator builds.
-- [May 2026] We introduce [**WALL-WM: Carving World Action Modeling at the Event Joints**](https://x2robot.com/api/files/file/WALL-WM.pdf), a World Action Model that couples future-video imagination with action prediction at semantic event boundaries.
-- [May 2026] We introduce [**Wall-OSS-0.5: A Deployment-Ready VLA with Gradient-Bridged Pretraining**](https://x2robot.com/api/files/file/wall_oss_05.pdf), an open-source model for directly deployable real-robot manipulation and downstream adaptation.
-- [Sept 2025] We introduce [**WALL-OSS: Igniting VLMs toward the Embodied Space**](https://x2robot.com/en/research/68bc2cde8497d7f238dde690), an end-to-end embodied foundation model that leverages large-scale multimodal pretraining to achieve embodiment-aware vision-language understanding, language-action association, and robust manipulation capability.
+- [May 2026] We introduce [**WALL-WM: Carving World Action Modeling at the Event Joints**](https://arxiv.org/pdf/2606.01955), a World Action Model that couples future-video imagination with action prediction at semantic event boundaries.
+- [May 2026] We introduce [**Wall-OSS-0.5: A Deployment-Ready VLA with Gradient-Bridged Pretraining**](https://arxiv.org/pdf/2605.30877v2), an open-source model for directly deployable real-robot manipulation and downstream adaptation.
+- [Sept 2025] We introduce [**WALL-OSS: Igniting VLMs toward the Embodied Space**](https://arxiv.org/pdf/2509.11766), an end-to-end embodied foundation model that leverages large-scale multimodal pretraining to achieve embodiment-aware vision-language understanding, language-action association, and robust manipulation capability.
 
 ## Models
 
diff --git a/README.zh-CN.md b/README.zh-CN.md
index 81e73d9..057cd0b 100644
--- a/README.zh-CN.md
+++ b/README.zh-CN.md
@@ -5,9 +5,9 @@ Wall-X 开源仓库提供 WALL 系列开源具身基础模型的训练与推理
 ## 最新动态
 
 - 2026 年 6 月：Wall-X 1.1.0 更新 Wall-OSS-0.5 训练与推理栈，包含公开 serving/evaluation runtime、DMuon 训练支持，以及安装期 CUDA 算子编译。
-- 2026 年 5 月：发布 [WALL-WM: Carving World Action Modeling at the Event Joints](https://x2robot.com/api/files/file/WALL-WM.pdf)。
-- 2026 年 5 月：发布 [Wall-OSS-0.5: A Deployment-Ready VLA with Gradient-Bridged Pretraining](https://x2robot.com/api/files/file/wall_oss_05.pdf)。
-- 2025 年 9 月：发布 [WALL-OSS: Igniting VLMs toward the Embodied Space](https://x2robot.com/en/research/68bc2cde8497d7f238dde690)。
+- 2026 年 5 月：发布 [WALL-WM: Carving World Action Modeling at the Event Joints](https://arxiv.org/pdf/2606.01955)。
+- 2026 年 5 月：发布 [Wall-OSS-0.5: A Deployment-Ready VLA with Gradient-Bridged Pretraining](https://arxiv.org/pdf/2605.30877v2)。
+- 2025 年 9 月：发布 [WALL-OSS: Igniting VLMs toward the Embodied Space](https://arxiv.org/pdf/2509.11766)。
 
 ## 模型
 
diff --git a/scripts/fake_inference.py b/scripts/fake_inference.py
index 1d69425..d431d2c 100755
--- a/scripts/fake_inference.py
+++ b/scripts/fake_inference.py
@@ -154,4 +154,3 @@ def main() -> int:
 
 if __name__ == "__main__":
     raise SystemExit(main())
-