From 1d16e476368fe00ab55fcb7382aa281f25ea7513 Mon Sep 17 00:00:00 2001
From: Qihang Zhang <qihangz.work@gmail.com>
Date: Sun, 9 Nov 2025 15:41:39 +0800
Subject: [PATCH 1/2] [fix] fix the problem that blog won't automaticlt
 redirect.

---
 _posts/2025-10-11-max-ent-rl.md | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/_posts/2025-10-11-max-ent-rl.md b/_posts/2025-10-11-max-ent-rl.md
index d6a6567..12d6000 100644
--- a/_posts/2025-10-11-max-ent-rl.md
+++ b/_posts/2025-10-11-max-ent-rl.md
@@ -15,3 +15,10 @@ authors:
       name: UBC
 
 ---
+
+<script>
+  window.location.replace("https://qihang-zhang.com/Learning-Sys-Blog/2025/10/06/max-ent-rl-and-boltzmann-distribution.html");
+</script>
+
+If you are not redirected automatically, you can read the full post here:
+[Why the Exponential? From Max‑Entropy RL to the Boltzmann Distribution](https://qihang-zhang.com/Learning-Sys-Blog/2025/10/06/max-ent-rl-and-boltzmann-distribution.html).

From 3b924dead23d4d22b37fe0206bfd51408c3ab6ed Mon Sep 17 00:00:00 2001
From: Qihang Zhang <qihangz.work@gmail.com>
Date: Sun, 9 Nov 2025 16:07:23 +0800
Subject: [PATCH 2/2] [doc] Add the blog of wpoe

---
 _posts/2025-10-11-max-ent-rl.md   |  2 +-
 _posts/2025-11-09-weighted-poe.md | 25 +++++++++++++++++++++++++
 2 files changed, 26 insertions(+), 1 deletion(-)
 create mode 100644 _posts/2025-11-09-weighted-poe.md

diff --git a/_posts/2025-10-11-max-ent-rl.md b/_posts/2025-10-11-max-ent-rl.md
index 12d6000..6912889 100644
--- a/_posts/2025-10-11-max-ent-rl.md
+++ b/_posts/2025-10-11-max-ent-rl.md
@@ -1,6 +1,6 @@
 ---
 layout: distill
-title:  Why the Exponential? From Max‑Entropy RL to the Boltzmann Distribution
+title: Why the Exponential? From Max‑Entropy RL to the Boltzmann Distribution
 description: This blog post explores why the exponential function appears ubiquitously across modern RL, energy-based modeling, and statistical mechanics. We examine the connection between max-entropy reinforcement learning and the Boltzmann distribution, uncovering the fundamental principles that make the exponential form inevitable and explaining what "temperature" actually does in these frameworks.
 tags: reinforcement-learning information-theory boltzmann-distribution
 giscus_comments: true
diff --git a/_posts/2025-11-09-weighted-poe.md b/_posts/2025-11-09-weighted-poe.md
new file mode 100644
index 0000000..d851e60
--- /dev/null
+++ b/_posts/2025-11-09-weighted-poe.md
@@ -0,0 +1,25 @@
+---
+layout: distill
+title: Test-Time Steering for Lossless Text Compression via Weighted Product of Experts
+description: >
+  When I was a child, I always wondered: if I keep compressing the same file, will it eventually shrink to nothing? Of course, the answer is no—once a file is optimally compressed by a lossless compressor, compressing it again with the same method gives a file of exactly the same size. Today I know this comes from the fundamental limits of lossless compression in information theory. But what if we use multiple compressors instead of one? If we combine them, can each remove a different part of the data’s redundancy—and how should such a combination be designed? In this blog we discussed the above questions and proposed a method called Weighted Product of Experts.
+tags: large-language-models lossless-compression mixture-of-experts information-theory
+giscus_comments: true
+date: 2025-11-09
+featured: true
+redirect: https://qihang-zhang.com/Learning-Sys-Blog/2025/10/15/weighted-product-of-experts.html
+
+authors:
+  - name: Qihang Zhang
+    url: "https://qihang-zhang.com/"
+    affiliations:
+      name: UBC
+
+---
+
+<script>
+  window.location.replace("https://qihang-zhang.com/Learning-Sys-Blog/2025/10/15/weighted-product-of-experts.html");
+</script>
+
+If you are not redirected automatically, you can read the full post here:
+[Test-Time Steering for Lossless Text Compression via Weighted Product of Experts](https://qihang-zhang.com/Learning-Sys-Blog/2025/10/15/weighted-product-of-experts.html).