<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Tong Niu — Thoughts</title><description>Short essays and notes on machine learning, mathematics, and whatever else I am thinking about.</description><link>https://wolfniu.github.io/</link><language>en</language><item><title>Neural networks need sleep</title><link>https://wolfniu.github.io/thoughts/neural-networks-need-sleep/</link><guid isPermaLink="true">https://wolfniu.github.io/thoughts/neural-networks-need-sleep/</guid><description>We borrowed neurons from the brain and left out sleep. In a toy test, a network that prunes and dreams every night keeps its old task at 96%. One that studies all night falls to 68%.</description><pubDate>Thu, 12 Nov 2026 00:00:00 GMT</pubDate><category>machine learning</category><category>continual learning</category><author>Tong Niu</author></item><item><title>Quantization is just a histogram</title><link>https://wolfniu.github.io/thoughts/quantization-is-a-histogram/</link><guid isPermaLink="true">https://wolfniu.github.io/thoughts/quantization-is-a-histogram/</guid><description>A 4-bit model is a 16-bar histogram of its own weights. Bit-width, calibration, and NF4 all fall out of that one picture.</description><pubDate>Thu, 12 Nov 2026 00:00:00 GMT</pubDate><category>machine learning</category><category>quantization</category><author>Tong Niu</author></item></channel></rss>