<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Language Learning on Chengyu Wang</title><link>https://chengyu.eu/tags/language-learning/</link><description>Recent content in Language Learning on Chengyu Wang</description><generator>Hugo -- gohugo.io</generator><language>en</language><copyright>© 2026 Chengyu</copyright><lastBuildDate>Sat, 18 Apr 2026 17:00:16 +0000</lastBuildDate><atom:link href="https://chengyu.eu/tags/language-learning/index.xml" rel="self" type="application/rss+xml"/><item><title>Testing Models Against a Real Use Case, Not Just Benchmarks</title><link>https://chengyu.eu/posts/testing-pt-pt-language-models-real-scenario/</link><pubDate>Sat, 18 Apr 2026 17:00:16 +0000</pubDate><guid>https://chengyu.eu/posts/testing-pt-pt-language-models-real-scenario/</guid><description>Blind-testing two Ollama models against real CIPLE A2 Portuguese-exam questions, the well-documented, benchmark-topping model lost badly to an obscure community fine-tune — because the benchmarks were measuring Brazilian Portuguese, not European.</description></item></channel></rss>