Gemma4-E4B More Prone to Errors in Multilingual Contexts
LChoshen · x · 2026-07-14
Researchers tested Gemma4-E4B on chess/TicTacToe-style tasks and found it is more prone to spatial relation errors in Arabic or Hebrew compared to English.
Specific failure modes include:
- Misidentifying a row of squares as a column or diagonal
- Failing to recognize an opponent's two-in-a-row threat
- This phenomenon was verified through large-scale testing
This indicates that the model's spatial reasoning capabilities can be inconsistent across different language inputs.
More from Models
- Gemini 3.6 Flash lands in Google AI Studio with cheaper output pricing — gaganghotra_ · 2026-07-21
- Jack Clark says OpenAI’s internal-deployment safety notes help the whole frontier community — jackclarkSF · 2026-07-21
- Mindlab Research puts Macaron-V1-Venti on Hugging Face — External_Mood4719 · 2026-07-21
- Google is surfacing Gemini 3.5 Flash-Lite and 3.6 Flash in AI Studio — Expensive_Syrup_6529 · 2026-07-21
- ChatGPT often explains the wall before answering whether it is tilting — Aware-sky-3489 · 2026-07-21
- Grok website traffic rose 38.15% YoY to 736 million Q2 visits — XFreeze · 2026-07-21