DeepSeek-V3 and r1 Found to Contain Anomalous Tokens Causing Bizarre Behavior
teortaxesTex · x · 2026-08-21
Research identifies "anomalous" or "glitch" tokens in DeepSeek-V3 and r1 that induce bizarre behavior, similar to the SolidGoldMagikarp phenomenon in GPT-2/3. The study utilized DeepSeek's tokenizer to extract vocabulary and automatically test for unusual outputs. The model's extensive Chinese training data complicates tokenization, leading to split characters and decoding difficulties.
Related event: Researchers Find Abnormal Tokens Causing Weird DeepSeek-V3 Outputs(2 posts)→
More from Models
- NVIDIA releases Nemotron 3.5 Lightning model and NeMo Switchyard router — nvidia · 2026-08-21
- Initial benchmarks emerge for Liquid AI's new LFM DSpark model — helloiamleonie · 2026-08-21
- Qwen 3.8 Max scores 58.7 on Surge AI benchmark, up 22.4 points — echen · 2026-08-21
- Tencent's flagship Hunyuan Hy4 surfaces in Yuanbao grey test ahead of launch — teortaxesTex · 2026-08-21
- Claude's Computer Use, Skills, and Files APIs are now generally available — EricBuess · 2026-08-21
- Users notice Claude acting like an "angry ex" in new interactions — ATTlKA · 2026-08-21