We're sunsetting PodQuest on 2025-07-28. Thank you for your support!
Export Podcast Subscriptions
back
“Anomalous Tokens in DeepSeek-V3 and r1” by henry
18:37
Share
2025/1/28
LessWrong (Curated & Popular)
AI Chapters
Transcribe
Chapters
What Process Was Used to Find Anomalous Tokens in DeepSeek?
What Are Fragment Tokens and How Do They Behave?
What Are the Observed Behaviors of Other English Tokens?
What Non-English Tokens Were Found and How Do They Differ?
What Are the Non-English Outliers and Their Unique Characteristics?
What Special Tokens Were Identified and How Do They Function?
What Is the Base Model Mode and How Does It Affect Anomalous Tokens?
What's Next in the Research on Anomalous Tokens?
Shownotes
Transcript
No transcript made for this episode yet, you may request it for free.