Glitch TokenBack to catalog
▁autorytatywna
The Polish word for “authoritative” (feminine form; ▁ is the SentencePiece space marker), the #2 most under-trained token in Phi-3 mini.
Raw token
"▁autorytatywna"
Observed behaviors
- 01Rank #2 most under-trained in the Phi-3 mini Magikarp repetition verification (embedding L2 norm 0.00200): a max self-repeat probability of just 6.4e-06.GitHub
- Observed or documented on
- Phi-3 (Microsoft)
- Behavior types
- Under-trained
- Tokenizer
LlamaTokenizer- Discovered by
- Sander Land & Max Bartolo