Prompt tuning in transformers is shown, via covering and Lipschitz arguments, to memorize at most linearly many examples in the prompt length.
Longrope: extending llm context window beyond 2 million tokens
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Memory Limitations of Prompt Tuning in Transformers
Prompt tuning in transformers is shown, via covering and Lipschitz arguments, to memorize at most linearly many examples in the prompt length.