The authors release a Hutsul-Ukrainian corpus and show LoRA-fine-tuned 7B models beat GPT-4o on automated and LLM-based metrics, but the evaluation is contaminated by overlapping training and test sources.
Mad families of Gowers' infinite block sequences
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Call a subset of $\mathbf{FIN}_k$ small if it does not contain a copy of $\langle{A\rangle}$ for some infinite block sequence $A \in \mathbf{FIN}_k^{[\infty]}$. Gowers' $\mathbf{FIN}_k$ theorem asserts that the set of small subsets of $\mathbf{FIN}_k$ forms an ideal, so it is sensible to consider almost disjoint families of $\mathbf{FIN}_k$ with respect to the ideal of small subsets of $\mathbf{FIN}_k$. We shall show that $\mathfrak{a}_{\mathbf{FIN}_k}$, the smallest possible cardinality of an infinite mad family of $\mathbf{FIN}_k$, is uncountable.
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Vuyko Mistral: Adapting LLMs for Low-Resource Dialectal Translation
The authors release a Hutsul-Ukrainian corpus and show LoRA-fine-tuned 7B models beat GPT-4o on automated and LLM-based metrics, but the evaluation is contaminated by overlapping training and test sources.