TY  - RPRT
TI  - Position: On the Methodological Pitfalls of Evaluating Base LLMs for Reasoning
AU  - Jason Chan
AU  - Zhixue Zhao
AU  - Robert Gaizauskas
PY  - 2025
UR  - https://arxiv.org/abs/2511.10381
ID  - 2511.10381
ER  -