@upennnlp
@Penn Natural Language Processing group
AI
The researcher says benchmarks often change one obvious detail, letting shortcut checks score well.
AI
