MCERF delivers a 41.1% relative accuracy gain on the DesignQA benchmark by combining ColPali vision-language retrieval with four specialized reasoning modes and dynamic routing.
StepI- deator: Utilizing Mixed Representations to Support Step-By-Step Design With Generative Artificial Intelligence,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MCERF: Advancing Multimodal LLM Evaluation of Engineering Documentation with Enhanced Retrieval
MCERF delivers a 41.1% relative accuracy gain on the DesignQA benchmark by combining ColPali vision-language retrieval with four specialized reasoning modes and dynamic routing.