Martin Ziqiao Ma
@marstin.bsky.social
https://mars-tin.github.io
phd<<<1,1>>>(UMich);
ex<<<3,1>>>({MIT_IBM_Watson, Adobe, Amazon});
Make the community better @ACLMentorship @GrowAI
Herborium Lover, Fortune Teller, Pokémon Trainer, Szechuan Cuisine Chef.
phd<<<1,1>>>(UMich);
ex<<<3,1>>>({MIT_IBM_Watson, Adobe, Amazon});
Make the community better @ACLMentorship @GrowAI
Herborium Lover, Fortune Teller, Pokémon Trainer, Szechuan Cuisine Chef.
Vision-Language Models are not yet pragmatically optimal.
We identify 3 key failures of pragmatic competence in referring expression generation with VLMs: (1) cannot uniquely refer to the referent, (2) include excessive or irrelevant information, and (3) misalign with human pragmatic preferences.
We identify 3 key failures of pragmatic competence in referring expression generation with VLMs: (1) cannot uniquely refer to the referent, (2) include excessive or irrelevant information, and (3) misalign with human pragmatic preferences.
April 23, 2025 at 5:55 PM
Vision-Language Models are not yet pragmatically optimal.
We identify 3 key failures of pragmatic competence in referring expression generation with VLMs: (1) cannot uniquely refer to the referent, (2) include excessive or irrelevant information, and (3) misalign with human pragmatic preferences.
We identify 3 key failures of pragmatic competence in referring expression generation with VLMs: (1) cannot uniquely refer to the referent, (2) include excessive or irrelevant information, and (3) misalign with human pragmatic preferences.