MetaEncoder represents a request and each candidate separately, then compares their vectors using a similarity score.
Bi-encoders can support multimodal decisions across small and massive choice sets
MetaEncoder uses one natural-language interface to score options in closed sets and retrieve items from collections containing millions of candidates.
Big Tech
Jianpeng Cheng · Guangyu Sun · Aashu Singh · Benyu Zhang · Haixing Dai · Hossein Mansour · +8 more
Meta AI
Research Digest··3 min read
Cheng et al.
Why this paper
From Meta AI
In one line
MetaEncoder turns a 30B multimodal decoder into a scalable bi-encoder that beats state-of-the-art multimodal embedders on decision tasks but remains weak on reasoning.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§