v0.3: AudioCaps in-domain fine-tune stage (a2t R@10 0.741, R@1 0.332)

Model card: architecture diagram; fix bank size to 484K (v0.2)

Add architecture diagram

Document version tags

from_pretrained: revision pinning

v0.2 retrieval examples in model card

v0.2 audio-to-image retrieval gallery

Bold best-per-column only in comparison tables

Bold v0.2 numbers in comparison tables

v0.2 model card with v0.1 comparison

v0.2 config

v0.2: 484K-pair corpus checkpoint (AudioCaps 0.717, Clotho-ZS 0.448)

Add VGGSound attribution to gallery

Embed retrieval gallery in model card

Add audio-to-image retrieval gallery

Add full cross-modal metrics and qualitative retrieval examples

Upload config.json with huggingface_hub

Upload inference.py with huggingface_hub

Upload README.md with huggingface_hub

Upload fusion-embedding-1-2b-preview.pt with huggingface_hub

initial commit