List your company

ResearchFormerly WizWand

Triplet

Benchmarks

TaskDatasetSOTA resultTrendResults

Image-referenced Sketch Colorization

Triplet 50K (val)

8.28FID

7

Audio-to-(Text+Image) Generation

triplet (text, image, audio)

36.79CLAP Score (A->T)

3

Image-to-(Text+Audio) Generation

triplet dataset text, image, audio

27.98CLIP Score (I->T)

3

Text-to-(Image+Audio) Generation

triplet (text, image, audio)

26.61CLIP Score (T->I)

3

Many-to-one Audio Generation ((T+I) → A)

Triplet text, image, audio

28.13CLAP Score (T->A)

3

Many-to-one Image Generation ((T+A) → I)

Triplet (text, image, audio)

25.57CLIP Score (T→I)

3

Many-to-one Text Generation ((I+A) → T)

Triplet (text, image, audio)

27.83CLIP Score (I->T)

3