M-CLIP/M-BERT-Distil-40

The model was tuned to match the embedding space for 40 languages, to the embedded space of the CLIP text encoder which accompanies the Res50x4 vision encoder . A full list of the 100 languages used during pre-training can be found here . The results can be viewed at Github.com/Gibo . The exception being that Greeks are apparently unable to recognize happy persons. The exception is that they seem to be incapable of recognizing happy persons, Greek and Spanish people. French, German, Spanish, Russian, Swedish and Greek it seems to yield respectable results for most instances. Kannada, a language which was included during pre training but not fine-tuning, performed,