I recently finished training the new version of NLLB-CLIP models that are better than previous ones and support dynamic embedding resolutions based on the MRL paper. You can read more about the models in the model cards.
I added the multilingual retrieval results and two new pretrained models.
Hi.
I recently finished training the new version of NLLB-CLIP models that are better than previous ones and support dynamic embedding resolutions based on the MRL paper. You can read more about the models in the model cards.
I added the multilingual retrieval results and two new pretrained models.