News for hackers. Stuff that matters.

A comment on Beating OpenAI CLIP with 100x less data and compute.

Re:Beating OpenAI CLIP with 100x less data and compute by ilaksh on Tuesday February 28, 2023 @12:03PM (#34971736)
This may be a dumb question, but would it be possible to apply these techniques to something like text completion and/or visual question answering? If you went ahead and used the optimizations but still scaled the model up?
[ Parent | on HN ]
Re:Beating OpenAI CLIP with 100x less data and compute by vov_or on Tuesday February 28, 2023 @12:23PM (#34972041)
Yes, it is possible. Approaches, on which our model is based, are capable to solve VQA and other similar tasks showing SOTA results.
[ Parent | on HN ]
Re:Beating OpenAI CLIP with 100x less data and compute by freediver on Tuesday February 28, 2023 @12:46PM (#34972370)
Do you have/plan to have a text embeddings model?
[ Parent | on HN ]
Re:Beating OpenAI CLIP with 100x less data and compute by vov_or on Tuesday February 28, 2023 @01:53PM (#34973246)
Yes, we are training text embedding models right now. And also have plans to open-source some of them! In addition, we train encoders for different modalities with retrieval purposes. For example, video data.
[ Parent | on HN ]
Re:Beating OpenAI CLIP with 100x less data and compute by ilaksh on Tuesday February 28, 2023 @04:18PM (#34974995)
Do you know anyone working on a large text completion model based on it?
[ Parent | on HN ]