3 ms·
I've been working on (3) embeddings translation with the goal being to translate something like OpenAI embeddings to UAE-Large. So far, I have had success using
by batch12 3y ago
I've been working on (3) embeddings translation with the goal being to translate something like OpenAI embeddings to UAE-Large. So far, I have had success using them for cosine similarity with around a 99.99% validation rate, but only 80% using Euclidean distance.
- brookst 3y agoI’m fascinated by embeddings translations and compatible embeddings with different numbers of dimensions. Can you share more about your work / findings?
- sdenton4 3y agoI mean, the simplest answer is a matmul... Given embedding x, y, find M such that Mx ~= y. Easy to train so long as you've got access to both models to compute embedding over whatever you're interested in... (easy to extend to two layers mlp as needed. maybe ensure that x and y are zero mean and unit length to make training the matmul a bit easier.)