HN user

worstestes

352 karma
Posts13
Comments65
View on HN

Very interesting research!

Given that you're using cosine similarity of text embeddings to approximate the influence of individual tokens in a prompt, how does this approach fare in capturing higher-order interactions between tokens, something that Integrated Gradients (allegedly) is designed to account for? Are there specific scenarios where the cosine similarity method might fall short in capturing the nuances that Integrated Gradients can reveal?

Ship of Theseus 4 years ago

I think that while the original composition of the team changes the traditions, attitudes and strategy definitely translates across the different generations of players, coaches and staff. It’d be different if a team was completely cycled through to a new set of members every period of time.

Sure, it’s a great idea and I’m onboard… however agriculture uses approximately 80% of the Colorado River's water. Let’s not pretend that personal water use for lawns has nearly as much impact as water demanding crops and inefficient allocation to livestock farms.