"Costco's credit card policy is all about saving money. It negotiates a partnership agreement with a single payment network in exchange for much lower processing fees. As part of its current deal with Visa, it pays transaction fees of only 0.4%. By saving on fees, Costco is able to keep prices low."
By not demanding better legislation, you're still being ripped off. What you describe as a "very nice low rate" is higher than the legislated EU wide limit.
But I get minimum 2% cash back when I use my card. Prices are 0.4% higher because of the fee, but I still come out way ahead. You can easily get 5% back on gas, groceries, Amazon purchases, etc. American credit cards are awesome if you know how to use them.
Funny thing is I'm still getting between 3% and 5% cashback on my US cards that I use in the EU. Those interchange fee limits are probably EU issued cards only.
That is probably your card issuer loosing money because US cards used in Europe is a small enough segment to not demand any policy changes.
In particular the articles talks about exemptions:
> provides for a limited number of exemptions, such as business cards used only for business expenses being charged directly to the account of the company;
Yes. We lost our beloved family golden retriever to VF back in 2009. It was a painful experience to see her degenerate, and the cough was something else. I remember treatment costing a pretty penny.
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?
Great question - there are currently (likely) tons of limitations to this approach as-is. We're planning on testing this on more capable models (e.g: integrated gradients on Llama2) to see how the relationship might change, but here are some initial thoughts:
1. The perturbation method could be improved to more directly capture long-range dependency information across tokens
2. The scoring method could _definitely_ be improved to capture more nuance across perturbations.
I think what we've found is that there does seem to be a relationship between the embedding space and attributions of LLMs, so the next step would be to figure out how to capture more nuance out of that relationship. This sort of side-steps the question you asked, because honestly we'd need to test a lot more to figure out the specific cases where an approach like this falls short.
Anecdotally - we've seen the greatest deviation between the estimation & integrated gradients as prompt "ambiguity" increases. We're thinking about ways to quantify & measure that ambiguity but that's its own can of worms.
