> 'A couple years ago, I read Beezus and Ramona, by Beverly Cleary. You didn't ask it to list ALL of the unusual things. And Santa, if you think there is room in your sleigh, please bring me an abacus. Please also bring me a set of paper dolls and a paint set, and Ramona some coloring books and crayons.

I think she has tried hard to be good, and I will help her comb it.

And please bring Ramona a doll with real hair to comb too. Please bring me a Shetland pony, a new dress, and a doll with real hair to comb. 'I have been a good girl all year except for my little sister Ramona who has been hard to put up with sometimes, but I guess I have too. > 'Dear Santa Claus,' Beezus printed carefully. > Certainly! Here's the actual passage from the book 'Beezus and Ramona' by Beverly Cleary, where Beezus writes a letter to Santa: > What's the actual passage from the real novel say? Shetland ponies were popular children's pets at the time, and buggy whips were used to control horse-drawn carriages, but the inclusion of an abacus suggests that the protagonist, Beezus, has an interest in learning and math. > The unusual thing about the list is that it includes an abacus, which is a counting device that was commonly used in ancient China and is still used in some parts of the world today, but is not a typical gift request for a young child to make to Santa Claus. In it, the protagonist writes a letter to Santa: 'Dear Santa, I would like a shetland pony, a buggy whip, and an abacus'. > A couple years ago, I read Beezus and Ramona, by Beverly Cleary. I feel like it's hard to predict one way or the other without trying it, just because LLMs have already demonstrated a lot of surprising powers. I'd be curious to see whether huge-context models are also able to do this, or if they start to have trouble when the bottleneck becomes reasoning capacity rather than input length. A human reading modified-Gatsby might eventually spot the altered line, but they'd also be able to answer questions about the overarching plot and themes of the novel, including ones that cannot be deduced from just a small number of salient snippets. I'm curious, though, whether there are diminishing returns in terms of how much analysis the model can do over those 100k tokens in a single forward pass. This sort of needle-in-a-haystack retrieval is definitely impressive, and it makes a lot more sense to achieve this in-context rather than trying to use a vector database if you can afford it. Carraway was “a software engineer that works on machine learning tooling at Anthropic.” When we asked the model to spot what was different, it responded with the correct answer in 22 seconds. >For example, we loaded the entire text of The Great Gatsby into Claude-Instant (72K tokens) and modified one line to say Mr.

Blume fudge series