I've seen it so this too. I had it keeping a running tally over many turns and occasionally it would say something like: "... bringing the total to 304.. 306, no 303. Haha, just kidding I know it's really 310." With the last number being the right one. I'm curious if it's an organic behavior or a taught one. It could be self learned through reinforcement learning, a way to correct itself since it doesn't have access to a backspace key.
The default outputs are considerably shorter even in thinking mode. Something that helped me get the thinking mode back to an acceptable state was to switch to the Nerd personality and in the traits customization setting tell it to be complete and add extra relevant details. With those additions it compares favorably to o3 on my recent chat history and even improved some cases. I prefer to scan a longer output than have the LLM guess what to omit. But I know many people have complained about verbosity so I can understand why they may have moved to less verbiage.
It's likely that the commenter has read less than 5 million posts worth of text though. So perhaps this still points to a lack of diversity in content.
You got me wondering. Supposing the average post is 10 words, and a typical page of text is 250 words, that would only be ~50 pages of text a day over the last 10 years. Which I don't think I manage, but over 20 years I am probably in that window.
I saw a very similar timely appeal here on Hacker News a few years ago and taught my son with this book at the age of 4. It has become my go-to comparison when prompting chat bots on what I want in a teaching material for other subjects. I listened to the entire article posted here and it makes me wonder if schools are getting something as foundational as reading wrong how can we trust the attention to research on anything else they're teaching? Don't get me wrong, I'm not going to pull my kid out of school but I'll dig a little deeper into how well he's learning. For math, we've been doing the Beast Academy books. It has gone... Okay. I like that they approach problems from many different ways which simulate the many different ways math is hidden in our interactions with the world. For my younger son I've recently started Teaching Your Child... because of how well it went for his brother but for math I may try something else to have a new data point. Something that occurred to me listening to the article is I wonder if certain skills are learned much faster with one on one instruction like the book has you do. Our schools pretty much never teach that way out of efficiency, though home schools often do. It may not be true for most subjects though or home school students would be so far ahead by college and that's not the impression I have.
"Here, we provide a quantifiable definition: A multimodal native model refers to a single model with strong understanding capabilities across multiple input modalities (e.g. text, code, image, video), that matches or exceeds the modality specialized models of similar capacities."
If you would build an in-memory cache, how would you do it?
It should have good performance and be able to hold many entries.
Reads are more common than writes. I know how I would do it already,
but I’m curious about your approach.
Was to add this requirement since it comes up so often:
Let's assume that keys accessed follow a power law, so some keys get
accessed very frequently and we would like them to have the fastest
retrieval of all.
I'm not sure if there are any efficient tweaks to hash tables or b-trees that might help with this additional requirement. Obviously we could make a hash table take way more space than needed to reduce collisions, but with a decent load factor is the answer to just swap frequently accessed keys to the beginning of their probe chain? How do we know it's frequently accessed? Count-Min sketch?
Even with that tweak, the hottest keys will still be scattered around memory. Wouldn't it be best if their entries could fit into fewer pages? So, maybe a much smaller "hot" table containing say the 1,000 most accessed keys. We still want a high load factor to maximize the use of cache pages so perhaps perfect hashing?
I've been doing this so long that my first thought was "use redis."
Why?
* it works
* it's available now
* it scales
* it's capable of HA
* it has bindings for every language you probably want to use
Why bother writing your own cache, unless it's for an exercise? Cache management is complicated and error prone. Unless the roundtrip kills you just use redis (or memcached).
In a typical LRU cache every read is a write in order to maintain access order. If this is a concurrent cache then those mutations would cause contention, as the skewed access distribution leads to serializing threads on atomic operations trying to maintain this ordering. The way concurrent caches work is by avoiding this work because popular items will be reordered more often, e.g. sample the requests into lossy ring buffers to replay those reorderings under a try-lock. This is what Java's Caffeine cache does for 940M reads/s using 16 thread (vs 2.3B/s for an unbounded map). At that point other system overhead, like network I/O, will dominate the profile so trying to rearrange the hash table to dynamically optimize the data layout for hot items seems unnecessary. As you suggest, one would probably be better served by using a SwissTable style approach to optimize the hash table data layout and instruction mix rather than muck with recency-aware structural adjustments.
The fastest retrieval will be a cache hit, so really once the data structures are not the bottleneck then the focus should switch to the hit rates. That's where the Count-Min sketch, hill climbing, etc. come into play in the Java case. There's also memoization to avoid cache stampedes, efficient expiration (e.g. timing wheels), async reloads, and so on that can become important. Or if a dedicated cache server like memcached, one has to worry about fragmentation, minimizing wasted space (to maximizing usable capacity), efficient I/O, etc. because every cache server can saturating the network these days so the goals shifts towards reducing the operational cost with stable tail latencies. What "good performance" means is actually on a spectrum because one should optimize for overall system performance rather than any individual, narrow metric.
You should check out the FASTER paper from Microsoft. It specifically covers how to create a K/V log that spills to disk for older keys, but keeps recent keys in memory.
Where I thought this might be going from the first paragraph:
Negative numbers are sometimes represented with parentheses: (234.58)
Tables sometimes tell you in the description that all numbers in are in 1000's or millions.
The dollar sign is used by many currencies, including in Australia and Canada.
I'd probably look around for some other gotchas. Here's one page on prices in general: https://gist.github.com/rgs/6509585 but interestingly doesn't quite cover the OP's problem or the ones I brought up, though the use cases are slightly different.
