Tim Kellogg
LightNews LightNews
banner
kellogh.hachyderm.io.ap.brid.gy
Tim Kellogg
@kellogh.hachyderm.io.ap.brid.gy
25 followers 2 following 67 posts
I'm a software engineer and sometimes manager. Currently #Raleigh but also #Seattle. Building ML platform for a healthcare startup. Previously, built an IoT platform for one of "those" companies.

Open source: dura, fossil, Jump-Location, Moq.AutoMock, ot…
Posts Replies Media Videos
kellogh.hachyderm.io.ap.brid.gy
can't do **this** with a junior engineer
screenshot of me using Cursor with an opening prompt: "alright dingbat, let's do this. I want to try a new strategy wit..."
October 9, 2024 at 7:50 PM
kellogh.hachyderm.io.ap.brid.gy
yooo tyro looks hot. i've never seen it before (and that's why we read open source code)

A #python tool for CLI options based on type annotations

https://github.com/brentyi/tyro
import tyro def main( foo: int, bar: str = "default", ) -> None: ... # Main body of a script. if __name__ == "__main__": # Generate a CLI and call `main` with its two arguments: `foo` and `bar`. tyro.cli(main)
October 9, 2024 at 2:28 AM
kellogh.hachyderm.io.ap.brid.gy
per CNN, voters have a clearer idea of who these guys are now

#walz
Favorable 46% -> 59% (+14)
Unfavorable 22% -> 32% (+10)

#vance
Favorable 30% -> 41% (+11)
Unfavorable 44% -> 52% (+8)
Following the debate, 59% of debate watchers said they had a favorable view of Walz, with just 22% viewing him unfavorably – an improvement from his already positive numbers among the same voters pre-debate (46% favorable, 32% unfavorable). Debate watchers came away from the debate with roughly neutral views of Vance: 41% rated him favorably and 44% unfavorably. That’s also an improvement from their image of Vance pre-debate, when his ratings among this group were deeply underwater (30% favorable, 52% unfavorable).
October 2, 2024 at 2:33 PM
kellogh.hachyderm.io.ap.brid.gy
Honestly this seems like a miss by Anthropic — **shouldn’t** an AI company be able to figure out the caching part on their own?

i get that substring isn’t exactly AI, but Anthropic’s implementation seemed low-level. Like you need extra machinery to manage the […]

[Original post on hachyderm.io]
API calls to supported models will automatically benefit from Prompt Caching on prompts longer than 1,024 tokens. The API caches the longest prefix of a prompt that has been previously computed, starting at 1,024 tokens and increasing in 128-token increments. If you reuse prompts with common prefixes, we will automatically apply the Prompt Caching discount without requiring you to make any changes to your API integration.
October 1, 2024 at 9:56 PM