I’m happy that this shows that hard work, understanding your codebase, having performant software, having actually working software, rigorously measuring and proving proof of results still matters.
There’s a huge difference between using LLMs to offload any hard work and for LLMs to be of some assistance while you are in control and take ownership of the output.
Unfortunately, the general public probably didn’t try a git clone and cargo build, and took the article at face value.
If you want it's power as a query engine but like to write python instead of SQL, I highly recommend using it as a backend for the Ibis dataframe library
It let's you interchange pythonic dataframe syntax (like Pandas and Polars) with SQL that 'compile' down to SQL in DuckDB dialect
And you can use those queries interchangably in postgres, sqlite, polars, spark, etc
Any business that needs convincing to move on from anything labeled NTLM does not care what "nerds" have to say. They are either one of those "I'm not spending money on something that works" or stuck with such legacy technical debt that at this point, removing it from environment is too costly to even consider so executives kick it down the road.
Except that I think it's clear that the motive was getting the data not taking the site offline. The evidence for that is that it stopped on its own without them doing anything to mitigate it.
Also I don't know why you think this is sophisticated, it's probably 40 lines of Python code max.
Are the folks still using ES simply unaware of the performance advantages of ClickHouse, or is there some use case that ES covers that CH is still missing?
That's swell if you have a balcony facing the sun and enough space for 4m2 of solar without blocking your neighbors' sun. In my unit for example, it's completely infeasible.
Z-Image is another open-weight image-generation model by Alibaba [1]. Z-Image Turbo was released around the same time as (non-Klein) FLUX.2 and received generally warmer community response [2] since Z-image Turbo was faster, also high-quality, and reportedly better at generating NSFW material. The full-size Z-Image is not yet released.
It's a shame that this 6K tunderbolt hub monitor does not support the latest thunderbolt 5 standard. Otherwise you can connect and daisy chain two of these 6K displays together.
How do you fit 10 A4 pages into 4:3 aspect ratio. 10 is 2 by 5 or 5 by 2 or you want to fit 9 or 12.
My arithmetic nodule is having a konniption fit. Does not compute. If this is 16:9 and you mistook your aspect ratio I can breathe again. 1:√2 says "not 4:3 or 3:4"
There seems to be grey deny button at top-right on first view but it disappears if you select the details. You need hide the details first if you want to click it.
You should do your own evals specific to your case. In my evals XML outperforms JSON on every model for out of distribution tasks (i.e. not for JSON that was in the data).
I use them both depending on which feels more natural for the task, often within the same project. The interop is easy and very high performance thanks to Apache Arrow: `df = duckdb.sql(sql).pl()` and `result = duckdb.sql("SELECT * FROM df")`.
I realize there's a lot out there on the subject, but do you have any specific reading recommendations summarizing the effect of the US dollar's reserve currency status on middle America? It's a history I've been interested in learning.
It is pure game theory. An aggressive person expects no bad outcomes from his passive victim. If they get a signal that their own outcome may be not that good, even marginally, this very often changes their behaviour.
That's why the advice to act submissively presented as "avoiding confrontation" is often the wrong advice.
You are not seeking confrontation, but you should signal that you are ready for confrontation. Stops aggressive behaviour very often.
Pre-alignment-meeting alignment! A good thing to explicitly exclude.
Definitely need a healthy culture when asking for honesty and openness about what people think, including they see as unclear, unknown, inconsistent with others, and/or outside the existing box.
This is par for the course with this AI slop. Most of the big claims about LLM productivity have completely lacked any backing evidence. Big claims require big evidence, but all I've seen so far is loud assertions and pathetic results.
My name is Omkar Kovvali and I've been wanting to share my CC sessions with friends / save + access them easily,so I decided to make an MCP server to do so!
/share -> Get a link
/import -> resume a conversation in your Claude Code
All shared sessions are automatically sanitized, removing api keys, tokens, and secrets.
Give it a try following the Github README instructions - would love feedback!
There’s a huge difference between using LLMs to offload any hard work and for LLMs to be of some assistance while you are in control and take ownership of the output.
Unfortunately, the general public probably didn’t try a git clone and cargo build, and took the article at face value.