philipodonnell

7 ms·

31.
by philipodonnell 3y ago
The most common one is that leadership wants to make a change but some people won't like it so it's easier to have a consulting firm work out the details, make the recommendation and then leave. Even when the outcome is predetermi
32.
by philipodonnell 3y ago
They’re waves. Going up the front it’ll change the world and going down the back it’s doom and gloom. Everyone is shouting all those things all the time, it’s just the balance shifting that we call a wave. But, as long every crest and troug
33.
by philipodonnell 3y ago
> Text to SQL that actually works. Ask questions in natural language across multiple tables and our engine knows where to join. Can you share any information about how it knows where to join?
34.
by philipodonnell 3y ago
Any similar GitHub for Python you recommend?
35.
by philipodonnell 3y ago
This is called Text2SQL or NL2SQL, it’s a surprisingly difficult problem even with RAG and GPT4 as soon as the query is non trivial, especially if there are semantic differences between the question and the db schema.
36.
by philipodonnell 3y ago
I tried something similar to this so I assume it’s to classify a table as being part of the resulting query or not, and using the schemas of the resulting tables in the prompt for the question. I found it just as useful to give it the full
37.
by philipodonnell 3y ago
Did you provide any db schema information in the prompts after the model was fine tuned?
38.
by philipodonnell 3y ago
Agree that it’s not representative of real world queries, but I think it’s more of a “measure progress against a simple but consistent baseline” and SOTA is still struggling so it’s clearly not ready for real queries. Most of the papers on
39.
by philipodonnell 3y ago
I don’t think it’s necessarily about a “universal” solution, just “better”. Make the column names more verbose, changing numeric enums to text ones, disambiguating column names, etc. One of the spider datasets is a stadium table and one of
40.
by philipodonnell 3y ago
Are you planning on submitting this model to be evaluated against the Spider holdout set? Also, wondering if anyone has found research on the inverse of this approach to the problem, i.e., instead of training the model to understand the dat
41.
by philipodonnell 3y ago
The forestry allegory is so relevant to what we go through in enterprise data. I recommend this book all the time to understand the motivations behind it.
42.
by philipodonnell 3y ago
> Taxi shifts the obligation of describing how things stitch together from Consumers to Producers. Traditionally, it falls to consumers to work out how to do this. And, it’s an expensive question to answer… it involves tracking down API
43.
by philipodonnell 4y ago
"Epiphanic rush" is such a accurate but bittersweet term for this. It must be so ingrained in human psychology from the days of shamans, prophets and sages.
44.
by philipodonnell 5y ago
If I remember the discussion at the time, OpenDoor and Offerpad are pure-play iBuying while Zillow was an already huge business pivoting into iBuying to make up for declines in its advertising business. Zillow had a reason to scale faster t
45.
by philipodonnell 5y ago
Many baseline versions of SAAS software will use username/passwords but in order to get SSO you need the premium version. Is it fair to accuse that company of profiting from cybersecurity threats because username/password is less
46.
by philipodonnell 5y ago
There is hope! > Two trucks were used for building "The People Eater Limousine". One for closeup shots and the second for stunts and the final explosion scene.
47.
by philipodonnell 5y ago
Anyone have experience with Apache Kylin? Seemed to me as a way to blend the cube concept with columnar stores. https://en.m.wikipedia.org/wiki/Apache_Kylin
48.
by philipodonnell 5y ago
Great writeup.
49.
by philipodonnell 6y ago
Its a fair point, just a bit of a tough spot to put IT in, you know? If I buy a tool for my employees to use I want to know if they are using it to see if I am getting my money's worth. How would I do that if any attempt to measure usa
50.
by philipodonnell 6y ago
I get the knee-jerk reaction to look for scripts that fake activity and call this spying, but despite the name on the page, these kinds of metrics seem more focused on adoption of "productivity tools" (industry jargon for the suit
51.
by philipodonnell 6y ago
No it was legit experimentation to affect emotional states, not AB testing. Just a random google search: https://www.forbes.com/sites/kashmirhill/2014/06/28/facebook...
52.
by philipodonnell 6y ago
> Is Apple actually taking $10 from the developer and giving it to the customer? That is exactly what is happening.
53.
by philipodonnell 6y ago
wow
54.
by philipodonnell 6y ago
Any update to the Azure list anyone has seen? 2017 is a long time ago.
55.
by philipodonnell 6y ago
Its at the bottom?
56.
by philipodonnell 6y ago
Insider trading By elected officials isn’t against the law. There’s no reason for them to hide it. There have been several recent examples of senators openly insider training and no repercussions for being caught.
57.
by philipodonnell 7y ago
Even if that's not the best options trader in the world, its still a much better explanation than the post being replied to.
58.
by philipodonnell 7y ago
For us it backs up the sewers and makes things unsanitary, for them it ruined their crops and destroyed their buildings. It was arguably worse for them, but different problems with the same cause.
59.
by philipodonnell 7y ago
Maybe they found something better than fossil fuels and used it all up. Or were smarter about the impact to the climate would be if they burned it.
60.
by philipodonnell 7y ago
I think the author uses these to imply a dramatic pause... like I sometimes do with ellipses. :-)

More ›