jmount

16 ms·

211.
by jmount 2y ago
You are right. We simplified the Arancini al Burru: Rice Balls with Butter on our own. Also "Frittata di riso" is a very good match (relaxing the regional requirement).
212.
by jmount 2y ago
Thanks for the book recommendation! I found an English translation of a book by the author and I am ordering that.
213.
by jmount 2y ago
I guess I should have emphasized how I didn't like the LLM contradicting itself.
214.
Four Causes(en.wikipedia.org)
1 points by jmount 2y ago|0 comments
215.
by jmount 2y ago
(author) Good point, sorry about that. I guess I was think the generation is low value was a bit orthogonal to retrieval being very high value.
216.
I want flexible queries, not RAG(win-vector.com)
237 points by jmount 2y ago|153 comments
217.
by jmount 2y ago
Syntactically, no. But functionally Polars is new in the space.
218.
by jmount 2y ago
There are several. I am tinkering with one called data algebra: https://pypi.org/project/data-algebra/ .
219.
by jmount 2y ago
STAN is an industrial grade Monte Carlo realization.
220.
Solving Recurrence Relations(win-vector.com)
64 points by jmount 2y ago|11 comments
221.
by jmount 2y ago
After inventing multi-token, one then invents a useful language oriented hierarchy (such as sections, paragraphs, sentences, and words).
222.
by jmount 2y ago
Futures are tricky. If not designed right they degenerate bucket shops (side bets that have nothing to do with the underlying- literally just betting on what numbers will appear on a tape) or depend only on interest rate (instead of capturi
223.
The AI Problem is not Alignment(win-vector.com)
2 points by jmount 2y ago|0 comments
224.
by jmount 2y ago
Strongly agree. No UI is really intuitive enough that there isn't a formal or informal learning and even training cost to it. Then the redesign invariably finds a new way to hide the controls (hamburger menus, ... menus, gosh knows wha
225.
The m = n Anomaly (video introduction)(youtube.com)
1 points by jmount 3y ago|0 comments
226.
by jmount 3y ago
Likely paying some tall dollars to deal with cheap first line support. Probably a "direct to human" fee available soon.
227.
by jmount 3y ago
Thanks, as always for the usable link. I really have to say how much one must be missing the plot to place an opinion piece behind a paywall. I know this is a convergence of pre-existing things (opinion pieces and paywalls), but the combo i
228.
by jmount 3y ago
Good culture probably also has good strategy.
229.
The m = n Machine Learning Anomaly(win-vector.com)
1 points by jmount 3y ago|0 comments
230.
How Data Quantity Drives Model Quality(github.com)
16 points by jmount 3y ago|0 comments
231.
by jmount 3y ago
> There is no franchise value in our industry and the work and product innovation we drive at this moment will define the next decade and beyond. So no bird in hand, but the two in the bush are worth more?
232.
Exact Sums over a Group Action(win-vector.com)
1 points by jmount 3y ago|0 comments
233.
Illustrating the F-Test in Action(win-vector.com)
1 points by jmount 3y ago|0 comments
234.
More Stupid LLM Tricks(ninazumel.com)
11 points by jmount 3y ago|2 comments
235.
What Good Is Analysis of Variance?(win-vector.com)
1 points by jmount 3y ago|0 comments
236.
Clarity, Not Magic(ninazumel.com)
1 points by jmount 3y ago|0 comments
237.
by jmount 3y ago
It is getting to the point that I feel we need a rule of the form: your UI is rejected unless you can tell a new user how to succeed with it over the phone. To my mind this means at least: fewer modalities, no hidden menus (hamburger), and,
238.
How to Use Jupyter Notebooks Inside For-Loops(win-vector.com)
3 points by jmount 3y ago|0 comments
239.
by jmount 3y ago
The "inside out" rocket engine is a great explanation.
240.
Disposing of People Appears to Be Priced Cheaper Than Parting with Cash(win-vector.com)
1 points by jmount 3y ago|0 comments

More ›