@antirez@mitsuhiko For what it’s worth, google claims they use synthetic data generation from teacher to train the student models.
The “hurr durr, chinese models are doing only that” is obviously false and in the end meaningless.
If its that simple, do it and give us cheaper SOTA
Exercising the vocal dynamics and shifting of Indian classical music within the octave (saptak) and seven notes (swara).
[📹 musicriyaaz]
https://t.co/TKOsWGOBGT
Amazon e-link:
https://t.co/p0dCVB5Ykx
E book link:
https://t.co/G0lkDURszm
Publishers : @bookleafpub . I am thankful for the guidance of Jaya, my consultant, in this process.
@b50 Late October 2023, my bank RM called to ask if I want to apply for a personal loan
I paid for transunion CIBIL subscription to be able to see what's going on saw 10 enquiries including Axis Bank.
I raised dispute and they refused all of them
So might not have changed anything
Here's a new post walking through an implementation of MVCC and major SQL transaction isolation levels, in 400 lines of Go code.
These ideas might sound esoteric, but they impact almost every developer using any database.
https://t.co/crFKM74R5h
@Stavros@iavins For John Scalzi, search for Amber Benson and you'll find ones from the Lock In series.
For Andy Weir's Martian, there's one narrated by RC Bray. Audible doesn't have rights to it anymore. But last I knew, there were still physical CDs available.
@iavins@eatonphil I was asking "why do you keep the entire page" with a guess on why.
Fixed size could make math and lookup easier?
Eg: log sequence number 100 is at 99*(4KB+fixed header size)?
Other potential benefit being you can point to the committed WAL page before checkpointing?
@eatonphil I'm sure I must have butchered that as memory is a bit spotty.
The CMU intro to DB course has pretty good lectures from Andy and Jignesh on the topic.
https://t.co/3x5TWK7Bxg
Check out the MVCC and database logging lectures.
@eatonphil You don't need to keep all historical versions.
You could:
Pipe the deltas to X-ish GB redo log AOFs on disk
Have a page manager with a buffer pool that keeps the modified pages in-mem
Checkpointing mechanism to flush the in-mem structure to disk and rotate rid of the redo log