catalog.ngc.nvidia.com/orgs/nvidia/resources/resnet_50_v1_5_for_pytorch explains:
The difference between v1 and v1.5 is that, in the bottleneck blocks which requires downsampling, v1 has stride = 2 in the first 1x1 convolution, whereas v1.5 has stride = 2 in the 3x3 convolution.This difference makes ResNet50 v1.5 slightly more accurate (~0.5% top1) than v1, but comes with a small performance drawback (~5% imgs/sec).
How many logical qubits are needed to run Shor's algorithm? by 
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01 Integer factorization algorithms better than Shor's algorithm by Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01- 2023 www.schneier.com/blog/archives/2023/01/breaking-rsa-with-a-quantum-computer.html comments on "Factoring integers with sublinear resources on a superconducting quantum processor” arxiv.org/pdf/2212.12372.pdf
A group of Chinese researchers have just published a paper claiming that they can—although they have not yet done so—break 2048-bit RSA. This is something to take seriously. It might not be correct, but it’s not obviously wrong.We have long known from Shor’s algorithm that factoring with a quantum computer is easy. But it takes a big quantum computer, on the orders of millions of qbits, to factor anything resembling the key sizes we use today. What the researchers have done is combine classical lattice reduction factoring techniques with a quantum approximate optimization algorithm. This means that they only need a quantum computer with 372 qbits, which is well within what’s possible today. (The IBM Osprey is a 433-qbit quantum computer, for example. Others are on their way as well.)
@cirosantilli/_file/python/python/sphinx by Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01To run each example and see the output run:
./build.sh
xdg-open out/index.html @cirosantilli/_file/python/sphinx/python/sphinx/hello by Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01Minimal example. Gives a hint at how boilerplate heavy Sphinx can be!
@cirosantilli/_file/python/sphinx/python/sphinx/union by Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01 @cirosantilli/_file/python/sphinx/python/sphinx/class by Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01Basic class example.
@cirosantilli/_file/python/sphinx/python/sphinx/virtual_method by Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01
Ciro Santilli 35 Updated 2025-03-28 +Created 1970-01-01OK, there's a billion questions:
- SQL Server
- stackoverflow.com/questions/485409/generating-a-histogram-from-column-values-in-a-database OP did not know the difference between count and histogram :-) But it's the number one Google result.
- stackoverflow.com/questions/19103991/create-range-bins-from-sql-server-table-for-histograms has a minor extra group by twist, but otherwise fine
- stackoverflow.com/questions/16268441/generate-histogram-in-sql-server
- SQLite
- stackoverflow.com/questions/67514208/how-to-optimise-creating-histogram-bins-in-sqlite perf only, benchmarking would be needed. SQLite.
- stackoverflow.com/questions/32155449/create-a-histogram-with-a-dynamic-number-of-partitions-in-sqlite variable bin size, same number of entries per bin
- stackoverflow.com/questions/60348109/histogram-for-time-periods-using-sqlite-regular-buckets-1h-wide time
- MySQL: stackoverflow.com/questions/1764881/getting-data-for-histogram-plot MySQL appears to extend
ROUNDto also round by integers:ROUND(numeric_value, -2), but this is not widely portable which is a shame - stackoverflow.com/questions/72367652/populating-empty-bins-in-a-histogram-generated-using-sql specifically asks about empty bins, which is amazing. Amazon Redshift dialect unfortunately, but answer provided works widely, and Redshift was forked from PostgreSQL, so there's hope. Those newb open source server focused projects that don't use AGPL!
Let's try it on SQLite 3.40.1, Ubuntu 23.04. Data setup:
sqlite3 tmp.sqlite 'create table t(x integer)'
sqlite3 tmp.sqlite <<EOF
insert into t values (
0,
2,
2,
3,
5,
6,
6,
8,
9,
17,
)
EOF
sqlite3 tmp.sqlite 'create index tx on t(x)'For a bin size of 5 ignoring empty ranges we can:which produces the desired:
sqlite3 tmp.sqlite <<EOF
select floor(x/5)*5 as x,
count(*) as cnt
from t
group by 1
order by 1
EOF0|4
5|5
15|1And to consider empty ranges we can use SQL which outputs the desired:
genenerate_series + as per stackoverflow.com/questions/72367652/populating-empty-bins-in-a-histogram-generated-using-sql:sqlite3 tmp.sqlite <<EOF
select x, sum(cnt) from (
select floor(x/5)*5 as x,
count(*) as cnt
from t
group by 1
union
select *, 0 as cnt from generate_series(0, 15, 5)
)
group by x
EOF0|4
5|5
10|0
15|1SQL's implementation of database triggers.
This feature is really cool, as it allows you to keep caches up to date!
In particular, everything that happens in a trigger happens as if it were in a transaction. This way, you can do less explicit transactions when you use triggers. It is a bit like the advantages of SQL CASCADE.
DBMS:
TODO what is the standard compliant syntax?
PostgreSQL requires you to define a SQL stored procedure: stackoverflow.com/questions/28149494/is-it-possible-to-create-trigger-without-execute-procedure-in-postgresql Their syntax may be standard compliant, not sure about the
EXECUTE part. Their docs: www.postgresql.org/docs/current/sql-createtrigger.htmlSQLite does not support SQL stored procedures at all, so maybe that's why they can't be standard compliant here: stackoverflow.com/questions/3335162/creating-stored-procedure-in-sqlite
SQL:1999 11.38 covers "Trigger definition". The Abstract syntax tree starts with the
CREATE TRIGGER and ends in:<triggered SQL statement> ::=
<SQL procedure statement>This is defined at 13.5 "SQL procedure statement", but that is humongous and I'm not sure what it is at all.
Basic example tested on SQLite 3.40.1, Ubuntu 23.04:output:
sqlite3 :memory: 'select 1 union select 2'1
2Two columns two rows:output:
sqlite3 :memory: <<EOF
select * from (values (1, 2), (2, 3))
union
select * from (values (2, 3), (3, 4))
EOF1|2
2|3
3|4Note how duplicates are removed, to keep them we output:
UNION ALL instead:sqlite3 :memory: <<EOF
select * from (values (1, 2), (2, 3))
union all
select * from (values (2, 3), (3, 4))
EOF1|2
2|3
2|3
3|4 There are unlisted articles, also show them or only show them.