torchvision ResNet by Ciro Santilli 35 Updated +Created
pytorch.org/vision/0.13/models.html has a minimal runnable example adapted to python/pytorch/resnet_demo.py.
That example uses a ResNet pre-trained on the COCO dataset to do some inference, tested on Ubuntu 22.10:
cd python/pytorch
wget -O resnet_demo_in.jpg https://upload.wikimedia.org/wikipedia/commons/thumb/6/60/Rooster_portrait2.jpg/400px-Rooster_portrait2.jpg
./resnet_demo.py resnet_demo_in.jpg resnet_demo_out.jpg
This first downloads the model, which is currently 167 MB.
We know it is COCO because of the docs: pytorch.org/vision/0.13/models/generated/torchvision.models.detection.fasterrcnn_resnet50_fpn_v2.html which explains that
FasterRCNN_ResNet50_FPN_V2_Weights.DEFAULT
is an alias for:
FasterRCNN_ResNet50_FPN_V2_Weights.COCO_V1
The runtime is relatively slow on P51, about 4.7s.
After it finishes, the program prints the recognized classes:
['bird', 'banana']
so we get the expected bird, but also the more intriguing banana.
By looking at the output image with bounding boxes, we understand where the banana came from!
Figure 1.
python/pytorch/resnet_demo_in.jpg
. Source.
Figure 2.
python/pytorch/resnet_demo_out.jpg
. The beak was of course a banana, not a beak!
Ubuntu 23.04 by Ciro Santilli 35 Updated +Created
Ubuntu 22.10 by Ciro Santilli 35 Updated +Created
The Oxford Student by Ciro Santilli 35 Updated +Created
Cherwell (newspaper) by Ciro Santilli 35 Updated +Created
Fish by Ciro Santilli 35 Updated +Created
This paraphyletic subgroup is easy to form the "acquatic only" (fishes) vs "things that come out of water" (tetrapods). Though mudfish make that distinction harder.
Which kind of makes sense, why would you want for limbs unless you are going to stay out of water!
ROW_NUMBER by Ciro Santilli 35 Updated +Created
sqlite3 ':memory:'  'WITH t (i) AS (VALUES (-1), (-1), (-2)) SELECT *, row_number() over () FROM t'
Possible output:
-1|1
-1|2
-2|3
Gives them unique IDs.
With a partition by:
sqlite3 ':memory:'  'WITH t (i) AS (VALUES (-1), (-1), (-2)) SELECT *, row_number() over ( partition by i ) FROM t'
possible output:
-2|1
-1|1
-1|2
SQL window RANGE by Ciro Santilli 35 Updated +Created
rm -f tmp.sqlite
sqlite3 tmp.sqlite "create table t (id integer, val integer)"
sqlite3 tmp.sqlite <<EOF
insert into t values
  (0, 0),
  (1, 5),
  (2, 10),
  (3, 14),
  (4, 15),
  (5, 16),
  (6, 20),
  (7, 25),
  (8, 29),
  (9, 30),
  (10, 30),
  (11, 31),
  (12, 35),
  (13, 40)
EOF
Show how many neighbours each column has with val between val - 2 and val + 2 inclusive:
sqlite3 tmp.sqlite <<EOF
SELECT id, val, COUNT(*) OVER (
  ORDER BY val RANGE BETWEEN 2 PRECEDING AND 2 FOLLOWING
) FROM t;
EOF
Output:
0|0|1
1|5|1
2|10|1
3|14|3
4|15|3
5|16|3
6|20|1
7|25|1
8|29|4
9|30|4
10|30|4
11|31|4
12|35|1
13|40|1
val - 1 and val + 1 inclusive instead:
sqlite3 tmp.sqlite <<EOF
SELECT id, val, COUNT(*) OVER (
  ORDER BY val RANGE BETWEEN 1 PRECEDING AND 1 FOLLOWING
) FROM t;
EOF
Output:
0|0|1
1|5|1
2|10|1
3|14|2
4|15|3
5|16|2
6|20|1
7|25|1
8|29|3
9|30|4
10|30|4
11|31|3
12|35|1
13|40|1
There seems to be no analogue to HAVING for window functions, so we can just settle for a subquery for once, e.g.:
sqlite3 tmp.sqlite <<EOF
SELECT * FROM (
  SELECT id, val, COUNT(*) OVER (
    ORDER BY val RANGE BETWEEN 1 PRECEDING AND 1 FOLLOWING
  ) as c FROM t
) WHERE c > 2
EOF
which outputs:
4|15|3
8|29|3
9|30|4
10|30|4
11|31|3
SQL contiguous ranges by Ciro Santilli 35 Updated +Created
stackoverflow.com/questions/17046204/how-to-find-the-boundaries-of-groups-of-contiguous-sequential-numbers/17046749#17046749 just works, even in SQLite which supports all quoting types known to man including [] for compatibility with insane RDBMSs!
Here's a slightly saner version:
rm -f tmp.sqlite
sqlite3 tmp.sqlite "create table mytable (id integer primary key autoincrement, number integer, status integer)"
sqlite3 tmp.sqlite <<EOF
insert into mytable(number, status) values
  (100,0),
  (101,0),
  (102,0),
  (103,0),
  (104,1),
  (105,1),
  (106,0),
  (107,0),
  (1014,0),
  (1015,0),
  (1016,1),
  (1017,0)
EOF
sqlite3 tmp.sqlite <<EOF
SELECT
  MIN(id) AS "id",
  MIN(number) AS "from",
  MAX(number) AS "to"
FROM (
  SELECT ROW_NUMBER() OVER (ORDER BY number) - number AS grp, id, number
  FROM mytable
  WHERE status = 0
)
GROUP BY grp
ORDER BY MIN(number)
EOF
output:
1|100|103
7|106|107
9|1014|1015
12|1017|1017
To get only groups of length greater than 1:
sqlite3 tmp.sqlite <<EOF
SELECT "id", "from", "to", "to" - "from" + 1 as "len" FROM (
  SELECT
    MIN("id") AS "id",
    MIN(number) AS "from",
    MAX(number) AS "to"
  FROM (
    SELECT ROW_NUMBER() OVER (ORDER BY "number") - "number" AS "grp", "id", "number"
    FROM "mytable"
    WHERE "status" = 0
  )
  GROUP BY "grp"
  ORDER BY MIN("number")
) WHERE "len" > 1
EOF
Output:
1|100|103|4
7|106|107|2
9|1014|1015|2
Cat qubit by Ciro Santilli 35 Updated +Created
Open source AlphaGo implementation by Ciro Santilli 35 Updated +Created
AlphaGo Zero by Ciro Santilli 35 Updated +Created
Figure 1.
AlphaGo Zero cheat sheet by David Foster (2017)
Source.
2013 DNS Census virtual host cleanup heuristic keyword searches by Ciro Santilli 35 Updated +Created
There are two keywords that are killers: "news" and "world" and their translations or closely related words. Everything else is hard. So a good start is:
grep -e news -e noticias -e nouvelles -e world -e global
iran + football:
  • iranfootballsource.com: the third hit for this area after the two given by Reuters! Epic.
3 easy hits with "noticias" (news in Portuguese or Spanish"), uncovering two brand new ip ranges:
  • 66.45.179.205 noticiasporjanua.com
  • 66.237.236.247 comunidaddenoticias.com
  • 204.176.38.143 noticiassofisticadas.com
Let's see some French "nouvelles/actualites" for those tumultuous Maghrebis:
  • 216.97.231.56 nouvelles-d-aujourdhuis.com
news + world:
  • 210.80.75.55 philippinenewsonline.net
news + global:
  • 204.176.39.115 globalprovincesnews.com
  • 212.209.74.105 globalbaseballnews.com
  • 212.209.79.40: hydradraco.com
OK, I've decided to do a complete Wayback Machine CDX scanning of news... Searching for .JAR or https.*cgi-bin.*\.cgi are killers, particularly the .jar hits, here's what came out:
  • 62.22.60.49 telecom-headlines.com
  • 62.22.61.206 worldnewsnetworking.com
  • 64.16.204.55 holein1news.com
  • 66.104.169.184 bcenews.com
  • 69.84.156.90 stickshiftnews.com
  • 74.116.72.236 techtopnews.com
  • 74.254.12.168 non-stop-news.net
  • 193.203.49.212 inews-today.com
  • 199.85.212.118 just-kidding-news.com
  • 207.210.250.132 aeronet-news.com
  • 212.4.18.129 sightseeingnews.com
  • 212.209.90.84 thenewseditor.com
  • 216.105.98.152 modernarabicnews.com
Wayback Machine CDX scanning of "world":
  • 66.104.173.186 myworldlymusic.com
"headline": only 140 matches in 2013-dns-census-a-novirt.csv and 3 hits out of 269 hits. Full inspection without CDX led to no new hits.
"today": only 3.5k matches in 2013-dns-census-a-novirt.csv and 12 hits out of 269 hits, TODO how many on those on 2013-dns-census-a-novirt? No new hits.
"world", "global", "international", and spanish/portuguese/French versions like "mondo", "mundo", "mondi": 15k matches in 2013-dns-census-a-novirt.csv. No new hits.
iraniangoals.com JavaScript reverse engineering by Ciro Santilli 35 Updated +Created
Notably, the password is hardcoded and its hash is stored in the JavaScript itself. The result is then submitted back via a POST request to /cgi-bin/goal.cgi.
TODO: how is the SHA calculated? Appears to be manual.
CTE insert values by Ciro Santilli 35 Updated +Created
sqlite3 :memory: 'WITH t (i, j) AS (VALUES (1, -1), (2, -2)) SELECT * FROM t'

Unlisted articles are being shown, click here to show only listed articles.