Application Development Discussions
Join the discussions or start your own on all things application development, including tools and APIs, programming models, and keeping your skills sharp.
cancel
Showing results for 
Search instead for 
Did you mean: 

Submissions for "SAP HANA Cloud multi-model" developer challenge: Week 3 - Image Embeddings

Vitaliy-R
Developer Advocate
Developer Advocate
43 REPLIES 43

Vitaliy-R
Developer Advocate
Developer Advocate
0 Kudos
3,243

Thank you for being the first @Cocquerel 

Will you try any of the additional challenges from the bottom of that notebook?

Cocquerel
Active Contributor
3,238

Thanks @Vitaliy-R , I liked your typos.
Yes, I'm planning to do the additional challenges but I have also one for you if you accept it.
For week2, I have suggested an alternative method for the analogy search that provides good results in some use cases (see https://community.sap.com/t5/application-development-discussions/submissions-for-quot-sap-hana-cloud... ) I have faced the issue that HANA Vector engine does not support SUM operation yet and decided to do it on python side. Would you be able to do it on HANA side using SQL Script ? The SQL functions TO_ARRAY, UNNEST and CARDINALITY may be useful...waiting for your submission 😁

Vitaliy-R
Developer Advocate
Developer Advocate
0 Kudos
3,233

Yes, I saw your post, but I am on the road right now and could not spend more time looking at it, I apologize.

Cocquerel
Active Contributor
0 Kudos
3,222

No issue, you have one full week to post your submission 🤣

Cocquerel
Active Contributor

Cocquerel
Active Contributor
1,621

I saw that in the last HANA Cloud release, CARDINALITY and MEMBER_AT Functions now support Vector data type. There is maybe no more need to convert to Array.

Cocquerel_0-1719470455704.png

 

Vitaliy-R
Developer Advocate
Developer Advocate
0 Kudos
1,615

It was too short a time between the release of Q2 HANA Cloud and week 4 of the dev challenge to prepare something based on that release 😭

Cocquerel
Active Contributor
0 Kudos
1,595

I also saw that Q2 HANA Cloud is also now supporting Vector in Graph. I would have been curious about which use cases you would have chosen to illustrate that. Example, use COMMUNITIES_LOUVAIN Built-In Graph Algorithm to create communities of similar pets...

mvinci
Product and Topic Expert
Product and Topic Expert
3,116

Here is my submission - I don't want to spoil too much I will send you a separate email to make a list of those errors really good for learnings ... 🙂 

 

mvinci_0-1718883101925.png

 

mvinci
Product and Topic Expert
Product and Topic Expert
2,289

Hi Vitaliy, i thought that I would cut out the rest of the team if I'd send you an email for the errors. So here you have my comments for everyone to see and to comment.

The whole idea to inject errors is excellent. I like it.

Some of them are very useful to understand what is missing such as a wrong data type name and the need to transform the string into the key data type we need to learn in these series. 

Some of them as environment related (raw versus python) and they were too evident although not immediate to resolve

Finally the last group of errors and funny (INSANE) and clearly misspelled to give time to recover easily with smile.

Having said that during the exercise I have to confess that I thought you went from zero to infinite and maybe having just a few errors very well designed as the TO_REAL_VECTOR() would have been more effective. 

In general I am learning a lot so I thank you very much. 

SaiNithesh_Gajula
Active Participant
3,082

Hi,

Here is my Week 3 Submissions. This weeks task made me so interesting.

sainithesh21_0-1718891738110.png

 

MioYasutake
Active Contributor
3,029

My submission for week 3. My challenge was to find the row number of the image I liked.

MioYasutake_1-1718916678898.png

raulruiz
Explorer
2,986

Insane similarity? LOL!! I loved it 

By the way, my responses. More similar:

raulruiz_0-1718919256651.png

Less similar

raulruiz_1-1718919760590.png

raulruiz_2-1718919776324.png

I have no pet for the last topic so I tried with an image from me just for fun. Well, the result.....  I guess I deserved it, I knew something like this could happen 🙂 🙂

raulruiz_4-1718921203736.png

 

raulruiz_3-1718921125554.png

 

geek
Participant
2,824

There must be a happy middle ground between just pressing buttons in a workbook and this much head scratching:

geek_0-1718989715493.png

 

Cocquerel
Active Contributor
2,389

From my perspective, I think your idea of introducing errors was brilliant. It indeed requires effort to understand and analyze the code, but that’s precisely the intended goal. For some errors, I wondered whether they were part of the challenge or due to my environment. Having a list of expected errors would have been less confusing. I also appreciated the open-ended questions at the end. 😊

2,038

@Vitaliy-R Think that I take back the earlier comment, was trying to solve the bugs from first principles and struggling but if one refers back to the previous week, the answers were almost all there. 

Alpesa1990
Participant
2,409

My Submission for this week.

More Simlar

Alpesa1990_0-1719058989581.png

Less Similar

Alpesa1990_0-1719092385509.png

and my dog which I always thought was a dog.

Alpesa1990_1-1719093204892.png

 

Congrats for the challenge!

gphadnis2000
Participant
2,225

Hi Vitaliy,

Indeed it was tough checking errors but also make you learn new things.I will try more different combinations but my trial account is about to expire.Here is my week3 submission.

gphadnis2000_0-1719147016438.png

 

DirkO
Participant
2,190

Hello Hello Witalij,

pls find my submission for week 3.

DirkO_0-1719155489266.png

Seems to be a popular combination here, right?

Kind regards,

DirkO

 

M-K
Participant
2,137

Here's my submission. The first combination was already good 😊

2024-06-23 19_51_19-Window.png

Good idea with those coding errors, fixing them has had taught me much more than last week's challenge.

moh_ali_square
Participant
1,999

My Submission for this week. I stumbled with one typo, with overthinking, it took many hours. but finally I landed here.

Week3_submission.png 

1,998

I hope it was still more of a learning experience besides those many hours...

jpenninkhof
Product and Topic Expert
Product and Topic Expert
1,955

Lola is actually a British shorthair. Ah well, nobody's perfect, right? 😜

jpenninkhof_0-1719264170064.png

Enjoyed the bug-hunting! 😁

 

0 Kudos
1,892

So, what were the next entries on the list of similarities for Lola, and with what similarity scores?

SushmitaG
Product and Topic Expert
Product and Topic Expert
1,842

Absolutely loved the challenge and all thanks to you...We are getting a good hands-on exercises.Screenshot 2024-06-26 at 9.17.04 AM.png

Vitaliy-R
Developer Advocate
Developer Advocate
0 Kudos
1,710

Hmmm... but the results you are getting are quite off 🤔