Testing AI by AIπ₯
Following up on the Telebirr assistant project I'm building I added real evaluation using RAGAS instead of just reading outputs and judging them myself.
Why RAGAS: it scores four separate things:
1. is the answer actually grounded in what was retrieved (faithfulness)
2.does it actually answer the question (relevancy)
3.was what got retrieved actually useful (precision)
4.and did retrieval find the right info in the first place (recall). Each one points at a different part of the pipeline if something's off.
Latest run: faithfulness 0.79, relevancy 0.81, precision 0.95, recall 0.79.
numbers shift a bit run to run some judgment calls timeout or fail on the free tier, so it's averaging over whatever completes. Still useful signal, just not perfectly stable yet.
The project will continue on more updates π₯π₯
#BuildInPublic
Following up on the Telebirr assistant project I'm building I added real evaluation using RAGAS instead of just reading outputs and judging them myself.
Why RAGAS: it scores four separate things:
1. is the answer actually grounded in what was retrieved (faithfulness)
2.does it actually answer the question (relevancy)
3.was what got retrieved actually useful (precision)
4.and did retrieval find the right info in the first place (recall). Each one points at a different part of the pipeline if something's off.
Latest run: faithfulness 0.79, relevancy 0.81, precision 0.95, recall 0.79.
numbers shift a bit run to run some judgment calls timeout or fail on the free tier, so it's averaging over whatever completes. Still useful signal, just not perfectly stable yet.
The project will continue on more updates π₯π₯
#BuildInPublic
π₯2β€1β‘1
Forwarded from Birhan Nega
To get something you never had, you have to do something you never did.
β€2
Forwarded from Lid's Verse (Lidiya)
This media is not supported in your browser
VIEW IN TELEGRAM
Forwarded from mr jc
α₯αα³α ααͺ α¨ααα°α³ α³ααα΅ α α°α¨α³α½α ααα«α α°αα α΅
β€3π₯°2
This is totally me when I'm not prepared to speak in public π π
ππππππ
@atocodes
@atocodes
π5
Comparison drains us because we measure our messy behind the scenes struggles against everyone else's polished highlight reels. When you tie your worth strictly to immediate results, every setback in your life or work feels like a personal failure rather than a natural part of growth. Wins and losses are just temporary experiences along the way, so stop comparing your raw draft to someone else's final edit and focus on living your own story.
Said by Fevenβ€οΈβπ©Ήπ€π½
CheckOut the Youtube :https://youtu.be/oaspA498zUE?si=dw2ZhtIg8378v9wn
#SundayThoughts
β€βπ₯3
Forwarded from α₯α¨αα¬
α α₯α₯αα
αα₯α
α α α₯α
α₯ α¨αααα΅ ααα« α¨α₯αα± αααα’
αα½αα αα³α 4:23
αα½αα αα³α 4:23
Forwarded from Coolio
BRAND NEW VIDEO IS OUT
Make sure you watch the entire video and drop a β€οΈ
Enjoyβ€οΈ
https://youtu.be/xVxLymVMrn4?si=hYKFKS-M2NfJNAdq
Make sure you watch the entire video and drop a β€οΈ
Enjoyβ€οΈ
https://youtu.be/xVxLymVMrn4?si=hYKFKS-M2NfJNAdq
YouTube
αα΅α΄α α¦αααα α αααα΅ || I Finding My Soulmate on ometv - Live session 101
join the telegram channel to never miss a live session
@CoolioCodes
@CoolioCodes
Forwarded from Mr. SeeFun.Dev
I built a 90-day Backend Mastery roadmap for developers who want to go deeper into backend engineering.
It covers 17 systems across Linux, networking, Docker, PostgreSQL, Redis, system design, distributed systems, Kafka, Kubernetes, cloud, CI/CD, observability, security, and more.
It includes curated resources, hands-on labs, checklists, and a day-by-day plan.
If you're serious about leveling up your backend skills, check it out:
backend-mastery.seefun.dev
@sifendev
It covers 17 systems across Linux, networking, Docker, PostgreSQL, Redis, system design, distributed systems, Kafka, Kubernetes, cloud, CI/CD, observability, security, and more.
It includes curated resources, hands-on labs, checklists, and a day-by-day plan.
If you're serious about leveling up your backend skills, check it out:
backend-mastery.seefun.dev
@sifendev
π₯1
Forwarded from Micky Codes πΌ
α αΆ Codes πΌ
Wait what ? Is it foreal Github is not working ? π³ Let's Support Abel β€οΈ
Yeah github experiancing some sort of high error rates around 20% for web experiences and api traffic.
@MickyCodes
@MickyCodes
Forwarded from Abroid ΓΔv (Abraham Addisu)
π August winner: Lidia Kibret (@Binary_04)
Live portfolio. Real GitHub work. An honest, specific reflection on what she actually built this month.
That's exactly what this scholarship exists to reward.
Her ALX Data Science subscription starts this week π₯
To everyone who applied, thank you. I read every single one, and I emailed each of you personally with real feedback.
Next round opens next month.
@abroid0
Live portfolio. Real GitHub work. An honest, specific reflection on what she actually built this month.
That's exactly what this scholarship exists to reward.
Her ALX Data Science subscription starts this week π₯
To everyone who applied, thank you. I read every single one, and I emailed each of you personally with real feedback.
Next round opens next month.
@abroid0
β€βπ₯1π₯1
β€1π1π₯°1
Forwarded from Chapi Dev Talks
so honestly speaking if you crack one idea that can work locally i bet you will find some other that will work for the abroad.
we usually got attract by US startup environment and we try harder to crack or to find one solution to a problem that were never existed. and we got attracted by the number people are posting.
So i really recommend you to start looking your own environment and find something that will work for the local market, i know it is hard that why it is not for everybody then you will meet a lot of people that will let you see more problem to be solved.
in my exprience i have seen many problem that needs to be solved in ethiopia for ethiopian market, but there are few people solving them. ofc if i tell you the problem to solve it will kill your creativity and the problem i see even might not aligned with your interest but if you really tackle the hardship you see the local market you definetely will know what to solve for the international market as well.
Also i really want you to try to involve and engage in networking event, it is something we can use to scale up our game.
we usually got attract by US startup environment and we try harder to crack or to find one solution to a problem that were never existed. and we got attracted by the number people are posting.
So i really recommend you to start looking your own environment and find something that will work for the local market, i know it is hard that why it is not for everybody then you will meet a lot of people that will let you see more problem to be solved.
in my exprience i have seen many problem that needs to be solved in ethiopia for ethiopian market, but there are few people solving them. ofc if i tell you the problem to solve it will kill your creativity and the problem i see even might not aligned with your interest but if you really tackle the hardship you see the local market you definetely will know what to solve for the international market as well.
Also i really want you to try to involve and engage in networking event, it is something we can use to scale up our game.
β€2