Recently went through a lot of discussions regarding "onset of AI and the paradigm shift at work".
Conclusion drawn:
Pre AI fundamentals and Post AI speed is the moat one can/should have today.
@Sheetal2205 i've been through AI assisted coding/debugging rounds (model under strict guardrails to not spit the code rather ask for approach, logic and give syntax) last month
as far as in person ones are concerned GCCs - jargons and how would one implement AI
@ElitzaVasileva 1.2 mil coming in annually at 12% market appreciation would give me headroom to build, sell, travel and explore ppl, diversity, cultures thus attracting more money
so its a cycle u see!
i just rushed through building a ML pipeline for amazon ML challenge and 5.5 performed exceptionally well in reasoning around the blocking/candidate generation strategies taking my F_0.5 score from 0.61 to 0.73 at full run within 3 hrs (on a free tier colab - memory contraints)
@waiting4_asi@abuxtalha i just rushed through building a ML pipeline for amazon ML challenge and 5.5 performed exceptionally well in reasoning around the blocking/candidate generation strategies taking my F_0.5 score from 0.61 to 0.73 at full run within 3 hrs (on a free tier colab - memory contraints)
@arpit_bhayani Would solicit a detailed take by you on Jev on YT/ preferably a stream where we leverage Jev and its classification capabilities hands on to build something.
I have a task (completed by agents) verification engine in my mind currently. @arpit_bhayani
@moiiikaaa reasoning brotha reasoning , loyalty follows !
just came on X after 72 hrs locked in with a business entity resolution ML pipeline and Opus 5.5
@safishamsii@graphify except for design and filmmaking all boxes ✅️
rather happy and convinced to see the roles existing where i thrive best with my multipotentiality !
Techie (not nerd) + people's person (oration , my forte)
would love to add value!
<Amazon ML Challenge>
done with iteration 3 of the complete ML pipeline boosting F_0.5 score :
0.61 -> 0.65 -> 0.72
vectorisation played a key role from 2 to 3.
wrapping up 72 hrs of grind with tons of knowledge and skill-ups pertaining to ML pipelines.
@amazonIN@awscloud
<Amazon ML Challenge>
finaly after 54 hrs of sleuthing ....
done with the submission of matching.tsv output
12.7 GB RAM, vectorizations, blocking optimisations, etc.
onto some feature engineering, threshold optimizations for remaining time
@awscloud@amazonIN#MachineLearning
@mayuri_3015 60 dollars can get you 2 claude pro instances and 2 opencode go instances - never ever you will run out of time window
just manage context through md files.
compute is the bottleneck!
exhausting token limits mid-project feels as if we are pre 2016 and our 1 GB data pack got exhausted :)
Working out Blocking strategies that work at scale - SNB failed.
#AmazonMLChallenge#MLTraining@awscloud@amazon@amazonIN