fareha

21,making my place in the world
Deep learning | DSA
|

[fareha @as_farrr]

You shouldn’t need a data center to understand your own videos.

Introducing gUrrt - a way to talk to your videos, running fully locally, without the soul-crushing VRAM requirements of large video-language models.

built by me and @muffBozo

https://t.co/gF0B8oKJxt

A thread 🧵

[fareha @as_farrr]

@deephivex The mathematical question section was too bad

[fareha @as_farrr]

@Oblivious9021 Claude will the part , he don't know opensource 😂😂

[fareha @as_farrr]

I can atleast do medium ( not really )

[fareha @as_farrr]

@Hikari_07_jp Lets connect , i am also doing local infrence engineering and stuff

[fareha @as_farrr]

@0xSero Currently reading this one :)

[fareha @as_farrr]

@analogalok i have RTX 4050 6GB , i was doing a highly intense batch image captioning task via llama.cpp that temp rise kinda scared me

[fareha @as_farrr]

@deephivex Yes

[fareha @as_farrr]

how to revise Dp , graph , whole ML in one day ??

[fareha @as_farrr]

@d3tx0nr 😭😭

[fareha @as_farrr]

@dharmeshrohit_ Only time dilation can save me now 😭😭

[fareha @as_farrr]

@Priyanshu__2109 I am jealous

[fareha @as_farrr]

Well I am waiting , not just for decentralization of power , but for more inference and ai infra job roles

[Ahmad @TheAhmadOsman]

Prediction

Opensource AI mass adoption within less than 18 months

[fareha @as_farrr]

@Utkarsh05367658 Long time not practised Dp , not fluent earlier also

[fareha @as_farrr]

@SashIs107 27 batch or 26 ?

[fareha @as_farrr]

@CodeLvM interview was related to AI engineering more accurately inference eng, They asked me some question related to my project , different inference engine when to use which one , paged and flash attention , vllm ,some agentic ai question

[fareha @as_farrr]

Given my first interview phone call
went just okay but at least i got a call 😭😭

[fareha @as_farrr]

@nikhil_thakur70 May 500 application and 100 breakdown later

[fareha @as_farrr]

@CodeLvM Not too good ,not too bad ,just okay okay

[fareha @as_farrr]

@TheAhmadOsman I was using ollama from past 7-8 month on my laptop 6gb vram and just switched to llama.cpp for my project (https://t.co/mAU42Tuq7K ) batch video captioning pipeline .
If any one want to give a read here is the link
https://t.co/8p7XSyyBFk