Yepp i can

Parallelize as in? Multiple models running together?

Aye thats going to help me out alot

Also which subreddits would you recommend me for posting about what I am doing? I guess writing blogs on medium is one way (Also I started to maintain a log book for my on-going project, though still local) On LinkedIn, its more like just normal text posts which I don't think is possible to do on reddit.

I do post alot of random stuff on twitter and sometimes LinkedIn as well

I am considering to make content on what I am working on (atleast on streams I do show what Im doing to the few viewers I get)

Might consider posting on reddit too but not LinkedIn for explaining what I'm doing (since there are some real assholes in my university and I don't think I want them to know what I'm doing in work)

(Also would it be possible for to send some opportunities to apply to? My career path sure is going to be in the ML development industry and later on to academia)

Gah never had friends in the first place to allow them to leave (joking)

But yeah gotta tune it I guess loss went to nan and I just woke up ;-;

Im working on video recognition Just started the project though so no direction as of now

Hahahaha I use weights and biases for this It just tells me the error rates and accuracy in a graph and if the program is running or not

Other than that I just stay in the illusion that everything is fine

Sounds fair enough Current evaluation time is like 1.5 hours so I didn't go ahead with an hourly evaluation plan

This sounds like a good plan to do while I wait for the model to train.

I'll start from tomorrow (since its 10:30 pm and I feel like I have burnt myself out for the day fixing the errors) Hope no more errors pop up while I sleep

Academia for now
Since Im a student (bachelors) and no one wants someone with just a bachelor's so I can't really enter the industry properly even if I want to

I would have as well but currently I have no clue what to do
Right now I am training the model(efficient-video-recognition) just to see if its resource friendly to our servers or not

So no clarity in which direction I have to move.

Ohh that sounds fun
I'll try working on some full stack development as well now that I have free time

Holy moly, Guess what, I am doing this now
Going to participate in kaggle competitions more often (I have a fear of competitions so I never participated) and everytime I fuck up
I increase the number of crunches by 10

Yea, thanks for the advice :D (I was going to wait like an idiot) Fixed it now and seems like it is running properly

Hahaha yeah I do spot mistakes as soon as I start the run

That sounds like a good idea A set or maybe just read a paper while it trains

Edit: Also I'll be going back to my University campus(still a bachelor's student) this week so physical activity is going to go off the charts

Doing this for a model that has an ETA of 5 days sounds a bit too much for a workout ;-;

Well well guess what I just got a signal kill at evaluation hohoho (I am evaluating the model after every 1000 steps and its takes an approximate 5 hours to go through each 1000 steps) This was the first eval check so fuck

Oh wow thats actually pretty good
I'll read it and also start maintaining my own log book too

Since I am working on two research projects, This will be fun