Join devRant
Do all the things like
++ or -- rants, post your own rants, comment on others' rants and build your customized dev avatar
Sign Up
Pipeless API

From the creators of devRant, Pipeless lets you power real-time personalized recommendations and activity feeds using a simple API
Learn More
Search - "rl"
-
Ask questions during interview.
Ask about trainings - it's usually a good sign when company offers training budget. Ask about specifics - sometimes it's a shared pluralsight account, and nothing else, which means that that had an idea and half assed it into existence.
Ask tech recruiter about overtime, a good sign is when they have no idea or say that it must be budgeted and scheduled - it means that it does not happen often.
Ask if it is possible to select and change projects, and how often it happens - if often, it may be bad low level management, or people learning new things and jumping between projects.
Also make sure to ask about rules for promotions and pay rises. Good company wił have a clear set of rules in place.
All of the above apply to mid to large companies.
For small company, i'm sure it will be different.3 -
Started a new job last week. Pays a tad below average for position, but i get training time and budget for anything i want.
So far i have had few days of company introduction, and now a week for training courses related to position.
I have not seen any code yet, brought no value in, just joined.
Massive green flag to me. -
Next level reinforcement learning:
Grab a baseball bat and show that damn machine who's the boss, i.e. reinforce that message by highfiving the said machine in the face with the aforementioned bat.3 -
Reinforcement learning is going to be my end. 😩😩😩☠️
(currently stuck at how to put images as well as a bunch of other -motor- values as input... and exactly what am I getting as output again?)
Pulling my own hair out... Ooooooof7 -
Not leaning on unit tests. I usually write them after testing my code manually, and lose time for testing feature by hand.1
-
Company i work for just posted updated pay rates.
It's good. They went from lower end of spectrum to a little above the average in terms of pay, and benefits stay the same - solid. -
The following paper combines recurrent neural nets for vision with methods from reinforcement learning research:
https://proceedings.neurips.cc/pape...
Apparently an agent learned to catch a ball 85% of the time, without being explicitly told to track the ball. The RL algorithm rewarded the agent *only* for successfully catching the ball. The system itself used this reward signal to set its *own* policy/goal, which was used to guide it toward the goal of tracking the ball itself--all on its own.
Behold, the very infancy of the paperclip maximizer problem.3 -
I spent ~12h working on a simple issue/bug.
7h was spent on rebuilding local dev environment which is a clusterfuck of maven profiles, tomcat, some autogenerated degeneracy, and 2 different build systems for JS.
5h spent on actual bug fixing, code reviews and so on.
FML2 -
Send an email.
Or, more seriously: invite only people who must be there, and can add something to the discussion, have an agenda, stick to it, and plan the meeting so that it ends at the start of lunch break. That way everybody will be interested in finishing on time or earlier. -
Get to know the new company better (Changed job shortly before Christmas).
Learn some DPs, DDD, k8s, finish introduction to hacking course, start doing htb and thm machines, finish and defend my thesis, finish books clean code, thinking in java (reading it to fill in gaps on knowledge), a few books about pentesting.
Among non tech goals: pass drivers license exam for cars, another one for motorcycles, go back to learning russian. -
Got pissed that my story delayed the whole project by 3 weeks (major story, lots of changes, and management decided to put me in a few extra activities outside of the project).
Stayed up until like 23:00 to deliver PR.1 -
Ros melodic in a strictly python 2.7 environment mixes horribly with a PyTorch based RL module... Time to work around with terminal calls from the latter
*sigh*1 -
Update of previous post
I just got assigned to a project.
Project-specific training is designed to take up to 20 days.
I got materials, videos, and a person who is in this project for a long time to supervise training and introduce me to the project.
Only problem is that the project apparently uses 5 years old version of base framework. -
!rant, but satisfying.
Got pulled in to a demo for some work that I had been apart of previously, but not recently since I had rolled on to a client. The Manager in charge of the work had fought me being pulled on to said client, as he wanted me committed to this project (which I didn't want to be invovled with). I had rolled off the engagement earlier this week, which is why I suspect I was included in this demo. So we are going through the motions, they are asking questions, I'm sitting quietly watching. out of the blue, Manager dude decides he wants to ask me a difficult question, because I'm sure he assumes he will stump me. I respond with "Not sure I'm the best to answer that specific question, since I haven't worked on this in a minute". He confirms that he only wants me to answer. So I do. And boy was I glad his camera was on, because he went from "Got em" to "Fuck, he got me" in a matter of seconds, and I could barely keep from smiling. After my answer, I respond with "Anything else you'd like to know?" to which he mutters "No, thank you" and quickly moves on. Talk about a victory. I'll ride this high through next week, I think.1