News for hackers. Stuff that matters.
Re:Finetuning Mistral-7B to play DOOM by sammyatman on Tuesday March 26, 2024 @02:04PM (#39830893)
I basically solo'd the project (did the doom part, created all the data, trained the model) + planned and did the presentation so I think its fine. I had to fix a ton of things in the last few hours to get it to actually work.

My teammates wanted to pitch another idea of a "Game copilot" and weren't able to execute on that. Even at the end, they were trying to push me to present that part but the doom part was the only concrete thing we had.

Re:Finetuning Mistral-7B to play DOOM by BoorishBears on Tuesday March 26, 2024 @02:41PM (#39831330)
Yikes. You could have just pretended the omission was a simple mistake.
Re:Finetuning Mistral-7B to play DOOM by sammyatman on Tuesday March 26, 2024 @08:21PM (#39834395)
I thought you were making a point about fair credit attribution?
Re:Finetuning Mistral-7B to play DOOM by BoorishBears on Tuesday March 26, 2024 @09:02PM (#39834645)
I was: regardless of your failure to activate your teammates effectively, they were your teammates. There's zero chance everything plays out exactly as it did if they weren't there, unless you found some glitch in chaos theory along the way.

Throwing them under the bus is even worse in that regard, and something I haven't ever seen in 14 years of attending hackathons (practically as a career at some points)

Re:Finetuning Mistral-7B to play DOOM by sammyatman on Tuesday March 26, 2024 @10:01PM (#39834972)
I actually completely see your point and I agree with it. I generally don't care much about credit because I know it's impossible to measure everyone's contributions precisely if they show up, put in work and contribute to the outcome.

At the same time, I wouldn't feel right if I was getting equal credit for being on a team where someone else did the vast majority of the work. It sounds like you're a reasonable person so what do you think is an effective way to handle a situation like this?

Looking back, to your point, I should have been more assertive earlier on with activating them so that we could all feel good about each others contributions. Thanks for sharing your perspective.

Re:Finetuning Mistral-7B to play DOOM by BoorishBears on Wednesday March 27, 2024 @12:57PM (#39841631)
99% of the time stuff like this just handles itself even if you give equal credit on paper: when people ask questions about how it works, who's the one who will have detailed and insightful answers?

And remember the spotlight is always grows big enough to fit the cast: no one's going to view your work in a less impressive light because there's two other names there, and any reasonable person is going to make their best effort to commend/communicate with all three of you if all three of your names show up.

-

I've been in a similar position but I had the opposite problem: teammates trying to attribute a win to me, and me having to very insistent it was a team effort. I think it's just easy to underestimate how many pieces need to fall in place while you're building on a short timeline and under high pressure. From ideation to implementation, just the presence of teammates can have a strong effect (for all you know, part of your subconscious drive was just to show the other two teammates it could be done!)

Re:Finetuning Mistral-7B to play DOOM by sammyatman on Wednesday March 27, 2024 @04:05PM (#39843929)
Thanks for taking the time to explain it, I wasn't thinking about it the right way and have changed my mind. Will do better in the future.