Andrea Panizza
@andreapanizza.bsky.social
ML, trekking, enjoying life
This is beautiful 🤩 I would have paid for a similar treatment from my team
April 10, 2025 at 7:49 PM
This is beautiful 🤩 I would have paid for a similar treatment from my team
PS I forgot an important caveat: the compute budget should be the same for both methods being HPO'ed. If the nihilist feels like being a PITA, I'll add a curve of the performance of the two methods as a function of the compute budget. If they still don't give up, I give up 🤷♂️ life goes on regardless.
April 1, 2025 at 12:23 PM
PS I forgot an important caveat: the compute budget should be the same for both methods being HPO'ed. If the nihilist feels like being a PITA, I'll add a curve of the performance of the two methods as a function of the compute budget. If they still don't give up, I give up 🤷♂️ life goes on regardless.
cross-validation precisely to answer this kind of objection. Is HPO is unfeasible, I agree that the result is just preliminary evidence. I note that since HPO is costly, there's merit in methods which show superior performance with default HP, but I don't insist b/c it's a valid objection. /2
April 1, 2025 at 12:17 PM
cross-validation precisely to answer this kind of objection. Is HPO is unfeasible, I agree that the result is just preliminary evidence. I note that since HPO is costly, there's merit in methods which show superior performance with default HP, but I don't insist b/c it's a valid objection. /2
I start by acknowledging the validity of their objection, because inn general it's true that with different HP, results may have been different (even random seeds may be considered HP, see arxiv.org/abs/2210.13393). Secondly, if we're in a situation where HPO is feasible, I use nested 1/
We need to talk about random seeds
Modern neural network libraries all take as a hyperparameter a random seed, typically used to determine the initial state of the model parameters. This opinion piece argues that there are some safe us...
arxiv.org
April 1, 2025 at 12:13 PM
I start by acknowledging the validity of their objection, because inn general it's true that with different HP, results may have been different (even random seeds may be considered HP, see arxiv.org/abs/2210.13393). Secondly, if we're in a situation where HPO is feasible, I use nested 1/
Ah, during the time of the big excitement about L5! I remember talking to people in Houston, expecting FSD to be solved soon...things proved to be harder, but a lot of progress has been made!
March 5, 2025 at 5:20 PM
Ah, during the time of the big excitement about L5! I remember talking to people in Houston, expecting FSD to be solved soon...things proved to be harder, but a lot of progress has been made!
Very well written! Did you work in the sector?
March 5, 2025 at 7:42 AM
Very well written! Did you work in the sector?
Looks like a book or a very long review paper! Can you share the link?
March 4, 2025 at 8:54 AM
Looks like a book or a very long review paper! Can you share the link?