All of abergal's Comments + Replies

Thanks for writing this! Would "fine-tune on some downstream task and measure the accuracy on that task before and after fine-tuning" count as measuring misalignment as you're imagining it? My sense is that there might be a bunch of existing work like that.

1William Saunders2y
I don't think all work of that form would measure misalignment, but some work of that form might, here's a description of some stuff in that space that would count as measuring misalignment. Let A be some task (e.g. add 1 digit numbers), B be a task that is downstream of A (to do B, you need to be able to do A, e.g. add 3 digit numbers), M is the original model, M1 is the model after finetuning. If the training on a downstream task was minimal, so we think it's revealing what the model knew before finetuning rather than adding knew knowledge, then better performance of M1 than M on A would demonstrated misalignment (don't have a precise definition of what would make finetuning minimal in this way, would be good to have a clearer criteria for that). If M1 does better on B after finetuning in a way that implicitly demonstrates better knowledge of A, but does not do better on A when asked to do it explicitly, that would demonstrate that the finetuned M1 is misaligned (I think we might expect some version of this to happen by default though, since M1 might overfit to only doing tasks of type B. Maybe if you have a training procedure where M1 generally doesn't get worse at any tasks then I might hope that it would get better on A and be disappointed if it doesn't).

This RFP is an experiment for us, and we don't yet know if we'll be doing more of them in the future. I think we'd be open to including research directions we think that are promising that apply equally well to both DL and non-DL systems-- I'd be interested in hearing any particular suggestions you have.

(We'd also be happy to fund particular proposals in the research directions we've already listed that apply to both DL and non-DL systems, though we will be evaluating them on how well they address the DL-focused challenges we've presented.)

1Ryan Carey3y
I imagine you could catch useful work with i) models of AI safety, or ii) analysis of failure modes, or something, though I'm obviously biased here.

Getting feedback in the next week would be ideal; September 15th will probably be too late.

Different request for proposals!

Thank you so much for writing this! I've been confused about this terminology for a while and I really like your reframing.

An additional terminological point that I think it would be good to solidify is what people mean when they refer to "inner alignment" failures. As you alude to, my impression is that some people use it to refer to objective robustness failures, broadly, whereas others (e.g. Evan) use it to refer to failures that involve mesa optimization. There is then additional confusion around whether we should think "inner alignment" failures that ... (read more)

Planned summary for the Alignment Newsletter:

This post describes the author’s insights from extrapolating the performance of GPT on the benchmarks presented in the <@GPT-3 paper@>(@Language Models are Few-Shot Learners@). The author compares cross-entropy loss (which measures how good a model is at predicting the next token) with benchmark performance normalized to the difference between random performance and the maximum possible performance. Since <@previous work@>(@Scaling Laws for Neural Language Models@) has shown that cross-entropy loss s

... (read more)

AI Impacts now has a 2020 review page so it's easier to tell what we've done this year-- this should be more complete / representative than the posts listed above. (I appreciate how annoying the continuously updating wiki model is.)

Thanks, added.

From Part 4 of the report:

Nonetheless, this cursory examination makes me believe that it’s fairly unlikely that my current estimates are off by several orders of magnitude. If the amount of computation required to train a transformative model were (say) ~10 OOM larger than my estimates, that would imply that current ML models should be nowhere near the abilities of even small insects such as fruit flies (whose brains are 100 times smaller than bee brains). On the other hand, if the amount of computation required to train a transformative model were
... (read more)
1Ajeya Cotra3y
Yes, it's assuming the scaling behavior follows the probability distributions laid out in Part 2, and then asking whether conditional on that the model size requirements could be off by a large amount. 

So exciting that this is finally out!!!

I haven't gotten a chance to play with the models yet, but thought it might be worth noting the ways I would change the inputs (though I haven't thought about it very carefully):

  • I think I have a lot more uncertainty about neural net inference FLOP/s vs. brain FLOP/s, especially given that the brain is significantly more interconnected than the average 2020 neural net-- probably closer to 3 - 5 OOM standard deviation.
  • I think I also have a bunch of uncertainty about algorithmic efficiency progress-- I could im
... (read more)
1Ajeya Cotra3y
Thanks! I definitely agree that the proper modeling technique would involve introducing uncertainty on algorithmic progress, and that this uncertainty would be pretty wide; this is one of the most important few directions of future research (the others being better understanding effective horizon length and better narrowing model size). In terms of uncertainty in model size, I personally find it somewhat easier to think about what the final spread should be in the training FLOP requirements distribution, since there's a fair amount of arbitrariness in how the uncertainty is apportioned between model size and scaling behavior. There's also semantic uncertainty about what it means to "condition on the hypothesis that X is the best anchor." If we're living in the world of "brain FLOP/s anchor + normal scaling behavior", then assigning a lot of weight to really small model sizes would wind up "in the territory" of the Lifetime Anchor hypothesis, and assigning a lot of weight to really large model sizes would wind up "in the territory" of the Evolution Anchor hypothesis, or go beyond the Evolution Anchor hypothesis.  I was roughly aiming for +- 5 OOM uncertainty in training FLOP requirements on top of the anchor distribution, and then apportioned uncertainty between model size and scaling behavior based on which one seemed more uncertain.

I'm a bit confused about this as a piece of evidence-- naively, it seems to me like not carrying the 1 would be a mistake that you would make if you had memorized the pattern for single-digit arithmetic and were just repeating it across the number. I'm not sure if this counts as "memorizing a table" or not.

1Daniel Kokotajlo4y
Excellent point! Well, they do get the answer right some of the time... it would be interesting to see how often they "remember" to carry the one vs. how often they "forget." It looks like the biggest model got basically 100% correct on 2-digit addition, so it seems that they mostly "remember."