Good post, aligns with a lot of my (anecodtal) expreience in a related but different field (biostatistics, still doing computational work but not ML and much more mature as a field).
Under communication: I think your missing actively reading papers (or listening to presentations). Each time you read a paper, ask yourself if it was easy to understand the idea, why or why not? A big problem in reasearch writing IMO is that the reader often is not reading the paper for the main reason you wrote it. Perhaps they care more about your methodology than your results, so they can apply it in a different context. Or their most interested in some implications or secondary result, which you didn’t elucidate or explain as clearly. I think this part is hard and I don’t have good ideas beyond release data* (where possible / ethical) and intermediatory results (for example, your trained model*) to allow secondary analyses.
Caveat: ethics, privacy, open-sourcing model weights might be bad, legal, etc. mean this isn’t always net-positive. I guess it is for the vast majority of research.
Good post, aligns with a lot of my (anecodtal) expreience in a related but different field (biostatistics, still doing computational work but not ML and much more mature as a field).
Under communication: I think your missing actively reading papers (or listening to presentations). Each time you read a paper, ask yourself if it was easy to understand the idea, why or why not? A big problem in reasearch writing IMO is that the reader often is not reading the paper for the main reason you wrote it. Perhaps they care more about your methodology than your results, so they can apply it in a different context. Or their most interested in some implications or secondary result, which you didn’t elucidate or explain as clearly. I think this part is hard and I don’t have good ideas beyond release data* (where possible / ethical) and intermediatory results (for example, your trained model*) to allow secondary analyses.
Caveat: ethics, privacy, open-sourcing model weights might be bad, legal, etc. mean this isn’t always net-positive. I guess it is for the vast majority of research.