I am incentivized to try and resolve this as YES. I will be betting in this market as well. I've left a 30,000 Mana limit order at 30% as a baseline.
Related questions
fuh is banned. Is his real name known so this market can still be resolved?
@firstuserhere Care to share your scores? It would inform chances for a possible NeurIPS re-submission in June.
@AlexbGoode I'd like to bet much more here. But without an answer the information asymmetry between us is to large.
@AlexbGoode I recently left my job a month ago so that I could freely take out some time and explore a few different projects, before joining somewhere new. Currently, I am involved in 3 projects that I am excited about.
Two of them are with independent researchers, and in the preliminary stage. I also am pursuing a project with a research group, the lead of which has published before to ICML, but not NeurIPS (to my knowledge, haven't looked it up yet).
@firstuserhere So you are not affiliated with a research lab? Is there a publication record for the project with the independent researchers? AIstats, AAAI, ICLR or JMLR? Any other less competitive venues? I am asking to get an idea of how far to deviate from base rates.
Are you planning to submit to ICML in January? Otherwise there is only NeuRIPS left as an option anyway.
Are you first author in these projects?
I am not yet affiliated with a research lab. There is track record for independent researchers, for both neurips and icml, though most of them I know personally, and don't know the actual stats. Can't imagine them being very high. However, also having talked to several people over the last month at both OpenAI and Anthropic about my projects in the mechanistic interpretability of LLMs (and other models), i feel confident in doing good research.
unrelated to these, a personal project i'm doing is similar to toy models of superposition by anthropic but for non language models. Another personal one is about studying the training dynamics in a suite of language models.
Yes, there are plans for 2 of the projects to submit to the ICML.
Yes, also considering other venues (being vague rn on purpose, will not be on a later date), as well as stuff like BlackboxNLP
Yes, I am either the first or second author.
Eh, why not?
@firstuserhere Not sure if much relevant but i have written two theses (one for a master's degree) and have pursued independent research projects in college
@firstuserhere for what it's worth, you seem to be providing some decent reasons for thinking you are doing "good research," but in the scope of all good research, even specifically the scope of all good technical AI/ML research, only a tiny fraction is a good fit for NeurIPS and ICML.
Also note that competitiveness varies a lot across subfield and method (e.g., theory, NLP, optimization, benchmarks).