ChatGPT is sufficiently well known to warrant critical scrutiny, and for this project we wrote a paper, developed a website where we track open-source instruction-tuned large language models , designed a poster for presentation at the ACM conference on Conversational User Interfaces (CUI’23) and, yes, even designed a logo that combines a key image of the open source movement with a variation on ChatGPT’s corporate logo.
2473932
UMBCUI8F
items
1
0
default
asc
1
2246
https://markdingemanse.net/wp-content/plugins/zotpress/
%7B%22status%22%3A%22success%22%2C%22updateneeded%22%3Afalse%2C%22instance%22%3A%22zotpress-53e6cf0f6c7c5c04933c449f3dc4a133%22%2C%22meta%22%3A%7B%22request_last%22%3A0%2C%22request_next%22%3A0%2C%22used_cache%22%3Atrue%7D%2C%22data%22%3A%5B%7B%22key%22%3A%22UMBCUI8F%22%2C%22library%22%3A%7B%22id%22%3A2473932%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Liesenfeld%20et%20al.%22%2C%22parsedDate%22%3A%222023%22%2C%22numChildren%22%3A3%7D%2C%22bib%22%3A%22%3Cdiv%20class%3D%5C%22csl-bib-body%5C%22%20style%3D%5C%22line-height%3A%202%3B%20padding-left%3A%201em%3B%20text-indent%3A-1em%3B%5C%22%3E%5Cn%20%20%3Cdiv%20class%3D%5C%22csl-entry%5C%22%3ELiesenfeld%2C%20A.%2C%20Lopez%2C%20A.%2C%20%26amp%3B%20Dingemanse%2C%20M.%20%282023%29.%20Opening%20up%20ChatGPT%3A%20tracking%20openness%2C%20transparency%2C%20and%20accountability%20in%20instruction-tuned%20text%20generators.%20%3Ci%3EACM%20Conference%20on%20Conversational%20User%20Interfaces%20%28CUI%20%26%23x2019%3B23%29%2C%20July%2019-21%2C%20Eindhoven%3C%5C%2Fi%3E.%20doi%3A%20%3Ca%20class%3D%27zp-doi-link%27%20href%3D%27https%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1145%5C%2F3571884.3604316%27%3E10.1145%5C%2F3571884.3604316%3C%5C%2Fa%3E%20%3Ca%20title%3D%27Download%27%20class%3D%27zp-DownloadURL%27%20href%3D%27https%3A%5C%2F%5C%2Fpure.mpg.de%5C%2Fpubman%5C%2Fitem%5C%2Fitem_3526897_1%5C%2Fcomponent%5C%2Ffile_3526898%5C%2FLiesenfeld%2520et%2520al_2023_Opening%2520up%2520ChatGPT.pdf%27%3EPDF%3C%5C%2Fa%3E%3C%5C%2Fdiv%3E%5Cn%3C%5C%2Fdiv%3E%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Opening%20up%20ChatGPT%3A%20tracking%20openness%2C%20transparency%2C%20and%20accountability%20in%20instruction-tuned%20text%20generators%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andreas%22%2C%22lastName%22%3A%22Liesenfeld%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alianda%22%2C%22lastName%22%3A%22Lopez%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mark%22%2C%22lastName%22%3A%22Dingemanse%22%7D%5D%2C%22abstractNote%22%3A%22Large%20language%20models%20that%20exhibit%20instruction-following%20behaviour%20represent%20one%20of%20the%20biggest%20recent%20upheavals%20in%20conversational%20interfaces%2C%20a%20trend%20in%20large%20part%20fuelled%20by%20the%20release%20of%20OpenAI%27s%20ChatGPT%2C%20a%20proprietary%20large%20language%20model%20for%20text%20generation%20fine-tuned%20through%20reinforcement%20learning%20from%20human%20feedback%20%28LLM%2BRLHF%29.%20We%20review%20the%20risks%20of%20relying%20on%20proprietary%20software%20and%20survey%20the%20first%20crop%20of%20open-source%20projects%20of%20comparable%20architecture%20and%20functionality.%20The%20main%20contribution%20of%20this%20paper%20is%20to%20show%20that%20openness%20is%20differentiated%2C%20and%20to%20offer%20scientific%20documentation%20of%20degrees%20of%20openness%20in%20this%20fast-moving%20field.%20We%20evaluate%20projects%20in%20terms%20of%20openness%20of%20code%2C%20training%20data%2C%20model%20weights%2C%20RLHF%20data%2C%20licensing%2C%20scientific%20documentation%2C%20and%20access%20methods.%20We%20find%20that%20while%20there%20is%20a%20fast-growing%20list%20of%20projects%20billing%20themselves%20as%20%60open%20source%27%2C%20many%20inherit%20undocumented%20data%20of%20dubious%20legality%2C%20few%20share%20the%20all-important%20instruction-tuning%20%28a%20key%20site%20where%20human%20annotation%20labour%20is%20involved%29%2C%20and%20careful%20scientific%20documentation%20is%20exceedingly%20rare.%20Degrees%20of%20openness%20are%20relevant%20to%20fairness%20and%20accountability%20at%20all%20points%2C%20from%20data%20collection%20and%20curation%20to%20model%20architecture%2C%20and%20from%20training%20and%20fine-tuning%20to%20release%20and%20deployment.%22%2C%22date%22%3A%222023%22%2C%22proceedingsTitle%22%3A%22ACM%20Conference%20on%20Conversational%20User%20Interfaces%20%28CUI%20%2723%29%2C%20July%2019-21%2C%20Eindhoven%22%2C%22conferenceName%22%3A%22%22%2C%22language%22%3A%22%22%2C%22DOI%22%3A%2210.1145%5C%2F3571884.3604316%22%2C%22ISBN%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fopening-up-chatgpt.github.io%22%2C%22collections%22%3A%5B%5D%2C%22dateModified%22%3A%222023-07-24T19%3A18%3A43Z%22%7D%7D%5D%7D
Liesenfeld, A., Lopez, A., & Dingemanse, M. (2023). Opening up ChatGPT: tracking openness, transparency, and accountability in instruction-tuned text generators.
ACM Conference on Conversational User Interfaces (CUI ’23), July 19-21, Eindhoven . doi:
10.1145/3571884.3604316 PDF
