Cannabis Ruderalis

Top edits to an article All edits made to a page by one user, in chronological order.

Article Reinforcement learning from human feedback (Log · Page History)
User PopoDameron (Edit Counter· Top Edits)
Total edits 91
Minor edits 14 (15.4%)
(Semi-)automated edits 5 (5.5%)
Reverted edits 0 (0%)
atbe1 4.3
Added (bytes)2 35,350
Deleted (bytes) -2,613
Minor edits · 14 (15.4%)
Major edits · 77 (84.6%)
(Semi-)automated edits · 5 (5.5%)
Manual edits · 86 (94.5%)
Reverted edits · 0 (0%)
Unreverted edits · 91 (100%)
1 Average time between edits (days)
2 Added text is any positive addition that wasn't reverted (approximate)
Date Links Size Edit summary
Diff · History 1,055 fixing minor things and adding clarifications
Diff · History 156 clarification
Diff · History -81 not the first time the term was used
Diff · History 12 ziegler 19 also continues text
Diff · History 36 first rlhf
Diff · History 5 ce
Diff · History 3 ce
Diff · History 0 ce
Diff · History 426 background
Diff · History 10 ce
Diff · History 344 simplification and delineation
Diff · History -249 ce
Diff · History 1,080 added new source and discussion
Diff · History 3 ce
Diff · History -13 added limitations
Diff · History 136 added a bit of detail
Diff · History -148 remove weakly sourced statement and explain overfitting
Diff · History 1,354 improved training section
Diff · History 188 Applications: stability clarification
Diff · History 36 clarity
Diff · History 842 more on video game bots
Diff · History 642 about the amount of comparison data
Diff · History 140 delineation
Diff · History -1 ce
Diff · History 445 Collecting human feedback: layperson explanation
Diff · History 221 suggested clarifications
Diff · History 316 clarity
Diff · History 1,047 change to sourced example
Diff · History 79 improved lede
Diff · History 0 moved sources
Diff · History -11 ce
Diff · History 3 ce
Diff · History 0 ugly split infinitive
Diff · History 97 ce, links, and switched online & offline because the latter is more important/common
Diff · History 1 ce
Diff · History 2 ce
Diff · History 0 ce
Diff · History 2 ce
Diff · History 4 ce
Diff · History 771 added limitation summary to the lede
Diff · History 410 made lede more accessible and moved some things around
Diff · History 0 change figure placement again. makes more sense here for phones
Diff · History 30 added template
Diff · History 6 wrong cite templates
Diff · History 103 added overview diagram
Diff · History -19 remove template
Diff · History 897 wrapping up
Diff · History 1,711 started RL policy training. just need to add the second term and probably do some ce
Diff · History -16 in use
Diff · History 16 to be continued later
Diff · History -30 Training: ce
Diff · History -8 ce
Diff · History 1,013 finished reward model training. next: training the policy using the RM
Diff · History -16 in use
Diff · History 1,154 started training section. still incomplete and needs a lot more on the reward model, plus haven't started the actual policy training
Diff · History 421 added another good source
Diff · History 191 clarity and ce
Diff · History 751 explained online vs offline distinction
Diff · History -422 Undid revision 1212481439 by Aldopacchiano (talk) citation is not relevant + WP:SELFCITE
Diff · History 307 Applications: +claude
Diff · History -29 See also: already in the lede, +alphabetical
Diff · History 2,169 added CV applications. will add more contextual technical detail soon
Diff · History 333 added gemini to applications
Diff · History 320 fixing up nlp applications
Diff · History -4 Undid revision 1210766352 by Ibnu Fulan (talk) I don't see enough evidence that this would be notable enough for an article
Diff · History 771 mostly ce
Diff · History 2,121 improved motivation
Diff · History 122 improved alternatives section
Diff · History 280 Collecting human feedback: improving section based on paper
Diff · History 79 ce & sections
Diff · History 329 clarifying and improving lede
Diff · History 21 ml template param
Diff · History 21 add machine learning template
Diff · History -20 acronym is fine since it was already expanded above + capitalization
Diff · History -1,132 cleanup and removing some unsourced examples etc
Diff · History -324 This might be simple, but it is not technically accurate
Diff · History 244 I believe that this should make things accessible enough to a more casual reader, but feel free to readd the tag upon disagreement
Diff · History -61 Oh, I see
Diff · History 22 wikilink
Diff · History -29 I can't imagine how someone might misinterpret this sentence. How could it be any more direct? (assuming at least very basic RL knowledge, of course)
Diff · History 9 clarified reward model
Diff · History 56 other common name
Diff · History 2 Cleaned up using AutoEd
Diff · History 313 some elaboration on NLP difficulties
Diff · History 166 minor improvements
Diff · History 0
Diff · History 55 added see also section
Diff · History 104 +Category:Reinforcement learning; +Category:Language modeling; +Category:Artificial intelligence using HotCat
Diff · History 31 added Category:Machine learning using HotCat
Diff · History 49 Adding short description: "Machine learning technique"
Diff · History 11,267 started article on RLHF. missing more technical details, which I plan to work on soon.
All times are in UTC.

Leave a Reply