Nimgwen's recent activity
-
Comment on The writers’ strike over AI is bigger than Hollywood in ~tv
-
Comment on Russia says Wagner Group’s leader will move to Belarus after his rebellious march challenged Putin in ~news
Nimgwen Link Parentyeah this has been my conclusion as well, because they were exposed traveling in column through the highways, the Kremlin knew where they were at all times. They could have been easily vaporized...yeah this has been my conclusion as well, because they were exposed traveling in column through the highways, the Kremlin knew where they were at all times. They could have been easily vaporized with rockets/artillery/jets, this was seemingly more akin to a protest than an actual civil war. Yet Prigozhin still had an advantage to negotiate, because it is unlikely the Kremlin would have wanted to obliterate a huge number of their soldiers, their infrastructure and all the civilians caught in the cross fire.... which might have ignited a real and longer revolution/insurgency.
So it was a stand-off and someone caved behind closed doors, so they achieved this "deal", which is not much of a deal, just a veiled way from one side to lose, which side lost? not quite clear yet, we will see now if Prigozhin gets assassinated (or his forces get arrested or sent to die in the front) or if Putin's power or his MOD's power diminishes. I think this is far from over.
-
Comment on Russia's FSB announces a felony investigation into Wagner Group's incitement to 'armed rebellion' in ~news
Nimgwen Link Parentyeah this a serious blow to the russian war effort, nothing good can come out of this for them.yeah this a serious blow to the russian war effort, nothing good can come out of this for them.
-
Comment on Prevalence of AI generated text in Tildes in ~tildes
Nimgwen Link ParentRough comment draft I wrote quickly without any editing and filled with mistakes and weird phrasing: From my experience GPT-4 is just as good if not better than Grammarly, mainly because it also...Rough comment draft I wrote quickly without any editing and filled with mistakes and weird phrasing:
From my experience GPT-4 is just as good if not better than Grammarly, mainly because it also does smarter corrections by rephrasing some sentences. This way it finds sentences than flow better or are more concise than what I had originally had given it. In that way it is less conservative in it's editing, but may end up being more useful by being able to polish rougher draft faster. Here is an article showing a comparison:
ChatGPT vs Grammarly: Who Wins the Battle of Grammatical Error Correction?
GPT-4 output (prompt was simply to "correct the text and make it flow better":
Based on my experience, GPT-4 can be just as effective, if not more so, than Grammarly. The key difference lies in its intelligent correction methods, which include rephrasing certain sentences to enhance their flow or make them more concise than the original input. In this regard, GPT-4 is less conservative in its editing approach, potentially providing more value by swiftly refining rough drafts. For a more detailed comparison, you may wish to consult the following article:
ChatGPT vs Grammarly: Who Wins the Battle of Grammatical Error Correction?
-
Comment on I gave Lemmy, Kbin, and Beehaw a chance. I think I'll be sticking with Tildes. in ~tech
Nimgwen Link ParentThe beta for Pulse an iOS/Android app for browsing Squabbles has recently released on both platforms, the UI is similar to Apollo, you might enjoy that. Although it is currently read-only, but...The beta for Pulse an iOS/Android app for browsing Squabbles has recently released on both platforms, the UI is similar to Apollo, you might enjoy that. Although it is currently read-only, but soon it will allow posting/commenting/liking, things are evolving very rapidly with Squabbles main-site and the 3rd party apps, it is quite exciting.
-
Comment on Some general advice for all the new Tildes users in ~tildes
Nimgwen Link ParentAs a new user who just joined, I find it incredibly refreshing to hear this. I have always strived for the ability to engage in reasonable arguments on the internet. From my early days on forums...As a new user who just joined, I find it incredibly refreshing to hear this. I have always strived for the ability to engage in reasonable arguments on the internet. From my early days on forums and through platforms like 4chan, YouTube comments, Facebook, and Reddit, I have made efforts to engage in nuanced discussions, albeit with varying results depending on the specific platform and community.
It's great to know that this platform as a whole shares the same goal, as I've often been frustrated by the culture on other platforms. Even on smaller forums (which were generally better than 4chan or YouTube comments, to be honest), I witnessed discussions turning into fights due to needless personal attacks overshadowing the actual points being made. While Reddit provided a somewhat better space for discussion, it was still heavily influenced by the culture of individual subreddits, which often made meaningful discussions impossible.
I'm looking forward to a different experience on Tildes.
the training process of the model is actually the analogue to previous experiences in human learning:
“A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.”
Humans can't create anything meaningful without extensive learning from previous experiences. Infants playing with crayons just make a garbled mess, with some hints at patterns. Interestingly AI is the same when the training is too shallow.