Embed Notice

HTML Code

<blockquote style="position: relative; padding-left: 55px;"><section><a href="https://toot.cafe/users/matt/statuses/111585593149227864">Matt Campbell (matt@toot.cafe)'s status on Saturday, 16-Dec-2023 06:16:22 JST</a><a href="https://toot.cafe/@matt" title="matt@toot.cafe"><img src="https://gnusocial.jp/theme/gnusocialjp/default-avatar-stream.png" width="48" height="48" alt="Matt Campbell" style="position: absolute; left: 0; top: 0;">Matt Campbell</a><div><a href="https://social.coop/@luis_in_brief/111585526194479974" rel="in-reply-to">in reply to</a><ul><li><li><a href="https://gnusocial.jp/user/40873" title="dalias@hachyderm.io">Rich Felker</a></li><li><a href="https://gnusocial.jp/user/55402" title="maria@thelife.boats">Maria Bustillos</a></li><li><a href="https://gnusocial.jp/user/75487" title="luis_in_brief@social.coop">Luis Villa</a></li><li><a href="https://gnusocial.jp/user/153529" title="danilo@hachyderm.io">Danilo Campos 🇵🇷</a></li></ul></div></section><article><p><a href="https://social.coop/@luis_in_brief">@luis_in_brief</a> <a href="https://hachyderm.io/@dalias">@dalias</a> <a href="https://fedi.simonwillison.net/@simon">@simon</a> <a href="https://hachyderm.io/@danilo">@danilo</a> <a href="https://thelife.boats/@maria">@maria</a> It would be interesting to see an LLM tuned for instruction-following, particularly iterative instruction following taking prior context into account, that is specifically trained not to emulate a person, e.g. no first-person pronouns.</p></article><footer><a rel="bookmark" href="https://gnusocial.jp/conversation/2458459#notice-4863931">In conversation</a><time datetime="2023-12-16T06:16:22+09:00" title="Saturday, 16-Dec-2023 06:16:22 JST">about a year ago</time> <span>from <span><a href="https://toot.cafe/@matt/111585593149227864" rel="external" title="Sent from toot.cafe via ActivityPub">toot.cafe</a></span></span><a href="https://toot.cafe/@matt/111585593149227864">permalink</a></footer></blockquote>

Corresponding Notice

Embed this notice
Matt Campbell (matt@toot.cafe)'s status on Saturday, 16-Dec-2023 06:16:22 JSTMatt Campbell
in reply to
@luis_in_brief @dalias @simon @danilo @maria It would be interesting to see an LLM tuned for instruction-following, particularly iterative instruction following taking prior context into account, that is specifically trained not to emulate a person, e.g. no first-person pronouns.
In conversationabout a year ago from toot.cafepermalink

Public

Embed Notice

HTML Code

Corresponding Notice