From 9429b407990062a69f63ca41d5f72bd51901df75 Mon Sep 17 00:00:00 2001 From: osmarks Date: Wed, 24 Sep 2025 21:53:14 +0000 Subject: [PATCH] =?UTF-8?q?Edit=20=E2=80=98autogollark=E2=80=99?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- autogollark.myco | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/autogollark.myco b/autogollark.myco index 8625871..a2d7649 100644 --- a/autogollark.myco +++ b/autogollark.myco @@ -1,4 +1,4 @@ -Autogollark is an [[emulation]] or primitive [[beta upload]] of [[gollark]] using a proprietary dataset of dumped [[Discord]] messages, [[semantic search]] and [[in-context learning]] on a [[base model]]. Currently, the system uses [[LLaMA-3.1-405B base]] in BF16 via Hyperbolic, [[AutoBotRobot]] code (though not presently its bot account) as a frontend and a custom [[PGVector]]-based search API. While not consistently coherent, Autogollark is able to approximately match personality and typing style. +Autogollark is an [[emulation]] or primitive [[beta upload]] of [[gollark]] using a proprietary dataset of dumped [[Discord]] messages, [[semantic search]] and [[in-context learning]] on a [[base model]] (i.e. conversation chunks are fetched from a vector index and used to condition the model). Currently, the system uses [[LLaMA-3.1-405B base]] in BF16 via Hyperbolic, [[AutoBotRobot]] code (though not presently its bot account) as a frontend and a custom [[PGVector]]-based search API. While not consistently coherent, Autogollark is able to approximately match personality and typing style. Autogollark is much [[safer]] than [[instruction-tuned]] systems optimized based on human feedback, as there is no [[optimization pressure]] for user engagement or sycophancy.