From d8fc0ecc80e6ec7fa269d08ea21bd28b35062fc4 Mon Sep 17 00:00:00 2001
From: osmarks <osmarks@mycorrhiza>
Date: Wed, 11 Dec 2024 10:30:37 +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 f89e17a..ff39603 100644
--- a/autogollark.myco
+++ b/autogollark.myco
@@ -11,7 +11,7 @@ Autogollark is much [[safer]] than [[instruction-tuned]] systems optimized based
 * Writeable memory?
 * {Fix lowercasing issue.
 * Due to general personality stability. Need finetune or similar.
-* One proposal: use internal finetune to steer big model somehow.
+* One proposal: use internal finetune to steer big model somehow. Possibly: use its likelihood (prefill-only) to evaluate goodness of big model output wrt. gollark personality, and if it is too bad then use finetune directly.
 }
 * {Increased autonomy (wrt. responses).
 * Use cheap classifier to evaluate when to respond.