From 84ac7d528830f48f10f0287c22d164ed005a14af Mon Sep 17 00:00:00 2001 From: Madjakul Date: Fri, 8 Nov 2024 17:53:43 +0100 Subject: [PATCH] Typo --- markdown/course5_inference.md | 2 +- markdown/course6_risks.md | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/markdown/course5_inference.md b/markdown/course5_inference.md index 168642b..a5e5aa9 100644 --- a/markdown/course5_inference.md +++ b/markdown/course5_inference.md @@ -199,7 +199,7 @@ How to reduce the distance between $q(x)$ and $p(x)$ when the assistance model i ### Layer skip: self speculative decoding -Speculative decoding comes with two inconvenients: +Speculative decoding comes with two inconveniences: - Loading two models in memory - Making sure the assistant model output a token distribution as close as possible to the main model diff --git a/markdown/course6_risks.md b/markdown/course6_risks.md index ffab2a5..26113a5 100644 --- a/markdown/course6_risks.md +++ b/markdown/course6_risks.md @@ -8,7 +8,7 @@ marp: true --- -# **Handling the Risks of Language Models** +# **Course 6: Handling the Risks of Language Models** ---