From fb9d0576e508988d0ac76f8d2a91fde825bcd301 Mon Sep 17 00:00:00 2001 From: malo-adler <87966500+malo-adler@users.noreply.github.com> Date: Thu, 7 Nov 2024 16:16:14 +0100 Subject: [PATCH] Correction de coquille --- II-Developpements/1_Anatomie_LLM.qmd | 1 + 1 file changed, 1 insertion(+) diff --git a/II-Developpements/1_Anatomie_LLM.qmd b/II-Developpements/1_Anatomie_LLM.qmd index bbafab0..b5c5015 100644 --- a/II-Developpements/1_Anatomie_LLM.qmd +++ b/II-Developpements/1_Anatomie_LLM.qmd @@ -251,5 +251,6 @@ L'algorithme de DPO (Direct Preference Optimization) permet de mettre à jour le - [Guide pratique / Implémentation HugginFace](https://huggingface.co/blog/dpo-trl) Liens des papiers originaux : + - [DPO](https://arxiv.org/abs/2305.18290) - [KTO](https://arxiv.org/abs/2402.01306)