From 4b0769d13721b7288a0bb0412e0107196a9cd862 Mon Sep 17 00:00:00 2001 From: Lewis Tunstall Date: Thu, 9 Nov 2023 14:42:57 +0000 Subject: [PATCH] Fix links --- recipes/zephyr-7b/README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/recipes/zephyr-7b/README.md b/recipes/zephyr-7b/README.md index 02746a1..4e783ad 100644 --- a/recipes/zephyr-7b/README.md +++ b/recipes/zephyr-7b/README.md @@ -3,8 +3,8 @@ As described in the Zephyr [technical report](https://huggingface.co/papers/2310.16944), training this model proceeds in two steps: -1. Apply SFT to fine-tune Mistral 7B on the UltraChat dataset. -2. Align the SFT model to AI feedback via DPO on the UltraFeedback dataset. +1. Apply SFT to fine-tune Mistral 7B on a filtered version of the UltraChat dataset ([link](https://huggingface.co/datasets/HuggingFaceH4/ultrachat_200k)). +2. Align the SFT model to AI feedback via DPO on a preprocessed version of the UltraFeedback dataset ([link](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized)). See below for commands to train these models using either DeepSpeed ZeRO-3 or LoRA.