From 25b8b8574c62667bd3ee19ec983c6e2a667b5dc4 Mon Sep 17 00:00:00 2001
From: Lxinyang <68582965+littleSunlxy@users.noreply.github.com>
Date: Mon, 19 Feb 2024 11:44:15 +0800
Subject: [PATCH] Update MobileLLaMA_SFT.md

update the detail description of MobileLLaMA SFT
---
 mobilellama/sft/MobileLLaMA_SFT.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/mobilellama/sft/MobileLLaMA_SFT.md b/mobilellama/sft/MobileLLaMA_SFT.md
index 13206c2..b55fded 100644
--- a/mobilellama/sft/MobileLLaMA_SFT.md
+++ b/mobilellama/sft/MobileLLaMA_SFT.md
@@ -1,7 +1,7 @@
 # MobileLLaMA SFT
 
 ## 🛠️ Installation
-Our MobileLLaMA SFT based on [FastChat](https://github.com/lm-sys/FastChat) (commit id: 81785d7ed1d6afb966b464a8ee4689b7413e6313)
+Our MobileLLaMA SFT training code is based on [FastChat](https://github.com/lm-sys/FastChat) (commit id: 81785d7ed1d6afb966b464a8ee4689b7413e6313)
 
 ### Install From Source.
 1. Clone the [FastChat](https://github.com/lm-sys/FastChat) repository and navigate to the FastChat folder
@@ -27,7 +27,7 @@ You can download MobileLLaMA-1.4B-Base / MobileLLaMA-2.7B-Base model from huggin
 ## Dataset
 We use the sft dataset in Vicuna fromat can be download from link: [ShareGPT_Vicuna_dataset](https://huggingface.co/datasets/Aeala/ShareGPT_Vicuna_unfiltered), and follow the steps:
 1. download the [json](https://huggingface.co/datasets/Aeala/ShareGPT_Vicuna_unfiltered/blob/main/ShareGPT_V4.3_unfiltered_cleaned_split.json) file to local data path.
-2. white the correct "--data_path" in the training script.
+2. write the correct "--data_path" in your SFT training scripts.
 
 ## 💎 Training
 Our training process can be reproduced by runing the scrips: