diff --git a/scripts/mplug_owl_2/README.md b/scripts/mplug_owl_2/README.md index ecbd8dd..89c503b 100644 --- a/scripts/mplug_owl_2/README.md +++ b/scripts/mplug_owl_2/README.md @@ -12,7 +12,7 @@ Install [mPLUG-Owl](https://github.com/X-PLUG/mPLUG-Owl/) under the main reposit ```shell git clone https://github.com/X-PLUG/mPLUG-Owl.git -cd mPLUG-Owl/mPLUG_Owl_2 +cd mPLUG-Owl/mPLUG-Owl2 pip install -e ".[train]" pip install flash_attn --no-build-isolation cd .. @@ -30,7 +30,7 @@ After that, you can conduct *low-level visual instruction tuning* as follows, un For the **Q-Instruct** dataset, download them directly via the following script: ```shell -cd mPLUG-Owl/mPLUG_Owl_2/playground/data +cd mPLUG-Owl/mPLUG-Owl2/playground/data wget https://huggingface.co/datasets/teowu/Q-Instruct/resolve/main/cleaned_labels.json wget https://huggingface.co/datasets/teowu/Q-Instruct/resolve/main/q-instruct-images.tar tar -xf q-instruct-images.tar @@ -56,7 +56,7 @@ If you choose the ***mix*** strategy, the high-level datasets also need to be do 1. Download the annotation of the final mixture our instruction tuning data [llava_v1_5_mix665k.json](https://huggingface.co/datasets/liuhaotian/LLaVA-Instruct-150K/blob/main/llava_v1_5_mix665k.json): ```shell -wget -P mPLUG-Owl/mPLUG_Owl2/playground/data https://huggingface.co/datasets/liuhaotian/LLaVA-Instruct-150K/blob/main/llava_v1_5_mix665k.json +wget -P mPLUG-Owl/mPLUG-Owl2/playground/data https://huggingface.co/datasets/liuhaotian/LLaVA-Instruct-150K/blob/main/llava_v1_5_mix665k.json ``` 2. Download the images from constituting datasets: @@ -86,7 +86,7 @@ After downloading all of them, organize the high-level data as follows in `LLaVA 3. Merge the **Q-Instruct** labels with labels from high-level datasets. ```shell -jq -s 'add' mPLUG-Owl/mPLUG_Owl2/playground/data/cleaned_labels.json mPLUG-Owl/mPLUG_Owl2/playground/data/llava_v1_5_mix665k.json > mPLUG-Owl/mPLUG_Owl2/playground/data/mix_cleaned_labels.json +jq -s 'add' mPLUG-Owl/mPLUG-Owl2/playground/data/cleaned_labels.json mPLUG-Owl/mPLUG-Owl2/playground/data/llava_v1_5_mix665k.json > mPLUG-Owl/mPLUG-Owl2/playground/data/mix_cleaned_labels.json ``` @@ -97,7 +97,7 @@ Please make sure you have enough computational resources before training. - [Must Do!] Replace all the `` token in the json into `<|image|>`, Otherwise the image will not be loaded into training. ```shell -sed -i 's//<|image|>/g' mPLUG-Owl/mPLUG_Owl2/playground/data/mix_cleaned_labels.json +sed -i 's//<|image|>/g' mPLUG-Owl/mPLUG-Owl2/playground/data/mix_cleaned_labels.json ``` #### Strategy (a): Mix with High-level Datasets