Install Emotion-LLaMA

1. folk & clone the repo
   • Recommend to use
     • folked repo ‣ (Joon’s modified ver. more easier to setup path configures)
2. Setup env & Install the required packages

cd Emotion-LLaMA
conda env create -f environment.yml
conda activate llama

pip install moviepy==1.0.3 soundfile==0.12.1 opencv-python==4.7.0.72

• Install torch based on your system

Download Models and put the location

<aside> 💡

• Path should be carefully specified ependeng on your own OS platform </aside>

1. Preparing the Pretrained LLM Weights

• Download the Llama-2-7b-chat-hf model from Huggingface to  Emotion-LLaMA/checkpoints/Llama-2-7b-chat-hf/

<aside> 💡

you have to ask permission at the huggingface

</aside>

• Specify the path to Llama-2 in the model config file:
  • minigpt4/configs/models/minigpt_v2.yaml

# Set Llama-2-7b-chat-hf path
llama_model: "/home/user/project/Emotion-LLaMA/checkpoints/Llama-2-7b-chat-hf"

2. HuBERT-large model

• 1.Download the HuBERT-large model from Huggingface to  Emotion-LLaMA/checkpoints/transformer/chinese-hubert-large/
• 2.Specify the path to HuBERT-large in the conversation file:
  • Emotion-LLaMA\minigpt4\conversation\conversation.py

# Set HuBERT-large model path
model_file = "checkpoints/transformer/chinese-hubert-large"

3. Emotion-LLaMA model for Local Demo