Emotion-LLaMA is structured as follows:
📦Dataset
┗ 📦Emotion
┗ 📂MER2023
┣ 📂video
┣ 📂HL-UTT
┣ 📂mae_340_UTT
┣ 📂maeV_399_UTT
┣ 📄transcription_en_all.csv
┣ 📄MERR_coarse_grained.txt
┣ 📄MERR_coarse_grained.json
┣ 📄MERR_fine_grained.txt
┗ 📄MERR_fine_grained.json
📦Emotion-LLaMA
┣ 📂checkpoints
┃ ┣ 📂Llama-2-7b-chat-hf
┃ ┣ 📂save_checkpoint
┃ ┃ ┣ 📂stage2
┃ ┃ ┃ ┣ 🔖checkpoint_best.pth
┃ ┃ ┃ ┗ 📄log.txt
┃ ┃ ┗ 🔖Emoation_LLaMA.pth
┃ ┣ 📂transformer
┃ ┃ ┗ 📂chinese-hubert-large
┃ ┗ 🔖minigptv2_checkpoint.pth
┣ 📂eval_configs
┃ ┣ 📜demo.yaml
┃ ┣ 📜eval_emotion.yaml
┃ ┗ 📜eval_emotion_EMER.yaml
┣ 📂train_configs
┃ ┣ 📜Emotion-LLaMA_finetune.yaml
┃ ┗ 📜minigptv2_tuning_stage_2.yaml
┣ 📂minigpt4
┣ 📑app.py
┣ 📜environment.yml
┣ 📑eval_emotion.py
┣ 📑eval_emotion_EMER.py
┗ 📑train.py