├──
    ├── ./data/  # Contains the keyword lists and preprocessing functions to create the JSONs combining file_name and corresponding GT captions
    ├── ./language_model/ # Stores the code of the language model
    ├── ./clip/  # Contains the code of the audio CLIP models. Their checkpoints can also be stored here
    ├── ./evaluation/ # Stores the code to create the plots and tables of the dissertation
    ├── ./sh_folder/ # Contains the shell-scripts to infer and to create the tables / plots
    ├── ./inference_result/ # Contains the inferenced results (one JSON, one CSV and one HTML per run) each run is uniquely identified by the timestamp
        ├── /facebook/opt-1.3b
            ├── /AudioCaps/excludes_prompt_MAGIC/
                ├── evaluation # csv files with aggregated NLG metrics
                ├── output_jsons # json files with individual predictions and metrics
                ├── output_tables # html files with individual predictions and metrics
            ├── /clotho_v2.1/excludes_prompt_MAGIC/
                ├── evaluation # csv files with aggregated NLG metrics
                ├── output_jsons # json files with individual predictions and metrics
                ├── output_tables # html files with individual predictions and metrics
    ├── ./softlinks/ # Store the softlinks to or the audio files themselves here!!!
            ├── AudioCaps_data  # stores AudioCaps .wav files
                ├── audiocaps_audio_clip_1.wav
                ...
            ├── evaluation_data_files # stores Clotho .wav files
                ├── clotho_audio_clip_1.wav
                ...
    
    ├── inference_magic.py # The central python script combining the data, the models and that stores the results
    └── sound_obj_generator.py # This program created the ChatGPT keyword list