Extract audio from video files and save as MP3. Args: video (str): Video identifier video_config (dict): Video configuration sett

Generate message list for chat completion from mixed inputs. Args: inputs (list): List of input dictionaries with 'type' and 'content' ke

Generate a test file name from sample and task. Args: sample (str): Sample name or path task (str): Task identifier Returns:

Generate transcripts from audio files using Whisper model. Args: video (str): Video identifier video_config (dict): Video configu

Get information about entities by retrieving connected episodic and semantic nodes. This function takes a list of anchor nodes and fi

Get video files matching the hashed title. Files can be .mp4, .mp3, .srt, .txt, depending on the base_path type. Args: base_path (str

Get chat completion response from specified model. Args: model (str): Model identifier messages (list): List of message dictionar

Extract unique video names in a given directory. Args: path (str): Directory path to search Returns: list: List of unique vi

Generate video and segment paths from URL and task. Args: video_url (str): URL of the video task (str): Task identifier Retu

Load and parse subtitle file, extracting only dialogue lines. Args: subtitle_path (str): Path to .srt subtitle file Returns:

Load transcript text from file. Args: transcript_path (str): Path to transcript file Returns: str: Content of transcript fil

ablation study, without equivalence

Process multiple messages in parallel using ThreadPoolExecutor. Messages are processed in batches, with each batch completing before starting the

Plot the distribution of values in an array. Args: values (array-like): Array of numeric values to plot save_path (str, optional)

Save video graph to pickle file. Args: video_graph (VideoGraph): Video graph to save config (dict): Configuration settings

Verify that a video was properly split into clips by checking the number of clips. Args: video_path (str): Path to original video fil

Visualize the video graph.