Query the vision model. Args: query: The query text images: Single image or list of images (paths or

Retrieve from all 4 RAG systems

Shorthand for query(). Args: images: Single image or list of images (paths or PIL Images) query: The

Initialize the similarity mapper. Args: model: ColPali model instance processor: ColPali processor i

Initialize the VLM wrapper. Args: model: The model identifier (e.g., "gpt-4o", "meta-llama/llama-4-scout-17b-16e

Initialize the LLM wrapper. Args: model: The model identifier (e.g., "gpt-4", "claude-3-opus-20240229")

Create visualization of aggregated heatmap.

Placeholder for future implementation of generation evaluation.

Evaluate retrieval performance using the generated evaluation dataset. Args: eval_dataset: List of evaluation sa

Flattens a PyTorch tensor and zero-pads it to a desired length. Args: tensor: The input PyTorch tensor. desired_lengt

Flattens a PyTorch tensor and zero-pads it to a desired length. Args: tensor: The input PyTorch tensor. desired_lengt

Generate an evaluation dataset by sampling chunks and creating QA pairs. Args: num_samples: Number of samples to

Generate similarity maps for all retrieved ColPali images

Ingest PDFs into all 4 RAG systems

Check if the current model supports vision inputs.

Convert PIL Image to base64 string