Skip to content

Latest commit

 

History

History
16 lines (13 loc) · 544 Bytes

README.md

File metadata and controls

16 lines (13 loc) · 544 Bytes
Raw

VSR Dataset

Prerequisites

Ollama is required to run the prepare_vsr_dataset.py script, which generates the VSR dataset. Moreover, the llama3 model must be pulled through Ollama.

The following requirements must be met to run the prepare_vsr_dataset.py script, which generates the VSR dataset:

  1. Download Ollama.
  2. ollama pull llama3
  3. ollama serve

How to run

Perform the following steps, executing from the vsr directory:

python ./prepare_vsr_dataset.py \
--vsr_questions [path to output JSON]