This repository contains the task examples, retrieval documents (in the archive evaluation_examples.zip), and virtual machine snapshots for benchmark Spider2-V (loaded by VMware depending on the machine architecture x86 or arm64).
You can find more information from our paper "Spider2-V: How Far Are Multimodal Agents From Automating Data Science and Engineering Workflows?"
paper Arxiv link:
https://arxiv.org/abs/2407.10956
project website:… See the full description on the dataset page:
https://huggingface.co/datasets/xlangai/ubuntu_spider2v.
