English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Instruction tuning, a new learning paradigm that fine-tunes pre-trainedlanguage models on tasks specified through instructions, has shown promisingzero-shot performance on various natural language processing tasks. However, ithas yet to be explored for vision and multimodal tasks. In this work, weintroduce MUL-TIINSTRUCT, the first multimodal instruction tuning benchmarkdataset that consists of 62 diverse multimodal tasks in a unified seq-to-seqformat covering 10 broad categories. The tasks are derived from 21 existingopen-source datasets and each task is equipped with 5 expert-writteninstructions. We take OFA as the base pre-trained model for multimodalinstruction tuning, and to further improve its zero-shot performance, weexplore multiple transfer learning strategies to leverage the large-scaleNATURAL INSTRUCTIONS dataset. Experimental results demonstrate strong zero-shotperformance on various unseen multimodal tasks and the benefit of transferlearning from a text-only instruction dataset. We also design a new evaluationmetric - Sensitivity, to evaluate how sensitive the model is to the variety ofinstructions. Our results indicate that fine-tuning the model on a diverse setof tasks and instructions leads to a reduced sensitivity to variations ininstructions for each task.

MultiInstruct: Improving Multi-Modal Zero-Shot Learning via Instruction Tuning