Maya: An Instruction Finetuned Multilingual Multimodal Model
This addresses the gap in handling low-resource languages and cultural contexts for users in diverse regions, though it is incremental as it builds on existing datasets and methods.
The paper tackles the problem of large Vision-Language Models struggling with low-resource languages and cultural nuances by introducing Maya, an open-source multilingual multimodal model, which includes a toxicity-free dataset in eight languages and improves cultural and linguistic comprehension in vision-language tasks.
The rapid development of large Vision-Language Models (VLMs) has led to impressive results on academic benchmarks, primarily in widely spoken languages. However, significant gaps remain in the ability of current VLMs to handle low-resource languages and varied cultural contexts, largely due to a lack of high-quality, diverse, and safety-vetted data. Consequently, these models often struggle to understand low-resource languages and cultural nuances in a manner free from toxicity. To address these limitations, we introduce Maya, an open-source Multimodal Multilingual model. Our contributions are threefold: 1) a multilingual image-text pretraining dataset in eight languages, based on the LLaVA pretraining dataset; 2) a thorough analysis of toxicity within the LLaVA dataset, followed by the creation of a novel toxicity-free version across eight languages; and 3) a multilingual image-text model supporting these languages, enhancing cultural and linguistic comprehension in vision-language tasks. Code available at https://github.com/nahidalam/maya.