Large language models (LLMs) have been recently leveraged as training data generators for various natural language processing (NLP) tasks. While previous research has explored different approaches to training models using generated data, they generally rely on simple class-conditional prompts, which may limit the diversity of the generated data and inherit systematic biases of LLM. Thus, we investigate training data generation with diversely attributed prompts (e.g.,

Baseline Methods As standard baselines, we first consider zero-shot CLIP (ZS) and vanilla fine-tuning (FT) with contrastive loss. We construct the label map for contrastive loss by regarding all of the samples from a class as positives. A.3 Multi-modal Classification Dataset To evaluate the multi-modal representation learning under video emotional classification, CMU-MOSEI consists of three modalities textual (T), visual (V), and audio (A), and contains 23,453 Y ouTube video clips about diverse movie reviews, and each clip is annotated with ordinal labels ranging from -3 (strong negative) to 3 (strong positive). While MulT learns the joint encoder only with standard classification loss (i.e., cross-entropy loss; Metric For image-text retrieval, we adopt top-1 and top-5 recalls likewise CLIP retrieval setup. Flickr30k for zero-shot transferred image-text retrieval.

A room's acoustic properties are a product of the room's geometry, the objects within the room, and their specific positions. A room's acoustic properties can be characterized by its impulse response (RIR) between a source and listener

For each (topic, subtopic) pair, we generate and solve 80 problems using GPT4.

Segmenting long videos into chapters enables users to quickly navigate to the information of their interest.

We build strong baseline models upon large pretrained language models, including GPT -3 and T5. Our benchmark is an ongoing effort, and this paper presents real-time evaluation results over the past year.