December 2023 Summaries

In 2023, curiosity and interest in AI spiked, especially after OpenAI launched ChatGPT. This led to a greater understanding and utilization of vector databases, which became a strong foundation for building effective AI applications. Weaviate's global online hackathons fostered innovation and collaboration, while in-person events brought the community together to celebrate successes. Community projects like Basepilot and BYTE showcased the boundless possibilities within the AI realm. Embracing open source and sharing knowledge through initiatives such as Hacktoberfest and Weaviate Academy further strengthened the community's growth and development. As we look forward to 2024, Weaviate is excited to continue building, learning, and collaborating with its global community of developers, data scientists, and AI enthusiasts.

Weaviate 1.23 introduces AutoPQ, which automatically triggers Product Quantization (PQ) for vector indexing, improving the developer experience. It also adds a new flat vector index type with Binary Quantization for small collections and multi-tenancy use cases. Additionally, this release integrates Weaviate with open-source large language models like Llama2-70b, CodeLlama-34b, or Mistral-7B-Instruct through the generative-anyscale module. The Python client has been updated to support new features and performance improvements include lazy shard loading and automatic resource limiting.

Humans learn differently from current state-of-the-art models like Large Language Models (LLMs). While LLMs are trained on trillions of tokens, they still lack a vivid understanding of the causal relationships in the world. In contrast, humans can efficiently form coherent understandings of the world by integrating multiple sensory inputs and using multimodal representations of information. To train models that understand multimodal data, one approach is Contrastive Learning. This involves training individual models for each modality separately and then unifying their representations through contrastive training. Another technique is Any-to-Any Search and Retrieval, which uses multimodal embedding models to perform any-to-any search and scaling these embeddings into production using Vector Databases. Multimodal Retrieval Augmented Generation (MM-RAG) augments the generation from Large Multimodal Models (LMMs) with multimodal retrieval of images and more. This process involves two steps: retrieving information from a multimodal knowledge base and generating content using a large multimodal model by grounding in the retrieved context. In summary, understanding how humans learn can help improve AI models' ability to understand causal relationships and utilize multimodal data effectively. Techniques like Contrastive Learning, Any-to-Any Search and Retrieval, and MM-RAG contribute to this goal by enabling efficient integration of multiple sensory inputs and improving the accuracy and scalability of AI models.

The Weaviate team attended AWS re:Invent in November, where they observed a strong focus on Generative AI and Retrieval Augmented Generation (RAG). They received positive feedback on their open-source RAG app, Verba. During the event, attendees expressed confusion about the rapidly evolving AI market and sought guidance on choosing the right vector database for their needs. Weaviate highlighted its purpose-built platform for AI applications and emphasized its security measures, including SOC 2 certification and encryption in transit and at rest. The team also introduced Weaviate Vector Club, a community initiative with merchandise and events to engage users.