Jul 3, 2024

• Meta 3D Gen (3DGen) introduces a fast, state-of-the-art pipeline for text-to-3D asset generation, achieving high prompt fidelity and quality in under a minute, supporting PBR for realistic relighting, and offering generative retexturing for 3D shapes.

• GraphRAG, a new tool by Microsoft Research for complex data discovery, leverages a large language model (LLM) to automate the extraction of knowledge graphs from text documents, enabling advanced question-answering capabilities over private or unseen datasets. GraphRAG on GitHub

• Pretzel is an open-source alternative to Jupyter, introducing enhancements like AI code generation, inline tab completion, and a sidebar for chat and error fixing, aiming to improve Jupyter's capabilities significantly.

• Palico AI is a development framework designed to streamline the process of building, experimenting with, and deploying LLM applications, enabling developers to rapidly test and iterate on various combinations of models, prompts, and architectures to achieve desired accuracy levels.

• AI's revenue gap has widened from $200B to $600B, reflecting a significant discrepancy between the revenue expectations from AI infrastructure investments and actual growth in the ecosystem.

• Hugging Face's transformers v4.42.0 introduces Gemma 2, RTDETR, InstructBLIP, LLAVa Next, and a new model adder, enhancing the library with models trained on 6T tokens, real-time object detection, visual instruction tuning, and improved video understanding.

• AI models used in medical imaging often exhibit biases, performing poorly across different demographic groups, notably in women and people of color, due to reliance on "demographic shortcuts."

• Rubra v0.1 introduces a collection of open-weight, tool-calling large language models (LLMs) including Llama, Mistral, Phi, Qwen, and Gemma, aiming to bridge the gap in function/tool calling capabilities between proprietary and open-source models. Try it out here

• Sparse MetA-Tuning (SMAT) introduces a novel approach by isolating subsets of pre-trained parameters for meta-tuning on each task, aiming to enhance transfer learning in vision beyond traditional methods.

• TSMC is pioneering chip scaling by stacking GPUs, a method that contrasts with Japanese researchers' approach of shrinking devices using a linear accelerator.

• Asynchronous Score Distillation (ASD) leverages text-to-image diffusion priors to synthesize 3D content efficiently without requiring paired text-3D data, overcoming the scalability issues faced by existing methods.

• MeMemo introduces on-device retrieval augmentation for text generation, addressing data privacy concerns in sensitive fields by utilizing client-side dense retrieval.

• Integrating PyTorch Geometric with OpenAI Gym's FrozenLake-v1 environment, the researcher encounters challenges in achieving model convergence without unique node features, such as positional encodings or unique indices.

• Language models, when fine-tuned on behavior data, can produce high-quality mixed embeddings by leveraging Retrieval-Augmented Generation (RAG) methods.

• Neurocache enhances large language models (LLMs) by using an external vector cache to store and retrieve past states, significantly extending effective context size and improving model performance.