“No matter how much data you train them on, they still don’t truly understand multiplication.” Written by Gary Marcus, who “feels really old whenever he has to write articles like this. He had really hoped to have said his last […]
VC investor Sarah Tavel suggests that LLM-based software companies should avoid selling end-user productivity, and instead sell a complete work output. ~ learn more
The Silicon Valley law firm Gunderson Dettmer built their own internal app for using large language models in their practice. The most exciting “component is the ability of lawyers to query documents they provide using retrieval-augmented generation (RAG), a method […]
“Med-PaLM M is a large multimodal generative model that flexibly encodes and interprets biomedical data including clinical language, imaging, and genomics with the same set of model weights. Med-PaLM M reaches performance competitive with or exceeding the state of the […]
“We hypothesize two failure modes of safety training: competing objectives and mismatched generalization. Competing objectives arise when a model’s capabilities and safety goals conflict, while mismatched generalization occurs when safety training fails to generalize to a domain for which capabilities […]
This seems like it might be something. “Gorilla is a LLM that can provide the appropriate API calls. It is trained on three massive machine learning hub datasets: Torch Hub, TensorFlow Hub and HuggingFace. … Zero-shot Gorilla outperforms GPT-4, Chat-GPT […]
Amazon researchers paired text with images as input to drive LLM ability ever higher. And they did it with a model that’s only 770 million params (vs the 175 Billion in GPT-3.5). “Our method achieves new state-of-the-art performance on the […]