Toggle light / dark theme

Bizarre, Evolutionary Missing Link Uncovered in Hubble Deep Survey of Galaxies The universe is so saturated with galaxies that even the weirdest things can go unnoticed for years after Hubble Space Telescope “deep-exposure” observations are taken. In sort of an intergalactic Where’s Waldo, an international team of astronomers uncovered in Hubble archival data a mysterious red dot nearly in the middle of the Great Observatories Origins Deep Survey-North (GOODS-North). As innocuous as it looks, it could be a rare missing link between some of the very earliest galaxies and the birth of supermassive black holes. The object, referred to as GNz7q, existed when the universe was just a toddler, only 750 million years after the big bang. The mixture of radiation from the object cannot be attributed to star formation alone. The best explanation is that it is a growing black hole shrouded in dust. Given time, the black hole will emerge from its dusty cocoon as a brilliant quasar, an intense beacon of light at the heart of an early galaxy. The pioneering Hubble telescope has provided a unique target for NASA ’s James Webb Space Telescope to use its spectroscopic instruments to study objects like GNz7q in unprecedented detail.

“Our batteries are designed to suit the needs of stationary power applications where safety, lifetime, levelized costs, and environmental footprints are key decision drivers,” the company said in a statement. “PolyJoule’s conductive polymer cells span the performance curve between traditional lead-acid batteries and modern lithium-ion cells, while enhancing service life and reducing balance of plant costs, due to their no-HVAC thermal management design.”

According to the manufacturer, the battery cells were tested to perform for 12,000 cycles at 100% depth of discharge. The device is based on a standard, two-electrode electrochemical cell containing the conductive polymers, a carbon-graphene hybrid, and a non-flammable liquid electrolyte. Alternating anodes and cathodes are interwoven and then connected in parallel to form a cell.

Large-scale language-based foundation models such as BERT, GPT-3 and CLIP have exhibited impressive capabilities ranging from zero-shot image classification to high-level planning. In most cases, these large language models, visual-language models and audio-language models remain domain-specific and rely highly on the distribution of their training data. The models thus obtain different although complementary common-sense knowledge within specific domains. But what if such models could effectively communicate with one another?

In the new paper Socratic Models: Composing Zero-Shot Multimodal Reasoning with Language, Google researchers argue that the diversity of different foundation models is symbiotic and that it is possible to build a framework that uses structured Socratic dialogue between pre-existing foundation models to formulate new multimodal tasks as a guided exchange between the models without additional finetuning.

This work aims at building general language-based foundation models that embrace the diversity of pre-existing language-based foundation models by levering structured Socratic dialogue, and offers insights into the applicability of the proposed Socratic Models on challenging perceptual tasks.