The NeurIPS Expo on Sunday focused on the theme of efficiency, exploring how to do more with less power, model parameters, and smaller chips. One highly attended session was led by Amazon, which discussed the effectiveness of smaller language models like Mistral 7B compared to larger models. Qualcomm showcased running Stable Diffusion on their Snapdragon mobile processor, Positron AI introduced their new inference chips for improved performance, and AMD discussed energy efficiency through quantization. The rest of the week will continue to emphasize efficiency, with sessions on Efficient Learning and the LLM Efficiency Challenge. Other popular talks covered topics such as AutoML, reinforcement learning, graph learning, and autonomous agents. The main conference begins tomorrow with Affinity Workshops and other exciting events lined up throughout the week. Stay tuned for daily NeurIPS 2023 updates.
https://mlcontests.com/neurips-2023/expo-day/