🚀This week, AI is stepping into new dimensions—becoming co-scientists, sculpting 3D avatars, and blending cloud and on-device models into a seamless dance of creativity and efficiency.🚨
- Towards an AI co-scientist
- Towards an AI co-scientist
Comments
- The FFT Strikes Back: An Efficient Alternative to Self-Attention
- Avat3r: Large Animatable Gaussian Reconstruction Model for High-fidelity 3D Head Avatars
- Fractal Generative Models
- HVI: A New color space for Low-light Image Enhancement
- ESPnet-SpeechLM: An Open Speech Language Model Toolkit
- Reasoning with Latent Thoughts: On the Power of Looped Transformers
This paper introduces an AI co-scientist, a multi-agent system built on Gemini 2.0, designed to assist researchers by generating and refining novel scientific hypotheses. The system employs a "generate, debate, and evolve" framework to iteratively improve hypotheses.
SWE-RL is a RL approach that enhances LLM reasoning for software engineering by learning from open-source software evolution data. It trains Llama3-SWE-RL-70B on GitHub PR data using a rule-based reward system.
FFTNet replaces self-attention with an efficient O(n log n) spectral filtering approach through FFT. By operating in the frequency domain, it captures long-range dependencies with a learnable spectral filter and nonlinear activations.