Kimi k1.5
Kimi k1.5 is an advanced multi-modal model trained with reinforcement learning (RL) that achieves state-of-the-art reasoning performance across multiple benchmarks and modalities. It outperforms existing models such as GPT-4o and Claude Sonnet 3.5 in short-CoT tasks by a large margin and matches OpenAI's o1 in long-CoT performance.
Detailed Introduction
Kimi k1.5 is an o1-level multi-modal model trained with reinforcement learning (RL) that achieves state-of-the-art reasoning performance across multiple benchmarks and modalities. It outperforms existing models such as GPT-4o and Claude Sonnet 3.5 in short-CoT tasks by a large margin and matches OpenAI's o1 in long-CoT performance. Key ingredients of Kimi k1.5 include long context scaling, improved policy optimization, a simplistic RL framework, and multi-modal training.
MoreAI
Roboschool
Roboschool is an open-source software for robot simulation, integrated with OpenAI Gym. It provides a variety of environments for training and testing robot controllers, including tasks familiar to Mujoco users and new challenges.
Earkick
Earkick is a free AI-powered chatbot designed to measure and improve mental health in real time, offering personalized support and privacy-focused features.
Depression: Twitter Dataset + Feature Extraction
This dataset contains 20,000 labelled English tweets of depressed and non-depressed users. The data is collected using the Twitter API and includes feature extraction techniques such as topic modelling and emoji sentiment analysis. It is designed for mental health classification at the tweet level.