Things & ideas I’ve made and dreamed to make computer agents come true.

I hope to make a small contribution on improving the agent evaluation, data benchmark, pre-training, and method to make the computer agents more intelligent and come to life. Stay tuned for more!

  • Agent Evaluation

    Creating trustworthy evaluation pipeline to evaluate agent performances in grounded environments with open-ended tasks.

    Coming soon...

  • Agent Data Benchmark

    Collecting large-scale, high-quality datasets for training and tuning agents.

    Coming soon

  • Agent Pre-training

    Coming soon...

    Coming soon...

  • Agent Method

    After we have the decent agent model, we can research on agentic methods to improve the agent performance, like In-context Learning, etc.

    In the future