wrapyfi VS zeroros

The authors present an example of combining Wrapyfi (https://github.com/fabawi/wrapyfi), a Python wrapper for message-oriented and robotics middleware, with LLaMA (https://github.com/facebookresearch/llama), a series of large language models from Meta AI. They demonstrate how Wrapyfi can enable running LLaMA on multiple mid-range machines with high inference speed and low cost. They also provide links to their GitHub repository (https://github.com/modular-ml/wrapyfi-examples_llama) and paper (https://arxiv.org/abs/2302.09648) for more details. They state that this example can revolutionize natural language processing tasks such as text generation, summarization, question answering, sentiment analysis, etc. without having to buy new hardware and use their existing infrastructure!

For this reason, I've developed ZeroROS (https://github.com/miquelmassot/zeroros) - a python-only, dependency-free middleware that shares the vibe of ROS (topics, subscribers, messages) providing the same coding patterns (callbacks) without all the hassle of dealing with the abovementioned side-materials.