A new open-source library called nanoVLM has been released, allowing users to train their vision-language models (VLMs) using only PyTorch. The library is designed to be lightweight and easy to use, making it a simpler alternative to other VLM training repositories. According to the developers, nanoVLM provides a streamlined experience for training VLMs, with features such as automatic data loading and model evaluation. The library has been released on Hugging Face's model hub. --- Why it matters: This matters because researchers and engineers working on vision-language models can now use a simpler and more lightweight library to train their models, potentially speeding up development time and improving collaboration. Source: https://huggingface.co/blog/nanovlm