My model writes incoherent stories at the first stage
Eli Levasseur built a custom C++ framework without external libraries to train a small language model on children's stories, which currently fails to understand context.
- Field
- training language models
- What they did
- The author wrote a neural network framework in C++ without using external math or ML libraries to train a small language model on children's stories.
- Why it matters
- The goal was to understand how every piece of the neural network works instead of creating another PyTorch data science project.
#neural network#c++#language model#student



