LLMGPT2 - Tiny GPT-2: Train & Export to GGUF (C# / ILGPU / OpenCL / CUDA / CPU)

[virex-84]

You can add this fully open-source project.
A C# implementation for creating and training extremely small GPT-2 style models from scratch, accelerated via OpenCL(Integrated AMD and Intel graphics)/CUDA(Nvidia) through ILGPU.
The final exported GGUF model is only ~440 KB

https://github.com/virex-84/LLMGPT2