ML Datasets & Models — Local Mirror

A small local mirror for offline study and reproducibility experiments. Last index update: 2026-05-06.

ML Datasets & Models

• Foundational papers — AlexNet to Mamba
• Pretrained models — BERT, ResNet
• Educational LLMs — GGUF format, runs on laptops
• Code archives — nanoGPT, llm.c, tinygrad
• Image datasets — MNIST, Fashion-MNIST, CIFAR-10
• Benchmark suites — COCO 2017
• Audio datasets — LibriSpeech

Software

• Software releases — PyTorch, TensorFlow wheels

Featured Large Files

• llama-3.1-8b-instruct.gguf — 8.0 GB
• qwen2.5-7b.safetensors — 4.7 GB
• imagenet-val.tar — 6.3 GB
• librispeech-train-clean-360.tar.gz — 23 GB
• coco2017-train.zip — 18 GB