Remove old custom-format checkpoint (replaced by model.safetensors)

Upload GPUburnout-1B base model (step 90K, loss 2.494)

test upload config.json

Upload pytorch_model.bin with huggingface_hub

initial commit