ellabettison/Logo-Detection-finetune

Full fine-tune vs hustvl/yolos-small. Fine-tune: 100% of the weights changed, across 12 of 12 layers.

Tensors changed: other 58/58, Embeddings 6/6, Attention 96/96, Norms 50/50

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse