activeDap/gpt2-xl_hh_harmful

Full fine-tune vs openai-community/gpt2-xl. Fine-tune: 100% of the weights changed, across 48 of 48 layers.

Tensors changed: Attention 192/192, Norms 194/194, MLP 192/192, Embeddings 1/2

1 derivatives checked

• mradermacher/gpt2-xl_hh_harmful-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse