nicholasKluge/Aira-2-1B5

Partial fine-tune vs openai-community/gpt2-xl. Partial change: 25% of the weights changed (attention, mlp, norms), in 12 of 48 layers.

Tensors changed: Attention 48/192, Norms 50/194, MLP 48/192, Embeddings 0/1

1 derivatives checked

• mradermacher/Aira-2-1B5-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse