zerodegress/NanoJev-bf16
Partial fine-tune vs C-Tianyu/NanoJev. Partial change: 74% of the weights changed (attention, mlp, norms, other), in 28 of 28 layers.
Tensors changed: Embeddings 0/1, Norms 106/115, MLP 84/84, Attention 116/116, other 5/6
Partial fine-tune vs C-Tianyu/NanoJev. Partial change: 74% of the weights changed (attention, mlp, norms, other), in 28 of 28 layers.
Tensors changed: Embeddings 0/1, Norms 106/115, MLP 84/84, Attention 116/116, other 5/6