rizla/rizla-9

Full fine-tune (fewer layers) vs CultriX/NeuralTrix-7B-dpo. CultriX/NeuralTrix-7B-dpo matches part of rizla/rizla-9 (291 of its 480 tensors; the rest of rizla/rizla-9 has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 32 of 32 layers.

Tensors changed: Embeddings 1/1, Norms 65/65, MLP 96/96, Attention 128/128, lm_head 1/1

On the Hub · Downloads on Model Pulse