MadameMoonflower/CitrusTea-Test

Partial fine-tune (fewer layers) vs grimjim/kukulemon-7B. grimjim/kukulemon-7B matches part of MadameMoonflower/CitrusTea-Test (291 of its 417 tensors; the rest of MadameMoonflower/CitrusTea-Test has no counterpart). On the shared part: Partial change: 26% of the weights changed (attention, lm_head, mlp, norms), in 8 of 32 layers.

Tensors changed: Embeddings 0/1, Norms 17/65, MLP 24/96, Attention 32/128, lm_head 1/1

On the Hub · Downloads on Model Pulse