DewEfresh/neo_7b-slerp
Full fine-tune (fewer layers) vs m-a-p/neo_7b. DewEfresh/neo_7b-slerp matches part of m-a-p/neo_7b (129 of its 255 tensors; the rest of m-a-p/neo_7b has no counterpart). On the shared part: Fine-tune: 90% of the weights changed, across 14 of 14 layers.
Tensors changed: Embeddings 0/1, Norms 28/29, MLP 42/42, Attention 56/56, lm_head 0/1
1 derivatives checked
- mradermacher/neo_7b-slerp-GGUF: Same weights, converted