fhai50032/Mistral-4B
Partial fine-tune (fewer layers) vs fhai50032/RolePlayLake-7B. fhai50032/Mistral-4B matches part of fhai50032/RolePlayLake-7B (147 of its 291 tensors; the rest of fhai50032/RolePlayLake-7B has no counterpart). On the shared part: Partial change: 47% of the weights changed (attention, mlp, norms), in 8 of 16 layers.
Tensors changed: MLP 24/48, Attention 32/64, Norms 16/33, Embeddings 0/1, lm_head 0/1
1 derivatives checked
- mradermacher/Mistral-4B-GGUF: Same weights, converted