mradermacher/saiga_gemma3_12b-GGUF
Full fine-tune (fewer layers) vs IlyaGusev/saiga_gemma3_12b. mradermacher/saiga_gemma3_12b-GGUF matches part of IlyaGusev/saiga_gemma3_12b (626 of its 1065 tensors; the rest of IlyaGusev/saiga_gemma3_12b has no counterpart). On the shared part: mradermacher/saiga_gemma3_12b-GGUF is a GGUF (Q8_0) of different weights. Fine-tune: 100% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/1, Norms 0/289, MLP 144/144, Attention 192/192