AGmind/strizh-ru-retriever
Full fine-tune (fewer layers) vs deepvk/RuModernBERT-small. AGmind/strizh-ru-retriever matches part of deepvk/RuModernBERT-small (26 of its 77 tensors; the rest of deepvk/RuModernBERT-small has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 4 of 4 layers.
Tensors changed: Embeddings 1/2, Norms 6/8, Attention 8/8, MLP 8/8
1 derivatives checked
- AGmind/strizh-ru-retriever-GGUF: Same weights, converted