tensopolis/virtuoso-lite-tensopolis-v2
LoRA-style vs arcee-ai/Virtuoso-Lite. Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of k_proj changed, in 8 of 40 layers; norms, biases and embeddings are untouched.
Tensors changed: Embeddings 0/1, Norms 0/81, MLP 0/120, Attention 8/160, lm_head 0/1
2 derivatives checked
- mradermacher/virtuoso-lite-tensopolis-v2-i1-GGUF: Quantization
- mradermacher/virtuoso-lite-tensopolis-v2-GGUF: Partial fine-tune