webmp3/Sakura-d1-3B-HighQuality-GGUF
Quantization (fewer layers) vs LiquidAI/d1-3B. webmp3/Sakura-d1-3B-HighQuality-GGUF matches part of LiquidAI/d1-3B (244 of its 707 tensors; the rest of LiquidAI/d1-3B has no counterpart). On the shared part: webmp3/Sakura-d1-3B-HighQuality-GGUF is a GGUF quantization (Q4) of the same weights: 77 tensors match exactly and 167 within quantization error.
Tensors changed: Embeddings 0/2, other 0/44, MLP 0/90, Norms 0/76, Attention 0/32