ggml-org/tinygemma3-GGUF
Quantization (fewer layers) vs ngxson/tinygemma3_cifar. ggml-org/tinygemma3-GGUF matches part of ngxson/tinygemma3_cifar (106 of its 188 tensors; the rest of ngxson/tinygemma3_cifar has no counterpart). On the shared part: ggml-org/tinygemma3-GGUF is a GGUF quantization (Q8_0) of the same weights: 49 tensors match exactly and 57 within quantization error.
Tensors changed: Embeddings 0/1, Norms 0/49, MLP 0/24, Attention 0/32