PollardWeights/Spark-X2.5-4B-Pollard
Quantization vs XHToken/Spark-X2.5-4B. PollardWeights/Spark-X2.5-4B-Pollard is a GGUF quantization (Q6_K) of the same weights: 73 tensors match exactly and 108 within quantization error.
Tensors changed: Norms 0/73, MLP 0/108