Myric/Spark-X2.5-4B-GGUF
Same weights, converted vs XHToken/Spark-X2.5-4B. Same weights stored in another dtype (BF16 → BF16/F32), plus 108 new tensors.
Tensors changed: Embeddings 0/1, Norms 0/73, MLP 0/108, Attention 0/72
Same weights, converted vs XHToken/Spark-X2.5-4B. Same weights stored in another dtype (BF16 → BF16/F32), plus 108 new tensors.
Tensors changed: Embeddings 0/1, Norms 0/73, MLP 0/108, Attention 0/72