hotchpotch/bekko-embedding-v1-a8m-GGUF
Same weights, converted vs hotchpotch/bekko-embedding-v1-a8m. Same weights stored in another dtype (BF16 → F16/F32).
Tensors changed: Embeddings 0/2, Norms 0/8, Attention 0/8, MLP 0/8
Same weights, converted vs hotchpotch/bekko-embedding-v1-a8m. Same weights stored in another dtype (BF16 → F16/F32).
Tensors changed: Embeddings 0/2, Norms 0/8, Attention 0/8, MLP 0/8