Sumitc13/sarvam-30b-GGUF
Same weights, converted vs sarvamai/sarvam-30b. Same weights stored in another dtype (F32 → BF16/F32), plus 72 new tensors.
Tensors changed: Attention 0/19, Norms 0/77, MLP 0/75, Embeddings 0/1, lm_head 0/1
Same weights, converted vs sarvamai/sarvam-30b. Same weights stored in another dtype (F32 → BF16/F32), plus 72 new tensors.
Tensors changed: Attention 0/19, Norms 0/77, MLP 0/75, Embeddings 0/1, lm_head 0/1