kingjones777/Sarvam-30B-ROCmFP4-GGUF
Quantization vs sarvamai/sarvam-30b. kingjones777/Sarvam-30B-ROCmFP4-GGUF is a GGUF quantization (Q8_0_ROCMFPX_AGENT) of the same weights: 95 tensors match exactly and 45 within quantization error.
Tensors changed: Attention 0/14, Norms 0/77, MLP 0/47, Embeddings 0/1, lm_head 0/1