Text Generation
GGUF
PyTorch
nvidia
nemotron-3
latent-moe
mtp
conversational

UD-IQ4XS is a duplicate of UD-IQ4NL

#4
by XZiar - opened

the sha256 is the same, and ffn is iq4nl+iq3, I guess the iq4xs version is uploaded incorrectly?

XZiar changed discussion title from UD-IQ4XS is a duplicate od UD-IQ4NL to UD-IQ4XS is a duplicate of UD-IQ4NL
Unsloth AI org

This is the case as we stated in our guide: "Some GGUFs end up similar in size because the model architecture (like gpt-oss) has dimensions not divisible by 128, so parts can’t be quantized to lower bits. Access GGUFs here."

See: https://unsloth.ai/docs/models/nemotron-3-super#run-nemotron-3-super-120b-a12b

Sign up or log in to comment