nvidia/NVIDIA-NemotronLabs-VoiceChat-11B
Full fine-tune (fewer layers) vs nvidia/NVIDIA-Nemotron-Nano-9B-v2. nvidia/NVIDIA-Nemotron-Nano-9B-v2 matches part of nvidia/NVIDIA-NemotronLabs-VoiceChat-11B (339 of its 1632 tensors; the rest of nvidia/NVIDIA-NemotronLabs-VoiceChat-11B has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 56 of 56 layers.
Tensors changed: other 147/189, Norms 65/84, MLP 50/50, Attention 16/16
4 derivatives checked
- hoidhxd/NVIDIA-NemotronLabs-VoiceChat-11B-GGUF: Same weights, converted
- mlx-community/NemotronLabs-VoiceChat-11B-8bit: Quantization
- mlx-community/NemotronLabs-VoiceChat-11B-4bit: Quantization
- mlx-community/NemotronLabs-VoiceChat-11B-mlx-8bit: Quantization
4 other repos hold byte-for-byte identical weight files.