xiaomi-research/MiLMMT-46-12B-v1.0
Full fine-tune vs xiaomi-research/MiLMMT-46-12B-v0.1. Fine-tune: 97% of the weights changed, across 48 of 48 layers.
Tensors changed: Embeddings 1/4, Norms 102/400, MLP 144/252, Attention 192/408, other 0/1
2 derivatives checked
- mradermacher/MiLMMT-46-12B-v1.0-GGUF: Quantization (fewer layers)
- mradermacher/MiLMMT-46-12B-v1.0-i1-GGUF: Quantization (fewer layers)
1 other repos hold byte-for-byte identical weight files.