allura-org/MS-Meadowlark-22B
Full fine-tune vs unsloth/Mistral-Small-Instruct-2409. Fine-tune: 99% of the weights changed, across 56 of 56 layers.
Tensors changed: Embeddings 1/1, Norms 9/113, MLP 168/168, Attention 224/224, lm_head 0/1
4 derivatives checked
- bartowski/MS-Meadowlark-22B-GGUF: Same weights, converted
- mradermacher/MS-Meadowlark-22B-i1-GGUF: Quantization
- mradermacher/MS-Meadowlark-22B-GGUF: Quantization
- ThomasComics/MS-ManciousWriter-22B-v0.2: Full fine-tune