allura-org/Koto-Small-7B-PT
Full fine-tune vs XiaomiMiMo/MiMo-7B-Base. Fine-tune: 100% of the weights changed, across 36 of 36 layers.
Tensors changed: Embeddings 1/1, Norms 52/73, MLP 108/108, Attention 252/252, lm_head 1/1
4 derivatives checked
- mradermacher/Koto-Small-7b-rc1-i1-GGUF: Quantization
- mradermacher/Koto-Small-7b-rc1-GGUF: Same weights, converted
- Aurore-Reveil/Koto-Small-7B-IT: Full fine-tune
- allura-forge/Koto-Small-7B-IT-ThonkTokens: Full fine-tune